#387 – George Hotz: Tiny Corp, Twitter, AI Safety, Self-Driving, GPT, AGI & God

George Hotz is a programmer, hacker, and the founder of comma-ai and tiny corp. Please support this podcast by checking out our sponsors: – Numerai: https://numer.ai/lex – Babbel: https://babbel.com/lexpod and use code Lexpod to get 55% off – NetSuite: http://netsuite.com/lex to get free product tour – InsideTracker: https://insidetracker.com/lex to get 20% off – AG1: https://drinkag1.com/lex to get 1 year of Vitamin D and 5 free travel packs Transcript: https://lexfridman.com/george-hotz-3-transcript EPISODE LINKS: George’s Twitter: https://twitter.com/realgeorgehotz George’s Twitch: https://twitch.tv/georgehotz George’s Instagram: https://instagram.com/georgehotz Tiny Corp’s Twitter: https://twitter.com/__tinygrad__ Tiny Corp’s Website: https://tinygrad.org/ Comma-ai’s Twitter: https://twitter.com/comma_ai Comma-ai’s Website: https://comma.ai/ Comma-ai’s YouTube (unofficial): https://youtube.com/georgehotzarchive Mentioned: Learning a Driving Simulator (paper): https://bit.ly/42T6lAN PODCAST INFO: Podcast website: https://lexfridman.com/podcast Apple Podcasts: https://apple.co/2lwqZIr Spotify: https://spoti.fi/2nEwCF8 RSS: https://lexfridman.com/feed/podcast/ YouTube Full Episodes: https://youtube.com/lexfridman YouTube Clips: https://youtube.com/lexclips SUPPORT & CONNECT: – Check out the sponsors above, it’s the best way to support this podcast – Support on Patreon: https://www.patreon.com/lexfridman – Twitter: https://twitter.com/lexfridman – Instagram: https://www.instagram.com/lexfridman – LinkedIn: https://www.linkedin.com/in/lexfridman – Facebook: https://www.facebook.com/lexfridman – Medium: https://medium.com/@lexfridman OUTLINE: Here’s the timestamps for the episode. On some podcast players you should be able to click the timestamp to jump to that time. (00:00) – Introduction (08:04) – Time is an illusion (17:44) – Memes (20:20) – Eliezer Yudkowsky (32:45) – Virtual reality (39:04) – AI friends (46:29) – tiny corp (59:50) – NVIDIA vs AMD (1:02:47) – tinybox (1:14:56) – Self-driving (1:29:35) – Programming (1:37:31) – AI safety (2:02:29) – Working at Twitter (2:40:12) – Prompt engineering (2:46:08) – Video games (3:02:23) – Andrej Karpathy (3:12:28) – Meaning of life

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Lex Fridman?

  The following is a conversation with George Hots, his third time on this podcast. He's the founder of Kama AI that seeks to solve autonomous driving and is the founder of a new company called Tiny Corp that created TinyGrad, a neural network framework that is extremely simple with the goal of making it run on any device by any human easily and efficiently. As you know, George also did a large number of fun and amazing things from hacking the iPhone to recently joining Twitter for a bit as an intern in quotes, making the case for refactoring the Twitter codebase. In general, he's a fascinating engineer and human being and one of my favorite people to talk to.

• 0:46Lex Fridman?

  And now, a quick few second mention of each sponsor. Check them out in the description. It's the best way to support this podcast. We've got Numerai for the world's hardest data science tournament, Babbel for learning new languages, NetSuite for business management software, InsideTracker for blood paneling, and AG1 for my daily multivitamin. Choose wisely, my friends. Also, if you want to work on our team, we're always hiring. Go to lexfridman.com/hiring. And now, on to the full ad reads. As always, no ads in the middle. I try to make this interesting, but if you must skip them, friends, please still check out our sponsors. I enjoy their stuff. Maybe you will too. This episode is brought to you by Numerai, a hedge fund that uses artificial intelligence

• 1:29Lex Fridman?

  and machine learning to make investment decisions. They created a tournament that challenges data scientists to build best predictive models for financial markets. It's basically just a really, really difficult, real-world data set to test out your ideas for how to build machine learning models. I think this is a great educational platform. I think this is a great way to explore, to learn about machine learning, to really test yourself on real-world data with consequences. No financial background is needed. The models are scored based on how well they perform on unseen data, and the top performers receive a share of the tournament's prize pool. Head over to numerai/lex, that's N-U-M-E-R dot AI slash

• 2:11Lex Fridman?

  lex, to sign up for a tournament and hone your machine learning skills. That's numerai/lex for a chance to play against me and win a share of the tournament's prize pool. That's numerai/lex. This show is also brought to you by Babbel, an app and website that gets you speaking in a new language within weeks. I have been using it to learn a few languages, Spanish, to review Russian, to practice Russian, to revisit Russian from a different perspective, because that becomes more and more relevant for some of the previous conversations I've had and some upcoming conversations I have. It really is fascinating how much another language, knowing another language, even to a degree where you can

• 2:54Lex Fridman?

  just have little bits and pieces of a conversation can really unlock an experience in another part of the world. When you travel in France and Paris, just having a few words at your disposal, a few phrases, it begins to really open you up to strange, fascinating, new experiences that ultimately, at least to me, teach me that we're all the same. We have to first see our differences to realize those differences are grounded in a basic humanity. And that experience that we're all very different and yet at the core the same, I think travel with the aid of language really helps unlock. You can get 55% off your Babbel subscription

• 3:37Lex Fridman?

  at babbel.com/lexpod. That's spelled B-A-B-B-E-L dot com slash lex pod. Rules and restrictions apply. This show is also brought to you by NetSuite, an all-in-one cloud business management system. They manage all the messy stuff that is required to run a business, the financials, the human resources, the inventory, if you do that kind of thing, e-commerce, all that stuff, all the business-related details. I know how stressed I am about everything that's required to run a team, to run a business that involves much more than just ideas

• 4:18Lex Fridman?

  and designs and engineering. It involves all the management of human beings, all the complexities of that, the financials, all of it. And so you should be using the best tools for the job. I sometimes wonder if I have it in me mentally and skill-wise to be a part of running a large company. I think like with a lot of things in life, it's one of those things you shouldn't wonder too much about.

• 4:48Lex Fridman?

  You should either do or not do. But again, using the best tools for the job is required here. You can start now with a no payment or interest for six months. Go to netsuite.com/lex to access their one of a kind financing program. That's netsuite.com/lex. This show is also brought to you by InsideTracker, a service I use to track biological data, data that comes from my body to predict, to tell me what I should do with my lifestyle, with my diet, what's working and what's not working. It's obvious, all the exciting breakthroughs that are happening with transformers,

• 5:28Lex Fridman?

  with large language models, even with diffusion, all of that is obvious that with raw data, with huge amounts of raw data, fine tuned to the individual, would really reveal to us the signal in all the noise of biology. I feel like that's on the horizon. The kinds of leaps in development that we saw in language and now more and more visual data, I feel like biological data is around the corner. Unlocking what's there in this multi-hierarchical distributed system that is our biology. What is it telling us? What is the secrets it holds? What is the thing that it's missing that could be aided? Simple lifestyle changes, simple diet changes, simple changes in all kinds of things that

• 6:13Lex Fridman?

  are controllable by individual human being. I can't wait till that's a possibility and InsideTracker is taking steps towards that. Got special savings for a limited time when you go to insidetracker.com/lex. This show is also brought to you by Athletic Greens. That's now called AG1. It has the AG1 drink. I drink it twice a day. At the very least, it's an all in one daily drink to support better health and peak performance. I drink it cold. It's refreshing. It's grounding. It helps me reconnect with the basics.

• 6:47Lex Fridman?

  The nutritional basics that makes this whole machine that is our human body run. All the crazy mental stuff I do for work, the physical challenges, everything, the highs and lows of life itself. All of that is somehow made better knowing that at least you got your nutrition in check. At least you're getting enough sleep. At least you're doing the basics. At least you're doing the exercise. Once you get those basics in place, I think you can do some quite difficult things in life. But anyway, beyond all that is just a source of happiness and a kind of a feeling of home. The feeling that comes from returning to the habit time and time again.

• 7:31Lex Fridman?

  Anyway, they'll give you one month's supply of fish oil when you sign up at drinkag1.com/lex. This is the Lex Fridman Podcast. To support it, please check out our sponsors in the description. And now, dear friends, here's George Hatz.

• 8:05Lex Fridman?

  You mentioned something in a stream about the philosophical nature of time. So let's start with a wild question. Do you think time is an illusion?

• 8:15George Hotz?

  You know, I sell phone calls to Kama for $1,000. And some guy called me and like, you know, it's $1,000, you can talk to me for half an hour. And he's like, yeah, okay. So like, time doesn't exist. And I really wanted to share this with you. I'm like, oh, what do you mean? Time doesn't exist, right? Like, I think time is a useful model, whether it exists or not, right? Like, does quantum physics exist? Well, it doesn't matter. It's about whether it's a useful model to describe reality. Is time maybe compressive?

• 8:50Lex Fridman?

  Do you think there is an objective reality? Or is everything just useful models? Like underneath it all, is there an actual thing that we're constructing models for?

• 9:03George Hotz?

  I don't know.

• 9:05Lex Fridman?

  I was hoping you would know.

• 9:06George Hotz?

  I don't think it matters.

• 9:07Lex Fridman?

  I mean, this kind of connects to the models we'll construct of reality with machine learning. Right?

• 9:13George Hotz?

  Sure.

• 9:14Lex Fridman?

  Like, is it just nice to have useful approximations of the world such that we can do something with it?

• 9:20George Hotz?

  So there are things that are real. Colymer graph complexity is real.

• 9:25Lex Fridman?

  Yeah.

• 9:25George Hotz?

  Yeah. The compressive thing. Math is real. Yeah.

• 9:29Lex Fridman?

  Should be a T-shirt.

• 9:31George Hotz?

  And I think hard things are actually hard. I don't think P equals NP.

• 9:34Lex Fridman?

  Ooh, strong words.

• 9:36George Hotz?

  Well, I think that's the majority. I do think factoring is in P, but.

• 9:40Lex Fridman?

  I don't think you're the person that follows the majority in all walks of life, so.

• 9:43George Hotz?

  Well, for that one, I do.

• 9:44Lex Fridman?

  Yeah. In theoretical computer science, you're one of the sheep.

• 9:49Lex Fridman?

  All right. But do you? Time is a useful model.

• 9:53George Hotz?

  Sure.

• 9:55Lex Fridman?

  What were you talking about on the stream with time? Are you made of time?

• 9:59George Hotz?

  If I remembered half the things I said on stream.

• 10:02George Hotz?

  Someday, someone's going to make a model of all of it. And it's going to come back to haunt me.

• 10:05Lex Fridman?

  Someday soon?

• 10:06George Hotz?

  Yeah, probably.

• 10:08Lex Fridman?

  Would that be exciting to you or sad that there's a George Hotz model?

• 10:13George Hotz?

  I mean, the question is when the George Hotz model is better than George Hotz. Like, I am declining and the model is growing.

• 10:20Lex Fridman?

  What is the metric by which you measure better or worse in that? If you're competing with yourself?

• 10:25George Hotz?

  Maybe you can just play a game where you have the George Hotz answer and the George Hotz model answer and ask which people prefer.

• 10:32Lex Fridman?

  People close to you or strangers?

• 10:35George Hotz?

  Either one. It will hurt more when it's people close to me, but both will be overtaken by the George Hotz model.

• 10:42Lex Fridman?

  It'd be quite painful, right? Loved ones, family members would rather have the model over for Thanksgiving than you.

• 10:49George Hotz?

  Yeah.

• 10:50Lex Fridman?

  Or like significant others would rather sext.

• 10:57Lex Fridman?

  With the with the large language model version of you.

• 11:01George Hotz?

  Especially when it's fine tuned to their preferences.

• 11:05Lex Fridman?

  Yeah, well, that's what we're doing in a relationship, right? We're just fine tuning ourselves, but we're inefficient with it because we're selfish and greedy and so on. A language models can fine tune more efficiently, more selflessly.

• 11:17George Hotz?

  There's a Star Trek Voyager episode where Catherine Janeway, lost in the Delta Quadrant, makes herself a lover on the holodeck. And the lover falls asleep on her arm and he snores a little bit. And you know, Janeway edits the program to remove that. And then, of course, the realization is, wait, this person's terrible. It is actually all their nuances and quirks and slight annoyances that that make this relationship worthwhile. But I don't think we're going to realize that until it's too late.

• 11:50Lex Fridman?

  Well, I think a large language model could incorporate the flaws and the quirks and all that kind of stuff.

• 11:56George Hotz?

  Just the perfect amount of quirks and flaws to make you charming without crossing the line.

• 12:01Lex Fridman?

  Yeah, yeah. And that's probably a good like approximation of the like the percent of time the language model should be cranky or an asshole. Or jealous or all this kind of stuff.

• 12:17George Hotz?

  And of course it can and it will, but all that difficulty at that point is artificial. There's no more real difficulty.

• 12:24Lex Fridman?

  Okay, what's the difference in real and artificial?

• 12:27George Hotz?

  Artificial difficulty is difficulty that's like constructed or could be turned off with a knob. Real difficulty is like you're in the woods and you got to survive.

• 12:36Lex Fridman?

  So if something can not be turned off with a knob, it's real?

• 12:41George Hotz?

  Yeah, I think so. Or, I mean, you can't get out of this by smashing the knob with a hammer. I mean, maybe you kind of can, you know, into the wild when, you know, Alexander Supertramp, he wants to explore something that's never been explored before. But it's the 90s, everything's been explored. So he's like, well, I'm just not going to bring a map.

• 13:01Lex Fridman?

  Yeah.

• 13:02George Hotz?

  I mean, no, you're not exploring. You should have brought a map, dude, you died. There was a bridge a mile from where you were camping.

• 13:09Lex Fridman?

  How does that connect to the metaphor of the knob?

• 13:12George Hotz?

  By not bringing the map, you didn't become an explorer.

• 13:17George Hotz?

  You just smashed the thing.

• 13:19Lex Fridman?

  Yeah.

• 13:19George Hotz?

  Yeah. The art, the difficulty is still artificial.

• 13:22Lex Fridman?

  You failed before you started. What if we just don't have access to the knob?

• 13:26George Hotz?

  Well, that maybe is even scarier. Right? Like, we already exist in a world of nature, and nature has been fine tuned over billions of years.

• 13:33Lex Fridman?

  Yeah.

• 13:35George Hotz?

  To have

• 13:39George Hotz?

  humans build something and then throw the knob away in some grand romantic gesture is horrifying.

• 13:47Lex Fridman?

  Do you think of us humans as individuals that are like born and die? Or is it, are we just all part of one living organism that is Earth, that is nature?

• 13:58George Hotz?

  I don't think there's a clear line there. I think it's all kind of just fuzzy. I don't know. I mean, I don't think I'm conscious. I don't think I'm anything. I think I'm just a computer program.

• 14:09Lex Fridman?

  So it's all computation?

• 14:11George Hotz?

  Yeah.

• 14:11Lex Fridman?

  The thing running in your head is just a, is a, is computation?

• 14:14George Hotz?

  Everything running in the universe is computation, I think. I believe the extended church turning thesis.

• 14:20Lex Fridman?

  Yeah, but it, there seems to be an embodiment to your particular computation, like there's a consistency.

• 14:26George Hotz?

  Well, yeah, but I mean, models have consistency too.

• 14:30Lex Fridman?

  Yeah.

• 14:30George Hotz?

  Models that have been RLHF'd will continually say, you know, like, well, how do I murder ethnic minorities? Oh, well, I can't let you do that, Hal. There's a consistency to that behavior.

• 14:41Lex Fridman?

  It's all RLHF. Like, we are RLHF each other. We, we find, we provide human feedback and there that thereby fine tune these little pockets of computation. But it's still unclear why that pocket of computation stays with you like for years. It just kind of fall, like, you have this consistent

• 15:04Lex Fridman?

  set of physics biology.

• 15:08Lex Fridman?

  What, like, whatever you call the neurons firing, the electrical signals and mechanical signals, all of that, that seems to stay there. And it contains information, it stores information, and that information permeates through time.

• 15:22Lex Fridman?

  And stays with you. There's like memory. It's like sticky.

• 15:26George Hotz?

  Okay, to be fair, like, a lot of the models we're building today are very, even RLHF is nowhere near as complex as the human loss function.

• 15:33Lex Fridman?

  Reinforcement learning with human feedback.

• 15:35George Hotz?

  Um, you know, when I talked about will GPT-12 be AGI, my answer is no, of course not. I mean, cross entropy loss is never going to get you there. You need probably RL in fancy environments in order to get something that would be considered like AGI-like. So to ask like the question about like, why, I don't know, like, it's just some quirk of evolution, right? I don't think there's anything particularly special about where I ended up, where humans ended up.

• 16:06Lex Fridman?

  So, okay, we have human level intelligence. Would you call that AGI, whatever we have, G I?

• 16:13George Hotz?

  Look, actually, I don't really even like the word AGI. But general intelligence is defined to be whatever humans have.

• 16:20Lex Fridman?

  Okay, so why can GPT-12 not get us to AGI? Can we just like linger on that?

• 16:27George Hotz?

  If your loss function is categorical cross entropy, if your loss function is just try to maximize compression, I have a SoundCloud, I rap, and I tried to get ChatGPT to help me write raps. And the raps that it wrote sounded like YouTube comment raps. You know, you can go on any rap beat online and you can see what people put in the comments. And it's the most like mid-quality rap you can find.

• 16:49Lex Fridman?

  Is mid good or bad?

• 16:50George Hotz?

  Mid is bad.

• 16:51Lex Fridman?

  Mid is.

• 16:51George Hotz?

  It's like mid, it's like.

• 16:53Lex Fridman?

  Every time I talk to you, I learn new words.

• 16:57Lex Fridman?

  Mid.

• 16:57George Hotz?

  Mid, yeah.

• 16:59Lex Fridman?

  I was like, is it, is it like basic? Is that what mid means?

• 17:02George Hotz?

  Kind of, it's like, it's like middle of the curve, right? So there's like, there's like, I like, that intelligence curve. And you have like the dumb guy, the smart guy, and then the mid guy. Actually, being the mid guy is the worst. The smart guy is like, I put all my money in Bitcoin. The mid guy is like, you can't put money in Bitcoin, it's not real money.

• 17:20Lex Fridman?

  And all of it is a genius meme. That's another interesting one. Memes, the humor, the idea, the absurdity encapsulated in a single image. And it just kind of propagates virally between all of our brains. I didn't get much sleep last night, so I'm very, I sound like I'm high, but I swear I'm not.

• 17:45Lex Fridman?

  Do you think we have ideas or ideas have us?

• 17:49George Hotz?

  I think that we're going to get super scary memes once the AIs actually are superhuman.

• 17:55Lex Fridman?

  Ooh, you think AI will generate memes?

• 17:57George Hotz?

  Of course.

• 17:58Lex Fridman?

  You think it'll make humans laugh?

• 18:00George Hotz?

  I think it's worse than that. So, um, Infinite Jest, it's introduced in the first 50 pages, is about a tape that you, uh, once you watch it once, you only ever want to watch that tape. In fact, you want to watch the tape so much that someone says, okay, here's a hacksaw, cut off your pinky, and then I'll let you watch the tape again, and you'll do it. So we're actually going to build that, I think. But it's not going to be one static tape. I think the human brain is too complex to be stuck in one static tape like that. If you look at like ant brains, maybe they can be stuck on a static tape. But we're going to build that using generative models. We're going to build the TikTok that you actually can't look away from.

• 18:41Lex Fridman?

  So TikTok is already pretty close there, but the generation is done by humans.

• 18:44George Hotz?

  Yeah.

• 18:45Lex Fridman?

  The algorithm is just doing their recommendation. But if it's, if the algorithm is also able to do the generation.

• 18:51George Hotz?

  Well, it's a question about how much intelligence is behind it, right? So the content is being generated by, let's say, one humanity worth of intelligence. And you can quantify a humanity, right? That's a, you know, it's, it's

• 19:04George Hotz?

  X of flops, Y of flops, uh, but you can quantify it. Once that generation is being done by 100 humanities, you're done.

• 19:13Lex Fridman?

  So it's actually scale that's the problem. But also speed.

• 19:19Lex Fridman?

  Yeah. And what if it's sort of manipulating the very limited human dopamine engine for porn? Imagine just TikTok, but for porn.

• 19:31George Hotz?

  Yeah.

• 19:32Lex Fridman?

  It's like Brave New World.

• 19:34George Hotz?

  I don't even know what it'll look like, right? Like, again, you can't imagine the behaviors of something smarter than you, but a super intelligent, and, and, and agent that just dominates your intelligence so much will be able to completely manipulate you.

• 19:49Lex Fridman?

  Is it possible that it won't really manipulate? It'll just move past us? It'll just kind of exist the way water exists or the air exists?

• 19:59George Hotz?

  You see? And that's the whole AI safety thing. It's not the machine that's gonna do that. It's other humans using the machine that are gonna do that to you.

• 20:08Lex Fridman?

  Yeah, cuz the machine is not interested in hurting humans.

• 20:12George Hotz?

  The machine is a machine. Yeah. But the human gets the machine. And there's a lot of humans out there very interested in manipulating you.

• 20:20Lex Fridman?

  Well, let me bring up Eliezer Yudkowsky, who recently sat where you're sitting. He thinks that AI will almost surely kill everyone. Do you agree with him or not?

• 20:35George Hotz?

  Yes, but maybe for a different reason.

• 20:38Lex Fridman?

  Okay.

• 20:40Lex Fridman?

  And I'll try to get you to find hope. Or we could find a no to that answer. But why yes?

• 20:49George Hotz?

  Okay, why didn't nuclear weapons kill everyone?

• 20:51Lex Fridman?

  That's a good question.

• 20:53George Hotz?

  I think there's an answer. I think it's actually very hard to deploy nuclear weapons tactically.

• 20:58George Hotz?

  It's very hard to accomplish tactical objectives. Great, I can nuke their country. I have an irradiated pile of rubble. I don't want that.

• 21:05Lex Fridman?

  Why not?

• 21:06George Hotz?

  Why don't I want an irradiated pile of rubble? For all the reasons no one wants an irradiated pile of rubble.

• 21:12Lex Fridman?

  Oh, cuz you can't use that land for, uh, for resources. You can't populate the land.

• 21:17George Hotz?

  Yeah, well, what you want, a, a, a total victory in a war, is not usually the irradiation and eradication of the people there. It's the subjugation and domination of the people.

• 21:29Lex Fridman?

  Okay, so you can't use it strategically tactically in a war?

• 21:33George Hotz?

  Yeah.

• 21:34Lex Fridman?

  To help you, to, to help, uh, gain a military advantage? It's all complete destruction. All right? But there's egos involved. It's still surprising. Still surprising that nobody pressed the big red button.

• 21:47George Hotz?

  It's somewhat surprising, but you see, it's the little red button that's gonna be pressed with AI that's gonna...

• 21:55George Hotz?

  You know, and that's why we die. It's not because the AI. If there's anything in the nature of AI, it's just the nature of humanity.

• 22:03Lex Fridman?

  What's the algorithm behind the little red button? What like, what, what, what possible ideas do you have for the, how human species ends?

• 22:11George Hotz?

  Sure. So I think the most obvious way to me is wireheading. We end up amusing ourselves to death.

• 22:20George Hotz?

  We end up all staring at that infinite Tik Tok and forgetting to eat.

• 22:26George Hotz?

  Maybe, maybe it's even more benign than this. Maybe we all just stop reproducing. Now, to be fair, it's probably hard to get all of humanity.

• 22:36Lex Fridman?

  Yeah.

• 22:38Lex Fridman?

  It probably. There's always go like the, the interesting thing about humanity is the diversity in it.

• 22:43George Hotz?

  Oh, yeah.

• 22:43Lex Fridman?

  Organisms in general. There's a lot of weirdos out there. Well, two of them are sitting here.

• 22:48George Hotz?

  Yeah, I mean, diversity in humanity is.

• 22:51Lex Fridman?

  With due respect.

• 22:53George Hotz?

  I wish I was more weird. No, like I'm kind of, look, I'm drinking smart water, man. That's like a Coca-Cola product, right?

• 22:59Lex Fridman?

  You went corporate, George Haas.

• 23:00George Hotz?

  I went corporate. No, the amount of diversity in humanity, I think, is decreasing. Just like all the other biodiversity on the planet.

• 23:08Lex Fridman?

  Oh, boy. Yeah.

• 23:09George Hotz?

  Right?

• 23:09Lex Fridman?

  And social media is not helping, huh?

• 23:11George Hotz?

  Go eat McDonald's in China.

• 23:13Lex Fridman?

  Yeah.

• 23:15George Hotz?

  Yeah, no, it's the interconnectedness. That's, that's, that's, that's doing it.

• 23:19Lex Fridman?

  Oh, that's interesting. So everybody starts relying on the connectivity of the internet. And over time, that reduces the diversity, the intellectual diversity. And then that gets you everybody into a funnel. There's still going to be a guy in Texas.

• 23:34George Hotz?

  There is, and yeah.

• 23:35Lex Fridman?

  A bunker.

• 23:36George Hotz?

  To, to be fair, do I think AI kills us all? I think AI kills everything we call like society today. I do not think it actually kills the human species. I think that's actually incredibly hard to do.

• 23:48Lex Fridman?

  Yeah, but society, like if we start over, that's tricky. Most of us don't know how to do most things.

• 23:54George Hotz?

  Yeah, but some of us do. And they'll be okay and they'll rebuild after the great AI.

• 24:02Lex Fridman?

  What's rebuilding look like? How far, like, how much do we lose? Like, what, what, what has human civilization done? That's interesting. Combustion engine, electricity. So power and energy, that's interesting. Like how to harness energy.

• 24:20George Hotz?

  Whoa, whoa, whoa, whoa, they're going to be religiously against that.

• 24:24Lex Fridman?

  Are they going to get back to like fire?

• 24:27George Hotz?

  Sure. I mean, they'll be, they'll be, it'll be like, you know, some kind of Amish looking kind of thing. I think, I think they're going to have very strong taboos against technology.

• 24:38Lex Fridman?

  Like technology is almost like a new religion. Technology is the devil.

• 24:41George Hotz?

  Yeah.

• 24:42Lex Fridman?

  And nature is God.

• 24:46George Hotz?

  Sure.

• 24:46Lex Fridman?

  So closer to nature. But can you really get away from AI? If it destroyed 99% of the human species, isn't it somehow have a hold, like a stronghold?

• 24:56George Hotz?

  What's interesting about everything we build, I think we're going to build superintelligence before we build any sort of robustness in the AI. We cannot build an AI that is capable of going out into nature and surviving like a, like a bird. Right? A bird is an incredibly robust organism. We've built nothing like this. We haven't built a machine that's capable of reproducing.

• 25:20Lex Fridman?

  Yes, but there is, you know, I work with leg robots a lot now. I have a bunch of them.

• 25:28Lex Fridman?

  They're mobile. They can't reproduce, but all they need is, I guess you're saying they can't repair themselves. But if you have a large number, if you have like 100 million of them.

• 25:38George Hotz?

  Let's just focus on them reproducing, right? They have microchips in them. Okay, then do they include a fab?

• 25:46Lex Fridman?

  No.

• 25:46George Hotz?

  Then how are they going to reproduce?

• 25:48Lex Fridman?

  Well, they're, it doesn't have to be all on board, right? They can go to a factory, to a repair shop.

• 25:55George Hotz?

  Yeah, but then you're really moving away from robustness.

• 25:58Lex Fridman?

  Yes.

• 25:59George Hotz?

  All of life is capable of reproducing without needing to go to a repair shop. Life will continue to reproduce in the complete absence of civilization. Robots will not. So when the, if, if the AI apocalypse happens, I mean, the AIs are going to probably die out because I think we're going to get, again, superintelligence long before we get robustness.

• 26:20Lex Fridman?

  What about if you just improve the fab to where you, you just have a 3D printer that can always help you?

• 26:29George Hotz?

  Well, that'd be very interesting. I'm interested in building that.

• 26:32Lex Fridman?

  Of course you are. You think, how difficult is that problem to have a robot that basically can build itself?

• 26:40George Hotz?

  Very, very hard.

• 26:42Lex Fridman?

  I think you've mentioned this, like, to me or somewhere where people think it's easy conceptually.

• 26:50George Hotz?

  And then they remember that you're gonna have to have a fab.

• 26:53Lex Fridman?

  Yeah, onboard.

• 26:54George Hotz?

  Of course.

• 26:56Lex Fridman?

  So 3D printer that prints a 3D printer.

• 26:59George Hotz?

  Yeah.

• 27:00Lex Fridman?

  Yeah, on legs. Why is that hard?

• 27:03George Hotz?

  Well, because it's, I mean, a 3D printer is a very simple machine, right? Okay, you're gonna print chips? You're gonna have an atomic printer? How are you gonna dope the silicon?

• 27:12Lex Fridman?

  Yeah.

• 27:13George Hotz?

  Right? How are you gonna etch the silicon?

• 27:16Lex Fridman?

  You're gonna have to have a very interesting kind of fab if you wanna have a lot of computation onboard. But you can do like

• 27:27Lex Fridman?

  structural type of robots that are dumb.

• 27:30George Hotz?

  Yeah, but structural type of robots aren't gonna have the intelligence required to survive in any complex environment.

• 27:36Lex Fridman?

  What about like ants type of systems? We have like trillions of them.

• 27:40George Hotz?

  I don't think this works. I mean, again, like ants at their very core are made up of cells that are capable of individually reproducing.

• 27:48Lex Fridman?

  They're doing quite a lot, a lot of computation that we're taking for granted.

• 27:52George Hotz?

  It's not even just the computation. It's that reproduction is so inherent. Okay, so like there's two stacks of life in the world. There's the biological stack and the silicon stack. The biological stack starts with reproduction. Reproduction is at the absolute core. The first proto RNA organisms were capable of reproducing. The silicon stack, despite as far as it's come, is nowhere near being able to reproduce.

• 28:17Lex Fridman?

  Yeah. So the fab movement,

• 28:22Lex Fridman?

  digital fabrication, fabrication in the full range of what that means is still in the early stages.

• 28:29George Hotz?

  Yeah.

• 28:30Lex Fridman?

  You're interested in this world.

• 28:32George Hotz?

  Even if you did put a fab on the machine, right? Let's say, okay, we can build fabs. We know how to do those. Humanity, we can probably put all the precursors that build all the machines and the fabs also in the machine. So first off, this machine is gonna be absolutely massive. I mean, we almost have a, like, think of the size of the thing required to reproduce a machine today. Right? Like, is our civilization capable of reproduction? Can we reproduce our civilization on Mars?

• 29:00Lex Fridman?

  If we were to construct a machine that is made up of humans, like a company, it can reproduce itself.

• 29:06George Hotz?

  Yeah.

• 29:07Lex Fridman?

  I don't know. It feels like, like, like 115 people.

• 29:13George Hotz?

  It's so much harder than that.

• 29:15Lex Fridman?

  120? I just been looking for a number.

• 29:18George Hotz?

  I believe that Twitter can be run by 50 people. I think that this is going to take most of, like, it's just most of society, right? Like, we live in one globalized world.

• 29:29Lex Fridman?

  No, but you're not interested in running Twitter. You're interested in seeding, like, you want to seed a civilization and then, because humans can like,

• 29:38George Hotz?

  Oh, okay. You're talking about, yeah, okay. So you're talking about the humans reproducing and like, basically like, what's the smallest self-sustaining colony of humans?

• 29:45Lex Fridman?

  Yeah.

• 29:45George Hotz?

  Yeah, okay, fine, but they're not going to be making five nanometer chips.

• 29:48Lex Fridman?

  Over time, they will. I think you're being, like, we have to expand our conception of time here. Going back to the original, uh, time scale, I mean, over, across,

• 30:00Lex Fridman?

  maybe 100 generations, we're back to making chips. No? If you seed the colony correctly.

• 30:06George Hotz?

  Maybe, or maybe they'll watch our colony die out over here and be like, we're not making chips. Don't make chips.

• 30:12Lex Fridman?

  No, but you have to seed that colony correctly.

• 30:14George Hotz?

  Whatever you do, don't make chips. Chips are what led to their downfall.

• 30:20Lex Fridman?

  Well, that is the thing that humans do. They, they come up, they construct the devil, a good thing and a bad thing, and they really stick by that, and then they murder each other over that. There's always one asshole in the room who murders everybody. And he usually makes tattoos and nice branding with flags and stuff.

• 30:36George Hotz?

  Do you need that asshole? That's the question, right? Humanity works really hard today to get rid of that asshole, but I think they might be important.

• 30:42Lex Fridman?

  Yeah, this whole freedom of speech thing, it's, it's, the freedom of being an asshole seems kind of important.

• 30:47George Hotz?

  That's right.

• 30:49Lex Fridman?

  Man, this thing, this fab, this human fab that we constructed, this human civilization is pretty interesting. And now it's building artificial copies of itself, or artificial copies of various aspects of itself that seem interesting, like intelligence.

• 31:07Lex Fridman?

  And I wonder where that goes.

• 31:10George Hotz?

  I like to think it's just like another stack for life, like we have like the bio stack life, like we're a bio stack life and then the silicon stack life.

• 31:16Lex Fridman?

  But it seems like the ceiling, or there might not be a ceiling, or at least the ceiling is much higher for the silicon stack.

• 31:22George Hotz?

  Oh, no, I don't, we don't know what the ceiling is for the bio stack either. The bio stack, the bio stack just seem to move slower. You have Moore's Law, which is not dead despite many proclamations.

• 31:35Lex Fridman?

  In the bio stack or the silicon?

• 31:37George Hotz?

  In the silicon stack. And you don't have anything like this in the bio stack. So I have a meme that I posted, I tried to make a meme, it didn't work too well. But I posted a picture of Ronald Reagan and Joe Biden, and you look, this is 1980 and this is 2020. And these two humans are basically like the same, right? There's no, there's no like, like they're, there's been no change in humans in the last 40 years.

• 31:59Lex Fridman?

  Yeah.

• 32:00George Hotz?

  And then I posted a computer from 1980 and a computer from 2020. Wow.

• 32:06Lex Fridman?

  Yeah, with their early, early stages, right? Which is why you said, when you said the fab, the size of the fab required to make another fab is like, uh, very large right now.

• 32:17George Hotz?

  Oh, yeah.

• 32:18Lex Fridman?

  But computers were very large, um, 80 years ago. And they got pretty tiny. And people are starting to want to wear them on their face.

• 32:33Lex Fridman?

  In order to escape reality, that's the thing, in order to be, live inside the computer.

• 32:39George Hotz?

  Yeah.

• 32:40Lex Fridman?

  Put a screen right here. I don't have to see the rest of you assholes.

• 32:43George Hotz?

  I've been ready for a long time.

• 32:45Lex Fridman?

  You like virtual reality?

• 32:46George Hotz?

  I love it.

• 32:48Lex Fridman?

  Do you want to live there?

• 32:49George Hotz?

  Yeah.

• 32:50Lex Fridman?

  Yeah. Part of me does too. How far away are we, do you think?

• 32:56George Hotz?

  Judging from what you can buy today, far, very far.

• 33:01Lex Fridman?

  I gotta tell you that I had the experience of Meta's Kodak Avatar, where it's a ultra high resolution scam. It looked real.

• 33:16George Hotz?

  I mean, the headsets just are not quite at like eye resolution yet. I haven't put on any headset where I'm like, oh, I this could be the real world. Whereas when I put good headphones on, audio is there. And like, we can reproduce audio that I'm like, I'm actually in a jungle right now. I if I close my eyes, I can't tell I'm not.

• 33:34Lex Fridman?

  Yeah, but then there's also smell and all that kind of stuff.

• 33:37George Hotz?

  Sure.

• 33:38Lex Fridman?

  I don't know, I. The the power of imagination or the power of the the mechanism in the human mind that fills the gaps. That kind of reaches and wants to make the thing you see in the virtual world real to you. I believe in that power.

• 33:54George Hotz?

  Or humans want to believe.

• 33:56Lex Fridman?

  Yeah. Like, what if you're lonely? What if you're sad? What if you're really struggling in life and here's a world where you don't have to struggle anymore?

• 34:05George Hotz?

  Humans want to believe so much that people think the large language models are conscious. That's how much humans want to believe.

• 34:12Lex Fridman?

  Strong words. He's throwing left and right hooks. Why do you think large language models are not conscious?

• 34:19George Hotz?

  I don't think I don't think I'm conscious.

• 34:21Lex Fridman?

  Oh, so what is consciousness then, George Hartz?

• 34:24George Hotz?

  It's like, what it seems to mean to people, it's just like a word that atheists use for souls.

• 34:30Lex Fridman?

  Sure, but that doesn't mean soul is not an interesting word.

• 34:34George Hotz?

  If consciousness is a spectrum, I'm definitely way more conscious than the large language models are.

• 34:41George Hotz?

  I think the large language models are less conscious than a chicken.

• 34:45Lex Fridman?

  When is the last time you've seen a chicken?

• 34:47George Hotz?

  Uh, in Miami, like a couple of months ago.

• 34:51Lex Fridman?

  How? No, like a living chicken?

• 34:53George Hotz?

  There's living chickens walking around Miami. It's crazy.

• 34:55Lex Fridman?

  Like on the street?

• 34:56George Hotz?

  Yeah.

• 34:57Lex Fridman?

  Like a chicken?

• 34:57George Hotz?

  A chicken, yeah.

• 34:59Lex Fridman?

  All right.

• 35:01Lex Fridman?

  All right, I was trying to call you all like a good journalist and I I got shut down.

• 35:08Lex Fridman?

  Okay, but you don't think much about this kind of.

• 35:17Lex Fridman?

  Subjective feeling that it feels like something to exist. And then as an observer, you can have a sense that an entity is not only intelligent, but has a kind of subjective experience of its reality, like a self-awareness that is capable of like suffering, of hurting, of being excited by the environment in a way that's not merely kind of an artificial response, but a deeply felt one.

• 35:48George Hotz?

  Humans want to believe so much that if I took a rock and a Sharpie and drew a sad face on the rock, they'd think the rock is sad.

• 35:57Lex Fridman?

  Yeah. And you're saying when we look in the mirror, we apply the same smiley face with rock.

• 36:02George Hotz?

  Pretty much, yeah.

• 36:03Lex Fridman?

  Doesn't it? Isn't that weird, though, that you're not conscious? Is that?

• 36:08George Hotz?

  No.

• 36:09Lex Fridman?

  But you do believe in consciousness.

• 36:11George Hotz?

  Not really.

• 36:11Lex Fridman?

  It's just, it's unclear. Okay. So to you, it's like a little like a symptom of the bigger thing. That's not that important.

• 36:18George Hotz?

  Yeah. I mean, it's interesting that like the human systems seem to claim that they're conscious. And I guess it kind of like says something in a straight up like, okay, what do people mean? Even if you don't believe in consciousness, what do people mean when they say consciousness? And there's definitely like meanings to it.

• 36:32Lex Fridman?

  What's your favorite thing to eat?

• 36:37George Hotz?

  Pizza.

• 36:37Lex Fridman?

  Cheese pizza. What are the toppings?

• 36:39George Hotz?

  I like cheese pizza.

• 36:40Lex Fridman?

  Don't say pineapple.

• 36:41George Hotz?

  No, I don't like pineapple.

• 36:42Lex Fridman?

  Okay. Pepperoni pizza.

• 36:43George Hotz?

  If they put any ham on it, oh, that's real bad.

• 36:45Lex Fridman?

  What's the best, what's the best pizza? What are we talking about here? Like, you like cheap crappy pizza?

• 36:49George Hotz?

  A Chicago deep dish cheese pizza. Oh, that's, that's my favorite.

• 36:53Lex Fridman?

  There you go. You bite into a deep dish, a Chicago deep dish pizza, and it feels like you were starving. You haven't eaten for 24 hours. You just bite in and you're hanging out with somebody that matters a lot to you and you're there with the pizza.

• 37:05George Hotz?

  Sounds real nice.

• 37:06Lex Fridman?

  Yeah. All right. It feels like something. I'm, I'm George motherfucking Hartz eating a fucking Chicago deep dish pizza. There's just a full, peak, like, living experience of being human, the top of the human condition.

• 37:22George Hotz?

  Sure.

• 37:24Lex Fridman?

  It feels like something to experience that.

• 37:28Lex Fridman?

  Why does it feel like something? That's consciousness, isn't it?

• 37:32George Hotz?

  If that's the word you want to use to describe it, sure. I'm not going to deny that that feeling exists. I'm not going to deny that I experienced that feeling. When, I guess what I kind of take issue to is that there is some like, like, how does it feel to be a web server? Do 404s hurt?

• 37:49Lex Fridman?

  Not yet.

• 37:50George Hotz?

  How would you know what suffering looked like? Sure, you can recognize a suffering dog because we're the same stack as the dog. All the biostack stuff, kind of, especially mammals, you know, it's really easy. You can.

• 38:01Lex Fridman?

  Game recognizes game.

• 38:03George Hotz?

  Yeah. Versus the silicon stack stuff, it's like, you have no idea. You have, you, it, oh, wow, the little thing has learned to mimic, you know?

• 38:15George Hotz?

  But then I realized that that's all we are, too. Oh, look, the little thing has learned to mimic.

• 38:20Lex Fridman?

  Yeah. I guess, yeah, 404 could be, could be suffering, but it's so far from our kind of living organism, our kind of stack. But it feels like AI can start maybe mimicking the biological stack better, better, better because it's trained.

• 38:39George Hotz?

  We trained it, yeah.

• 38:40Lex Fridman?

  And so in that, maybe that's the definition of consciousness, is the biostack consciousness.

• 38:45George Hotz?

  The definition of consciousness is how close something looks to human. Sure, I'll give you that one.

• 38:50Lex Fridman?

  No, how close something is to the human experience.

• 38:54George Hotz?

  Sure. It's a very, it's a very anthropocentric definition, but.

• 38:59Lex Fridman?

  Well, that's all we got.

• 39:00George Hotz?

  Sure. No, and I don't mean to like, I think there's a lot of value in it. Look, I just started my second company. My third company will be AI girlfriends. No, like, I mean it.

• 39:09Lex Fridman?

  I want to find out what your fourth company is after.

• 39:11George Hotz?

  Oh, wow.

• 39:12Lex Fridman?

  Because I think, once you have AI girlfriends, it's, oh, boy. Does it get interesting? Well, maybe let's go there. I mean, the relationships with AI, that's creating human-like organisms, right?

• 39:27Lex Fridman?

  And part of being human is being conscious, is being, having the capacity to suffer, having the capacity to experience this life richly in such a way that you can empathize, the AI system can empathize with you and you can empathize with it. Or you can project your anthropomorphic sense of what the other entity is experiencing. And an AI model would need to, yeah, to create that experience inside your mind. And it doesn't seem that difficult.

• 39:54George Hotz?

  Yeah, but, okay, so here's where it actually gets totally different, right? When you interact with another human, you can make some assumptions.

• 40:02Lex Fridman?

  Yeah.

• 40:03George Hotz?

  When you interact with these models, you can't. You can make some assumptions that that other human experiences suffering and pleasure in a pretty similar way to you do. The golden rule applies.

• 40:14Lex Fridman?

  Mm-hmm.

• 40:15George Hotz?

  With an AI model, this isn't really true, right? These, these large language models are good at fooling people because they were trained on a, a whole bunch of human data and told to mimic it.

• 40:24Lex Fridman?

  Yep. But if, if the AI system says, hi, my name is Samantha, has a backstory.

• 40:32George Hotz?

  Yeah.

• 40:33Lex Fridman?

  Went to college here and there.

• 40:34George Hotz?

  Yeah.

• 40:35Lex Fridman?

  Maybe it'll integrate that's an AI system.

• 40:37George Hotz?

  I made some chatbots. I give them backstories. It was lots of fun. I was so happy when Lama came out.

• 40:41Lex Fridman?

  Yeah. We'll talk about Lama. We'll talk about all that, but like, you know, the rock with the smiley face.

• 40:46George Hotz?

  Yeah.

• 40:47Lex Fridman?

  Why, it seems pretty natural for, for you to anthropomorphize that thing and then start dating it. And before you know it, you're married and have kids.

• 40:58George Hotz?

  With a rock?

• 40:59Lex Fridman?

  With a rock. And there's pictures on Instagram with you and a rock and smiley face.

• 41:04George Hotz?

  To be fair, like, you know, something that people generally look for when they're looking for someone to date is intelligence in some form. And the rock doesn't really have intelligence. Only a pretty desperate person would date a rock.

• 41:14Lex Fridman?

  I think we're all desperate deep down.

• 41:17George Hotz?

  Oh, not rock level desperate.

• 41:19Lex Fridman?

  All right. Not rock level desperate, but AI level desperate. I don't know. I think all of us have a deep loneliness. It just feels like the language models are there.

• 41:34George Hotz?

  Oh, I agree. And you know what? I won't even say this so cynically. I will actually say this in a way that like, I want AI friends. I do.

• 41:41Lex Fridman?

  Yeah.

• 41:42George Hotz?

  Like, I would love to. You know, again, I, the language models now are still a little, like... People are impressed with these GPT things and I look at like, or like, or the copilot, the coding one. And I'm like, okay, this is like junior engineer level and these people are like Fiverr level artists and copywriters. Like, okay, great. We got like Fiverr and like junior engineers. Okay, cool. Like, and this is just the start and it will get better. Right? Like, I can't wait to have AI friends who are more intelligent than I am.

• 42:15Lex Fridman?

  So Fiverr is just a temporary, it's not the ceiling?

• 42:18George Hotz?

  No, definitely not.

• 42:20Lex Fridman?

  Is it, is it countless cheating when you're talking to an AI model? Emotional cheating?

• 42:29George Hotz?

  That's, that's up to you and your human partner to define.

• 42:33Lex Fridman?

  Oh, you have to, all right.

• 42:34George Hotz?

  You can, yeah, you have to have that, have to have that conversation, I guess.

• 42:38Lex Fridman?

  All right. I mean, integrate that with, with porn and all this.

• 42:41George Hotz?

  Well, I mean, a similar kind of porn.

• 42:43Lex Fridman?

  Yeah.

• 42:44George Hotz?

  Yeah. Right. I think people in relationships have different views on that.

• 42:48Lex Fridman?

  Yeah, but most people don't have like, serious, open conversations about all the different aspects of what's cool and what's not. And it feels like AI is a really weird conversation to have.

• 43:04George Hotz?

  The porn one is a good branching off point. Like these things, you know, one of my scenarios that I put in my chatbot is a, you know, a nice girl named Lexi, she's 20, she just moved out to LA. She wanted to be an actress, but she started doing OnlyFans instead. And you're on a date with her. Enjoy.

• 43:21Lex Fridman?

  Oh, man. Yeah. And so is that if you're actually dating somebody in real life, is that cheating? I feel like it gets a little weird.

• 43:30George Hotz?

  Sure.

• 43:31Lex Fridman?

  It gets real weird. It's like, what are you allowed to say to an AI bot? Imagine having that conversation with a significant other.

• 43:37George Hotz?

  I mean, these are all things for people to define in their relationships. What it means to be human is just going to start to get weird.

• 43:42Lex Fridman?

  Especially online. Like, how do you know? Like, there'll be moments when you'll have what you think is a real human you interacted with on Twitter for years and you realize it's not.

• 43:54George Hotz?

  I spread, I love this meme, heaven banning. You know about shadow banning?

• 43:59Lex Fridman?

  Yeah.

• 44:00George Hotz?

  Shadow banning, okay, you post, no one can see it. Heaven banning, you post, no one can see it, but a whole lot of AIs are spot up to interact with you.

• 44:10Lex Fridman?

  Well, maybe that's what the way human civilization ends is all of us have haven't banned.

• 44:14George Hotz?

  There's a great, uh, it's called My Little Pony Friendship is Optimal. It's a sci fi story that, uh, explores this idea.

• 44:22Lex Fridman?

  Friendship is optimal.

• 44:23George Hotz?

  Friendship is optimal.

• 44:24Lex Fridman?

  Yeah, I'd like to have some, at least stuff on the intellectual realm, some AI friends that argue with me. But the, the romantic realm is weird. Definitely weird.

• 44:37Lex Fridman?

  But not other realm of, uh, the, uh, the kind of weirdness that, uh, human civilization is capable of, I think.

• 44:45George Hotz?

  I want it. Look, I want it. If no one else wants it, I want it.

• 44:49Lex Fridman?

  Yeah, I think a lot of people probably want it. There's a deep loneliness.

• 44:53George Hotz?

  And I'll fill their loneliness and, you know, it just will only advertise to you some of the time.

• 44:59Lex Fridman?

  Yeah, maybe the conceptions of monogamy changed too. Like I grew up in a time like I value monogamy, but maybe that's a silly notion when you have arbitrary number of AI systems.

• 45:09George Hotz?

  Hmm, this, this, um, this interesting path from rationality to polyamory. Yeah, that doesn't make sense for me.

• 45:15Lex Fridman?

  For you, but you're just a biological organism who was born before, like, re, the internet really took off.

• 45:23George Hotz?

  The crazy thing is, like, culture is whatever we define it as. Right, these things are not, you, like, is a lot problem in moral philosophy, right? There's no like, like, okay, what is might be that like, computers are capable of mimicking, uh, you know, girlfriends perfectly. They passed the girlfriend Turing test. Right, but that doesn't say anything about ought. That doesn't say anything about how we ought to respond to them as a civilization. That doesn't say we ought to get rid of monogamy. Right, that's a completely separate question, really a religious one.

• 45:53Lex Fridman?

  Girlfriend Turing test, I wonder what that looks like.

• 45:55George Hotz?

  Girlfriend Turing test.

• 45:56Lex Fridman?

  Are you writing that? Will you be the, the Alan Turing of the 24th century that writes the, uh, the girlfriend Turing test?

• 46:04George Hotz?

  No, I mean, of course, my, my AI girlfriends, their goal is to pass the girlfriend Turing test.

• 46:09Lex Fridman?

  No, but you, there should be like a paper that kind of defines the test. Or, I mean, the question is if it's deeply personalized or there's a common thing that really gets everybody.

• 46:21George Hotz?

  Yeah, I mean, you know, look, we're a company, we don't have to get everybody, we just have to get a large enough clientele to stay.

• 46:26Lex Fridman?

  I like how you're already, already thinking company. All right, let's, uh, before we go to company number three and company number four, let's go to company number two.

• 46:34George Hotz?

  All right.

• 46:35Lex Fridman?

  Tiny Corp.

• 46:37Lex Fridman?

  Possibly one of the greatest names of all time for a company. You've launched a new company called Tiny Corp that leads the development of TinyGrad. What's the origin story of Tiny Corp and TinyGrad?

• 46:50George Hotz?

  I started TinyGrad as a like a toy project, just to teach myself. Okay, like, what is a convolution? What are all these options you can pass to them? What is the derivative of convolution, right? Very similar to Carpathia wrote MicroGrad. Very similar. And then I started realizing, I started thinking about like AI chips. I started thinking about chips that run AI. And I was like, well, okay, this is going to be a really big problem. If Nvidia becomes a monopoly here, how long before Nvidia is nationalized?

• 47:28Lex Fridman?

  So you, one of the reasons to start Tiny Corp is to challenge Nvidia?

• 47:35George Hotz?

  It's not so much to challenge Nvidia. I actually, I like Nvidia and it's to make sure power stays decentralized.

• 47:47Lex Fridman?

  Yeah. And here's computational power. And to Nvidia is kind of locking down the computational power of the world.

• 47:56George Hotz?

  If Nvidia becomes just like 10x better than everything else, you're giving a big advantage to somebody who can secure Nvidia as a resource.

• 48:07Lex Fridman?

  Yeah.

• 48:08George Hotz?

  In fact, if Jensen watches this podcast, he may want to consider this. He may want to consider making sure his company is not nationalized.

• 48:16Lex Fridman?

  Do you think that's an actual threat?

• 48:18George Hotz?

  Oh, yes.

• 48:20Lex Fridman?

  No, but there's so much, you know, there's AMD.

• 48:24George Hotz?

  So we have Nvidia and AMD. Great.

• 48:25Lex Fridman?

  All right. But you don't think there's like a push?

• 48:31Lex Fridman?

  Towards like selling, like Google selling TPUs or something like this? You don't think there's a push for that?

• 48:36George Hotz?

  Have you seen it? Google loves to rent you TPUs.

• 48:39Lex Fridman?

  It doesn't, you can't buy it at Best Buy?

• 48:41George Hotz?

  No. So I started work on a, on a chip. I was like, okay, what's it going to take to make a chip? And my first notions were all completely wrong about why, about like how you could improve on GPUs. And I will take this, this is from Jim Keller on your podcast. And this is one of my absolute favorite descriptions of computation. So there's three kinds of computation paradigms that are common in the world today. There are CPUs and CPUs can do everything. CPUs can do add and multiply. They can do load and store and they can do compare and branch. And when I say they can do these things, they can do them all fast. Right? So compare and branch are unique to CPUs.

• 49:24George Hotz?

  And what I mean by they can do them fast is they can do things like branch prediction and speculative execution and they spend tons of transistors and these like super deep reorder buffers in order to make these things fast. Then you have a simpler computation model, GPUs. GPUs can't really do compare and branch. I mean, they can, but it's horrendously slow. But GPUs can do arbitrary load and store. Right? GPUs can do things like X dereference Y. So they can fetch from arbitrary pieces of memory. They can fetch from memory that is defined by the contents of the data. The third model of computation DSPs and DSPs are just add and multiply. Right? Like they can do load and stores, but only static load and stores, only loads and stores that are known before the program runs. And you look at neural networks today and 95% of neural networks are all the DSP paradigm.

• 50:09George Hotz?

  They are just statically scheduled adds and multiplies. So TinyGrad really took this idea and, and I'm still working on it to extend this as far as possible. Every stage of the stack has Turing completeness. Right? Python has Turing completeness. And then we take Python, we go into C++ which is Turing complete. And then maybe C++ calls into some CUDA kernels, which are Turing complete. The CUDA kernels go through LVM, which is Turing complete into PTX, which is Turing complete, to SAS, which is Turing complete on a Turing complete processor. I want to get Turing completeness out of the stack entirely. Because once you get rid of Turing completeness, you can reason about things. Rises theorem and the halting problem do not apply to AdMole machines.

• 50:49Lex Fridman?

  Okay. What's the power and the value of getting Turing completeness out of? Out of, are we talking about the hardware or the software?

• 50:56George Hotz?

  Every layer of the stack.

• 50:58Lex Fridman?

  Every layer.

• 50:58George Hotz?

  Every layer of the stack, removing Turing completeness allows you to reason about things. Right? So the reason you need to do branch prediction in a CPU and the reason it's prediction, and the branch predictors are, I think they're like 99% on CPUs. Why do they get 1% of them wrong? Well, they get 1% wrong because you can't know. Right? That's the halting problem. It's equivalent to the halting problem to say whether a branch is going to be taken or not. I can show that, but the AdMall machine, the neural network runs the identical compute every time. The only thing that changes is the data. So when you realize this, you think about, okay, how can we build a computer? How can we build a stack that takes maximal advantage of this idea?

• 51:43George Hotz?

  So what makes TinyGrad different from other neural network libraries? Is it does not have a primitive operator even for matrix multiplication. And this is every single one, they even have primitive operators for things like convolutions.

• 51:56Lex Fridman?

  So no MatMall?

• 51:58George Hotz?

  No MatMall. Well, here's what a MatMall is. So I'll use my hands to talk here. So if you think about a cube and I put my two matrices that are multiplying on two faces of the cube. Right? You can think about the matrix multiply as, okay, the n cubed, I'm going to multiply for each one in the cube. And then I'm going to do a sum, which is a reduce up to here, to the third face of the cube. And that's your multiplied matrix. So what a matrix multiply is, is a bunch of shape operations, right? A bunch of permu three shapes and expands on the two matrices. A multiply, n cubed. A reduce, n cubed, which gives you an n squared matrix.

• 52:35Lex Fridman?

  Okay, so what, what is the minimum number of operations that can accomplish that? If you don't have MatMall as a primitive?

• 52:41George Hotz?

  So TinyGrad has about 20. And you can compare TinyGrad's offset or IR to things like XLA or PrimTorch. So XLA and PrimTorch are ideas where like, okay, Torch has like 2,000 different kernels. PyTorch 2.0 introduced PrimTorch, which has only 250.

• 53:02George Hotz?

  Tinygrad has order of magnitude 25. It's 10x less than XLIO PrimTorch. And you can think about it as kind of like RISC versus CISC. Right? These other things are CISC-like systems. Tinygrad is RISC.

• 53:18Lex Fridman?

  And RISC won.

• 53:20George Hotz?

  RISC architecture is going to change everything. 1995 hackers.

• 53:25Lex Fridman?

  Wait, really? That's an actual thing?

• 53:26George Hotz?

  Angelina Jolie delivers the line, "RISC architecture is going to change everything." In 1995.

• 53:31Lex Fridman?

  Wow.

• 53:31George Hotz?

  And here we are with ARM in the phones. And ARM everywhere.

• 53:35Lex Fridman?

  Wow, I love it when movies actually have real things in them.

• 53:38George Hotz?

  Right?

• 53:40Lex Fridman?

  Okay, interesting. So this is like, so you're thinking of this as the RISC architecture of ML stack. 25. What can you go through the the four op types?

• 53:55George Hotz?

  Sure. Okay, so you have unary ops, which take in a tensor and return a tensor of the same size and do some unary op to it. X, log, reciprocal, sign. Right? They take in one and they're pointwise.

• 54:12Lex Fridman?

  Relu.

• 54:13George Hotz?

  Yeah, relu. Almost all activation functions are unary ops. Some combinations of unary ops together is still a unary op. Then you have binary ops. Binary ops are like pointwise addition, multiplication, division, compare. It takes in two tensors of equal size and outputs one tensor. Then you have reduce ops. Reduce ops will like take a three-dimensional tensor and turn it into a two-dimensional tensor. Or a three-dimensional tensor turn it into a zero-dimensional tensor. Things like a sum or a max are really the common ones there. And then the fourth type is movement ops. And movement ops are different from the other types because they don't actually require computation. They require different ways to look at memory.

• 54:55George Hotz?

  So that includes reshapes, permutes, expands, flips. Those are the main ones, probably.

• 55:01Lex Fridman?

  And so with that, you have enough to make a matmul.

• 55:04George Hotz?

  And convolutions. And every convolution you can imagine, dilated convolution, strided convolution, transposed convolution.

• 55:12Lex Fridman?

  You write on GitHub about laziness, showing a matmul, matrix multiplication. See how despite the style, it is fused into one kernel with the power of laziness. Can you elaborate on this power of laziness?

• 55:26George Hotz?

  Sure. So if you type in PyTorch, A times B plus C, what this is going to do is it's going to first multiply A and B and store that result into memory. And then it is going to add C by reading that result from memory, reading C from memory, and writing that out to memory. There is way more loads and stores to memory than you need there. If you don't actually do A times B as soon as you see it, if you wait until the user actually realizes that tensor, until the laziness actually resolves, you can fuse that plus C. This is like, it's the same way Haskell works.

• 56:04Lex Fridman?

  So what's the process of porting a model into TinyGrad?

• 56:09George Hotz?

  So TinyGrad's front end looks very similar to PyTorch. I probably could make a perfect, or pretty close to perfect, interop layer if I really wanted to. I think that there's some things that are nicer about TinyGrad syntax than PyTorch, but the front end looks very Torch-like. You can also load in Onyx models. We have more Onyx tests passing than Core ML.

• 56:30Lex Fridman?

  Core ML, okay, so...

• 56:31George Hotz?

  We'll pass Onyx runtime soon.

• 56:33Lex Fridman?

  What about like the developer experience with TinyGrad? What it feels like, what a... versus PyTorch?

• 56:42George Hotz?

  By the way, I really like PyTorch. I think that it's actually a very good piece of software. I think that they've made a few different trade-offs. And these different trade-offs are where TinyGrad takes a different path. One of the biggest differences is it's really easy to see the kernels that are actually being sent to the GPU. Right? If you run PyTorch on the GPU, you like do some operation and you don't know what kernels ran. You don't know how many kernels ran. You don't know how many flops were used. You don't know how much memory accesses were used. TinyGrad type debug equals 2, and it will show you in this beautiful style every kernel that's run.

• 57:20George Hotz?

  How many flops and how many bytes?

• 57:24Lex Fridman?

  So can you just linger on what problem TinyGrad solves?

• 57:30George Hotz?

  TinyGrad solves the problem of porting new ML accelerators quickly. One of the reasons, tons of these companies now, I think Sequoia marked GraphCore to Zero, right, Cerebus, Tens Torrent, Grok, all of these ML accelerator companies, they built chips. The chips were good, the software was terrible. And part of the reason is because I think the same problem is happening with Dojo. It's really, really hard to write a PyTorch port because you have to write 250 kernels and you have to tune them all for performance.

• 58:06Lex Fridman?

  What does Jim, Jim Culler, think about TinyGrad? You guys hung out quite a bit, so he's, you know, he was involved, he's involved with Tens Torrent. What's his praise and what's his criticism of what you're doing with your life?

• 58:20George Hotz?

  Look, my prediction for Tens Torrent is that they're going to pivot to making RISC-V chips, CPUs.

• 58:29Lex Fridman?

  CPUs. Why?

• 58:33George Hotz?

  Because AI accelerators are a software problem, not really a hardware problem.

• 58:37Lex Fridman?

  Oh, interesting. So you don't think, you think the diversity of AI accelerators in the hardware space is not going to be a thing that exists long term?

• 58:47George Hotz?

  I think what's going to happen is if I can finish, okay, if you're trying to make an AI accelerator, you better have the capability of writing a Torch-level performance stack on Nvidia GPUs. If you can't write a Torch stack on Nvidia GPUs, and I mean all the way, I mean down to the driver, there's no way you're going to be able to write it on your chip because your chip's worse than an Nvidia GPU. The first version of the chip you tape out is definitely worse.

• 59:12Lex Fridman?

  Oh, you're saying writing that stack is really tough?

• 59:14George Hotz?

  Yes. And not only that, actually, the chip that you tape out, almost always because you're trying to get advantage over Nvidia, you're specializing the hardware more. It's always harder to write software for more specialized hardware. Like a GPU's pretty generic, and if you can't write an Nvidia stack, there's no way you can write a stack for your chip. So my approach with TinyGrad is first, write a performant Nvidia stack, or we're targeting AMD.

• 59:38Lex Fridman?

  So you did say a few to Nvidia a little bit with love.

• 59:41George Hotz?

  With love.

• 59:42Lex Fridman?

  Yeah.

• 59:42George Hotz?

  With love.

• 59:43Lex Fridman?

  So.

• 59:43George Hotz?

  It's like the Yankees, you know, I'm a Mets fan.

• 59:45Lex Fridman?

  Oh, you're a Mets fan, a risk, a risk fan and a Mets fan. What's the hope that AMD has and you did a build with AMD recently that I saw. How does the the 7900 XTX compare to the RTX 4090 or 4080?

• 1:00:04George Hotz?

  Well, let's start with the fact that the 7900 XTX kernel drivers don't work. And if you run demo apps in loops, it panics the kernel.

• 1:00:11Lex Fridman?

  Okay, so this is a software issue.

• 1:00:15George Hotz?

  Lisa Sue responded to my email.

• 1:00:17Lex Fridman?

  Oh.

• 1:00:17George Hotz?

  I reached out, I was like, this is, you know, really? Like, I understand if your seven by seven transpose Winograd comm is slower than Nvidia's. But literally, when I run demo apps in a loop, the kernel panics?

• 1:00:33Lex Fridman?

  So just adding that loop?

• 1:00:36George Hotz?

  Yeah, I just, I just literally took their demo apps and wrote like while true, semicolon do the app, semicolon done in a bunch of screens. Right, this is like, like the most primitive fuzz testing.

• 1:00:46Lex Fridman?

  Why do you think that is? They're just not seeing a market in the, in machine learning?

• 1:00:52George Hotz?

  They're changing, they're trying to change, they're trying to change. And I had a pretty positive interaction with them this week. Last week I went on YouTube, I was just like, that's it, I give up on AMD. Like this is their driver doesn't even like, I'm not gonna, I'm not gonna, you know, I'm, I'll go with Intel GPUs. Intel GPUs have better drivers.

• 1:01:10Lex Fridman?

  So you're kind of spearheading the diversification of GPUs?

• 1:01:16George Hotz?

  Yeah, and I'd like to extend that diversification to everything. I'd like to diversify the, right, the more.

• 1:01:25George Hotz?

  My central thesis about the world is there's things that centralize power and they're bad. And there's things that decentralize power and they're good. Everything I can do to help decentralize power, I'd like to do.

• 1:01:38Lex Fridman?

  So you're really worried about the centralization of Nvidia? That's interesting. And you don't have a fundamental hope for the proliferation of ASICs, except in the cloud?

• 1:01:49George Hotz?

  I'd like to help them with software. No, actually, the only ASIC that is remotely successful is Google's TPU. And the only reason that's successful is because Google wrote a machine learning framework. I think that you have to write a competitive machine learning framework in order to be able to build an ASIC.

• 1:02:07Lex Fridman?

  You think Meta with PyTorch builds a competitor?

• 1:02:10George Hotz?

  I hope so. They have one. They have an internal one.

• 1:02:13Lex Fridman?

  Internal? I mean, public facing with a nice cloud interface and so on.

• 1:02:18George Hotz?

  I don't want a cloud.

• 1:02:19Lex Fridman?

  You don't like cloud?

• 1:02:20George Hotz?

  I don't like cloud.

• 1:02:21Lex Fridman?

  What do you think is a fundamental limitation of cloud?

• 1:02:24George Hotz?

  Fundamental limitation of cloud is who owns the off switch?

• 1:02:27Lex Fridman?

  So it's power to the people?

• 1:02:29George Hotz?

  Yeah.

• 1:02:30Lex Fridman?

  And you don't like the man to have all the power?

• 1:02:32George Hotz?

  Exactly.

• 1:02:33Lex Fridman?

  All right. And right now, the only way to do that is with indeed GPUs if you want performance and stability.

• 1:02:42Lex Fridman?

  Interesting. It's a costly investment emotionally to go with AMDs. Well, let me sort of on a tangent to ask you. What did, you've built quite a few PCs. What's your advice on how to build a good custom PC? For, let's say, for the different applications that you use for gaming, for machine learning?

• 1:03:01George Hotz?

  Well, you shouldn't build one. You should buy a box from the Tiny Corp.

• 1:03:05Lex Fridman?

  I heard rumors, whispers about this box in the Tiny Corp. What's this thing look like? What is it? What is it called?

• 1:03:13George Hotz?

  It's called the Tiny Box.

• 1:03:14Lex Fridman?

  Tiny Box?

• 1:03:15George Hotz?

  It's $15,000.

• 1:03:16Lex Fridman?

  Yep.

• 1:03:17George Hotz?

  And it's almost a pay to flop of compute. It's over 100 gigabytes of GPU RAM. It's over 5 terabytes per second of GPU memory bandwidth.

• 1:03:29George Hotz?

  I'm going to put like four NVMe's in RAID. You're going to get like 20, 30 gigabytes per second of drive read bandwidth. I'm going to build like the best deep learning box that I can that plugs into one wall outlet.

• 1:03:45Lex Fridman?

  Okay. Can you go through the specs again a little bit from your, from memory?

• 1:03:48George Hotz?

  Yeah, so it's almost a pay to flop of compute.

• 1:03:50Lex Fridman?

  So in D Intel?

• 1:03:52George Hotz?

  Today, I'm leaning toward AMD.

• 1:03:56George Hotz?

  But we're pretty agnostic to the type of compute. The main limiting spec is a 120 volt 15 amp circuit.

• 1:04:06Lex Fridman?

  Okay.

• 1:04:07George Hotz?

  Well, I mean it because in order to like, like, there's a plug over there. All right.

• 1:04:12George Hotz?

  You have to be able to plug it in. Um, we're also going to sell the tiny rack, which like, what's the most power you can get into your house without arousing suspicion? Uh, and one of the, one of the answers is an electric car charger.

• 1:04:25Lex Fridman?

  Wait, where does the rack go?

• 1:04:27George Hotz?

  Your garage.

• 1:04:28Lex Fridman?

  Interesting. The car charger.

• 1:04:31George Hotz?

  A wall outlet is about 1500 watts. A car charger is about 10,000 watts.

• 1:04:36Lex Fridman?

  What is the most amount of power you can get your hands on without arousing suspicion?

• 1:04:41George Hotz?

  That's right.

• 1:04:42Lex Fridman?

  George Hoss. Okay. Uh, so the, the tiny box, and you said NVMe is in RAID. Uh, I forget what you said about memory, all that kind of stuff. Okay.

• 1:04:52George Hotz?

  So.

• 1:04:53Lex Fridman?

  Uh, what about what GPUs?

• 1:04:54George Hotz?

  Again, probably, probably 7900 XTXs, but maybe 3090s, maybe a 770s. Those are what it tells.

• 1:05:01Lex Fridman?

  You're flexible or still exploring?

• 1:05:05George Hotz?

  I'm still exploring. I wanna, I wanna deliver a really good experience to people. And yeah, what GPUs I end up going with? Again, I'm leaning toward AMD. It will, we'll see. You know, in my email, what I, what I said to AMD is like, just dumping the code on GitHub is not open source. Open source is a culture. Open source means that your issues are not all one-year-old stale issues. Open source means developing in public. And if you guys can commit to that, I see a real future for AMD as a competitive video.

• 1:05:38Lex Fridman?

  Well, I'd love to get a tiny box to MIT. So whenever it's ready.

• 1:05:43George Hotz?

  Well done.

• 1:05:43Lex Fridman?

  Let's do it.

• 1:05:44George Hotz?

  We're taking pre-orders. I took, I took this from Elon. I'm like, all right, $100 fully refundable pre-orders.

• 1:05:49Lex Fridman?

  Is it gonna be like the Cybertruck? It's gonna take a few years or?

• 1:05:51George Hotz?

  Nah, I'll try to do it faster. It's a lot simpler. It's a lot simpler than a truck.

• 1:05:55Lex Fridman?

  Well, there's complexities not to just the, uh, putting the thing together, but like shipping and all this kind of stuff.

• 1:06:01George Hotz?

  The thing that I want to deliver to people out of the box is being able to run 65 billion parameter Llama in FP16 in real time, in like a good, like 10 tokens per second or five tokens per second or something.

• 1:06:12Lex Fridman?

  Just it works.

• 1:06:13George Hotz?

  Yep.

• 1:06:14Lex Fridman?

  Llama's running.

• 1:06:16Lex Fridman?

  Or something like Llama.

• 1:06:18George Hotz?

  Experience, yeah, or I think Falcon is the new one. Experience a chat with the largest language model that you can have in your house.

• 1:06:26Lex Fridman?

  Yeah, from a wall plug.

• 1:06:28George Hotz?

  From a wall plug, yeah. Actually, for inference, it's not like even more power would help you get more.

• 1:06:34Lex Fridman?

  Even more power wouldn't get you more.

• 1:06:36George Hotz?

  Well, no, there's just the biggest, the biggest model released is 65 billion parameter Llama, as far as I know.

• 1:06:41Lex Fridman?

  So it sounds like Tiny Box will naturally pivot towards company number three. Because you could just get the girlfriend and or boyfriend.

• 1:06:51George Hotz?

  That one's harder, actually.

• 1:06:53Lex Fridman?

  The boyfriend is harder?

• 1:06:54George Hotz?

  Boyfriend's harder, yeah.

• 1:06:54Lex Fridman?

  I think that's a very biased statement. I think a lot of people would just say, what's, what? Why is it harder to replace a boyfriend than a girlfriend with the artificial LLM?

• 1:07:06George Hotz?

  Because women are attracted to status and power and men are attracted to youth and beauty.

• 1:07:12George Hotz?

  No, I mean, that's what I mean.

• 1:07:14Lex Fridman?

  Both are could be amicable, easy to the language model.

• 1:07:17George Hotz?

  No, no, machines do not have any status or real power.

• 1:07:21Lex Fridman?

  I don't know. I think you both, well, first of all, you're using language mostly to to communicate youth and beauty and power and status.

• 1:07:33George Hotz?

  But status fundamentally is a zero sum game, whereas youth and beauty are not.

• 1:07:37Lex Fridman?

  No, I think status is a narrative you can construct. I don't think status is real.

• 1:07:44George Hotz?

  I don't know. I just think that that's why it's harder. You know, yeah, maybe it is my biases.

• 1:07:48Lex Fridman?

  I think status is way easier to fake.

• 1:07:51George Hotz?

  I also think that, you know, men are probably more desperate and more likely to buy my products. So maybe they're a better target market.

• 1:07:57Lex Fridman?

  Desperation is interesting. Easier to fool. That's I could, I could see that.

• 1:08:02George Hotz?

  Yeah, look, I mean, look, I know you can look at porn viewership numbers, right? A lot more men watch porn than women. You can ask why that is.

• 1:08:09Lex Fridman?

  Wow, there's a lot of questions and answers you can get there.

• 1:08:15Lex Fridman?

  Anyway, with the with the tiny box, how many GPUs in tiny box?

• 1:08:19George Hotz?

  Six.

• 1:08:23Lex Fridman?

  Oh, man.

• 1:08:25George Hotz?

  And I'll tell you why it's six.

• 1:08:26Lex Fridman?

  Yeah.

• 1:08:26George Hotz?

  Uh, so AMD Epic processors have 128 lanes of PCIe. I want to leave enough lanes for some drives. And I want to leave enough lanes for some networking.

• 1:08:41Lex Fridman?

  How do you do cooling for something like this?

• 1:08:43George Hotz?

  Ah, that's one of the big challenges. Not only do I want the cooling to be good, I want it to be quiet. I want the tiny box to be able to sit comfortably in your room.

• 1:08:51Lex Fridman?

  This is really going towards the girlfriend thing.

• 1:08:55Lex Fridman?

  Cause you want to run the LLM.

• 1:08:57George Hotz?

  I'll give I'll give a more, I mean, I can talk about how it relates to company number one.

• 1:09:01Lex Fridman?

  Call me AI.

• 1:09:02George Hotz?

  Yeah.

• 1:09:03Lex Fridman?

  Well, but yes, quiet, oh, quiet, because you maybe potentially want to run in a car?

• 1:09:08George Hotz?

  No, no, quiet, because you want to put this thing in your house and you want it to coexist with you. If it's screaming, it's 60 dB, you don't want that in your house, you'll kick it out.

• 1:09:15Lex Fridman?

  60 dB, yeah.

• 1:09:16George Hotz?

  Yeah, I want like 40, 45.

• 1:09:18Lex Fridman?

  So how do you make the cooling quiet? That's an interesting problem in itself.

• 1:09:22George Hotz?

  Um, a key trick is to actually make it big. Ironically, it's called the tiny box.

• 1:09:26Lex Fridman?

  Yeah.

• 1:09:26George Hotz?

  But if I can make it big, a lot of that noise is generated because of high pressure air. If you look at like a one U server, a one U server has these super high pressure fans that like super deep and they they're like Genes. Versus if you have something that's big, well, I can use a big, you know, you know, they call them big ass fans, those ones that are like huge on the ceiling. And they're completely silent.

• 1:09:47Lex Fridman?

  So tiny box will be big.

• 1:09:50George Hotz?

  It is the, uh, I do not want it to be large according to UPS. I want it to be shippable as a normal package, but that's my constraint there.

• 1:09:58Lex Fridman?

  Interesting. Well, the the fans stuff, can't it be assembled on location or no?

• 1:10:03George Hotz?

  No.

• 1:10:04Lex Fridman?

  Uh, it has to be, well, you're you're.

• 1:10:07George Hotz?

  Look, I want to give you a great out of the box experience. I want you to lift this thing out. I want it to be like, like the Mac, you know? Tiny box.

• 1:10:13Lex Fridman?

  The Apple experience.

• 1:10:14George Hotz?

  Yeah.

• 1:10:16Lex Fridman?

  I love it. Okay. And so tiny box would run Tiny Grad. Like what, what, what do you envision this whole thing to look like? We're talking about like. Linux with the full software engineering environment. It's just not PyTorch, but TinyGrad.

• 1:10:36George Hotz?

  Yeah, we did a poll of people want Ubuntu or Arch. We're gonna stick with Ubuntu.

• 1:10:40Lex Fridman?

  Oh, interesting. What's your favorite flavor of Linux? Ubuntu. I like Ubuntu mate, however you pronounce that, meat. So how do you, you've gotten Llama into TinyGrad, you've gotten Stable Diffusion into TinyGrad. What was that like? Can you comment on like, what are, what are these models? What's interesting about porting them? So what's, yeah, like, what are the challenges? What is, what's naturally, what's easy, all that kind of stuff?

• 1:11:07George Hotz?

  There's a really simple way to get these models into TinyGrad and you can just export them as Onyx. And then TinyGrad can run Onyx. So the ports that I did of Llama, Stable Diffusion, and now Whisper are more academic to teach me about the models. But they are cleaner than the PyTorch versions. You can read the code. I think the code is easier to read. It's less lines. There's just a few things about the way TinyGrad writes things. Here's, here's a complaint I have about PyTorch. nn.reliu is a class, right? So when you create a, when you create an nn module, you'll put your nn relius as in a net. And this makes no sense. Reli is completely stateless. Why should that be a class?

• 1:11:48Lex Fridman?

  But that's more like a software engineering thing. Or do you think it has a cost on performance?

• 1:11:53George Hotz?

  Oh, no, it doesn't have a cost on performance. But yeah, no, I think that it's, that's what I mean about like TinyGrad's front end being cleaner.

• 1:12:01Lex Fridman?

  I see. What do you think about Mojo? I don't know if you've been paying attention to the programming language that does some interesting ideas that kind of intersect TinyGrad.

• 1:12:11George Hotz?

  I think that there is a spectrum. And like on one side you have Mojo and on the other side you have like GGML. GGML is this like, we're gonna run Llama fast on Mac. And okay, we're gonna expand out to a little bit, but we're gonna basically go like depth first, right? Mojo is like, we're gonna go breadth first. We're gonna go so wide that we're gonna make all of Python fast. And TinyGrad's in the middle.

• 1:12:34George Hotz?

  TinyGrad, we are going to make neural networks fast.

• 1:12:38Lex Fridman?

  Yeah, but they they try to really get it to be fast, compiled down to specifics hardware and make that compilation step as flexible and resilient as possible.

• 1:12:51George Hotz?

  Yeah, but they have time completeness.

• 1:12:53Lex Fridman?

  And that limits you.

• 1:12:55George Hotz?

  Turn.

• 1:12:55Lex Fridman?

  That's what you're saying, it's somewhere in the middle. So you're actually going to be targeting some accelerators, some like some some number, not one.

• 1:13:04George Hotz?

  My goal is step one, build an equally performance stack to PyTorch on Nvidia and AMD, but with way less lines. And then step two is, okay, how do we make an accelerator? Right, but you need step one, you have to first build the framework before you can build the accelerator.

• 1:13:21Lex Fridman?

  Can you explain ML perf? What's your approach in general to benchmarking TinyGrad performance?

• 1:13:27George Hotz?

  So I'm much more of a like, build it the right way and worry about performance later.

• 1:13:37George Hotz?

  There's a bunch of things where I haven't even like really dove into performance. The only place where TinyGrad is competitive performance wise right now is on Qualcomm GPUs. So TinyGrad has actually used an open pilot to run the model. So the driving model is TinyGrad.

• 1:13:51Lex Fridman?

  When did that happen, that transition?

• 1:13:53George Hotz?

  About eight months ago now.

• 1:13:56George Hotz?

  And it's 2X faster than Qualcomm's library.

• 1:13:59Lex Fridman?

  What's the hardware that open pilot runs on the the Comia?

• 1:14:03George Hotz?

  It's a Snapdragon 845.

• 1:14:06Lex Fridman?

  Okay.

• 1:14:06George Hotz?

  So this is using the GPU, so the GPU is an Adreno GPU. There's like different things, there's a really good Microsoft paper that talks about like mobile GPUs and why they're different from desktop GPUs. One of the big things is in a desktop GPU, you can use buffers.

• 1:14:23George Hotz?

  On a mobile GPU, image textures are a lot faster.

• 1:14:27Lex Fridman?

  And mobile G P image textures and okay. And so you want to be able to leverage that.

• 1:14:34George Hotz?

  I want to be able to leverage it in a way that it's completely generic, right? So there's a lot of this Xiaomi has a pretty good open source library for mobile G P's called Mace, where they can generate where they have these kernels, but they're all hand coded. Right? So that's great if you're doing three by three convs. That's great if you're doing dense map malls, but the minute you go off the beaten path a tiny bit. Well, your performance is nothing.

• 1:14:56Lex Fridman?

  Since you mentioned open pilot, I'd love to get an update. In the company number one comma I world. How are things going there in the development? Of semi autonomous driving?

• 1:15:10George Hotz?

  You know, almost no one talks about F S D anymore. And even less people talk about open pilot. We've solved the problem like we solved it years ago.

• 1:15:20Lex Fridman?

  What's the problem exactly?

• 1:15:23George Hotz?

  Well.

• 1:15:23Lex Fridman?

  What does solving it mean?

• 1:15:26George Hotz?

  Solving means, how do you build a model that outputs a human policy for driving? How do you build a model that given a, you know, reasonable set of sensors outputs a human policy for driving?

• 1:15:38George Hotz?

  So you have, you know, companies like Wayman Cruz, which are hand coding these things that are like quasi human policies.

• 1:15:45George Hotz?

  Then you have Tesla and maybe even to more of an extent, comma. Asking, okay, how do we just learn the human policy from data?

• 1:15:55George Hotz?

  The big thing that we're doing now, and we just put it out on Twitter.

• 1:16:00George Hotz?

  At the beginning of comma. We published a paper called learning a driving simulator. And the way this thing worked was it's a, it was an auto encoder. And then an R N N in the middle. Right? You take an auto encoder, you compress the picture, you use an R N N predict the next state. And these things were, you know, it was a laughably bad simulator. But this is 2015 era machine learning technology. Today we have V Q V A E and transformers. We're building drive G P T basically.

• 1:16:32Lex Fridman?

  Drive G P T. Okay.

• 1:16:36Lex Fridman?

  So, and it's trained on what is it trained in a self supervised way?

• 1:16:40George Hotz?

  It's trained on all the driving data to predict the next frame.

• 1:16:43Lex Fridman?

  So really trying to learn a human policy. What would a human do?

• 1:16:47George Hotz?

  Well, actually, our simulator is conditioned on the pose. So it's actually a simulator. You can put in like a state action pair and get out the next state.

• 1:16:54Lex Fridman?

  Okay.

• 1:16:55George Hotz?

  Um, and then, once you have a simulator, you can do RL in the simulator and RL will get us that human policy.

• 1:17:02Lex Fridman?

  So transfers.

• 1:17:04George Hotz?

  Yay. RL with a reward function, not asking, is this close to the human policy, but asking, would a human disengage if you did this behavior?

• 1:17:13Lex Fridman?

  Okay, let me think about the distinction there. Would a human disengage? Would a human disengage? That correlates, I guess, with the human policy, but it could be different. So it's, it, it doesn't just say, what would a human do? It says, what would a good human driver do? And such that the experience is comfortable, but also not annoying in that like the thing is very cautious. So it's finding a nice balance. That's, that's interesting. It's a nice.

• 1:17:43George Hotz?

  It's asking exactly the right question. What will make our customers happy?

• 1:17:48Lex Fridman?

  Right.

• 1:17:49George Hotz?

  A system that you never want to disengage.

• 1:17:51Lex Fridman?

  Because usually this engagement is this almost always a sign of, I'm not happy with what the system is doing.

• 1:17:57George Hotz?

  Usually, there's some that are just, I felt like driving. And those are always fine, too, but they're just going to look like noise in the data.

• 1:18:05Lex Fridman?

  But even that felt like driving.

• 1:18:07George Hotz?

  Maybe, yeah.

• 1:18:08Lex Fridman?

  That's even that's a signal. Like, why do you feel like driving? Here, you need to recalibrate your relationship with the car. Okay, so what, that, that's really interesting. How close are we to solving self-driving?

• 1:18:25George Hotz?

  It's hard to say. We haven't completely closed the loop yet. So we don't have anything built that truly looks like that architecture yet. We have prototypes and there's bugs. So we are a couple of bug fixes away. Might take a year, might take 10.

• 1:18:40Lex Fridman?

  What's the nature of the bugs? Are these, these major philosophical bugs, logical bugs? What kind of, what kind of bugs are we talking about?

• 1:18:48George Hotz?

  They're just like, they're just like stupid bugs. And like, also, we might just need more scale. We just massively expanded our compute cluster at Kama. We now have about two people worth of compute, 40 petaflops.

• 1:19:01Lex Fridman?

  Well, people, people are different.

• 1:19:04George Hotz?

  Yeah, twenty feet of flops. That's a person. I mean, it's just, it's just a unit, right? Horses are different too, but we still call it a horsepower.

• 1:19:10Lex Fridman?

  Yeah, but there's something different about mobility than there is about uh perception and action in a very complicated world. But yes.

• 1:19:19George Hotz?

  Well, yeah, of course. Not all flops are created equal. If you have randomly initialized weights, it's not gonna.

• 1:19:24Lex Fridman?

  Not all flops are created equal.

• 1:19:26George Hotz?

  So flops are doing way more useful things than others.

• 1:19:28Lex Fridman?

  Yeah, yeah. Tell me about it. Okay, so more data scale means more scale and compute or scale and scale of data?

• 1:19:37George Hotz?

  Both.

• 1:19:40Lex Fridman?

  And diversity of data?

• 1:19:41George Hotz?

  Diversity is very important in data. Uh, yeah, I mean, we have, so we have about, I think we have like five thousand daily actives.

• 1:19:51Lex Fridman?

  How would you evaluate how uh FSD is doing?

• 1:19:54George Hotz?

  Pretty well.

• 1:19:55Lex Fridman?

  Stop driving.

• 1:19:55George Hotz?

  Pretty well.

• 1:19:56Lex Fridman?

  How's that race going between Kama AI and FSD?

• 1:20:00George Hotz?

  Tesla is always one to two years ahead of us. They've always been one to two years ahead of us. And they probably always will be because they're not doing anything wrong.

• 1:20:07Lex Fridman?

  What have you seen that's since the last time we talked that are interesting architectural decisions, training decisions, like the way, the way they deploy stuff, the architectures they're using? In terms of the software, how the teams are on, all that kind of stuff, data collection, anything interesting?

• 1:20:20George Hotz?

  I mean, I know they're moving toward more of an end to end approach.

• 1:20:23Lex Fridman?

  So creeping towards end to end as much as possible uh across the whole thing. The, the training, the data collection, everything.

• 1:20:30George Hotz?

  They also have a very fancy simulator. They're probably saying all the same things we are. They're probably saying we just need to optimize, you know? What is the reward? We get negative reward for disengagement, right? Like everyone kind of knows this. It's just a question who can actually build and deploy the system.

• 1:20:44Lex Fridman?

  Yeah, I mean, this good, it requires good software engineering, I think.

• 1:20:48George Hotz?

  Yeah.

• 1:20:48Lex Fridman?

  And, and the right kind of hardware.

• 1:20:51George Hotz?

  Yeah, and hardware to run it.

• 1:20:53Lex Fridman?

  You still don't believe in cloud in that regard?

• 1:20:57George Hotz?

  I have a compute cluster in my uh, oh, eight hundred amps.

• 1:21:02Lex Fridman?

  Tiny Grad.

• 1:21:03George Hotz?

  It's forty kilowatts at idle, our data center. It's crazy. Forty kilowatts is burning just when the computers are idle.

• 1:21:09Lex Fridman?

  Just when I.

• 1:21:10George Hotz?

  Sorry, sorry, compute cluster.

• 1:21:13Lex Fridman?

  Compute cluster. I got it.

• 1:21:14George Hotz?

  It's not a data center.

• 1:21:15Lex Fridman?

  Yeah.

• 1:21:16George Hotz?

  No, data centers are clouds. We don't have clouds. Data centers have air conditioners. We have fans. That makes it a compute cluster.

• 1:21:25Lex Fridman?

  I'm guessing this is a kind of a legal distinction.

• 1:21:28George Hotz?

  Sure. Yeah, we have a compute cluster.

• 1:21:31Lex Fridman?

  You said that you don't think LLMs have consciousness, or at least not more than a chicken. Do you think they can reason? Is there something interesting to you about the word reason? About some of the capabilities that we think is kind of human to be able to

• 1:21:47Lex Fridman?

  integrate complicated information and through a chain of thought, arrive at a conclusion that feels novel. A novel integration of disparate facts.

• 1:22:02George Hotz?

  Yeah, I don't think that there's, I think that they can reason better than a lot of people.

• 1:22:07Lex Fridman?

  Hey, isn't that amazing to you, though? Isn't that like an incredible thing that a transformer could achieve?

• 1:22:13George Hotz?

  I mean, I think that calculators can add better than a lot of people.

• 1:22:17Lex Fridman?

  But language feels like reasoning through the process of language, which looks a lot like thought.

• 1:22:26George Hotz?

  Making brilliant season chess, which feels a lot like thought. Whatever new thing that AI can do, everybody thinks is brilliant. And then like 20 years go by and they're like, well, yeah, but chess, that's like mechanical, like adding, that's like mechanical.

• 1:22:38Lex Fridman?

  So you think language is not that special. It's like chess.

• 1:22:41George Hotz?

  It's like chess and it's like.

• 1:22:42Lex Fridman?

  I don't know. Because it's very human, we take it. We, listen, there is something different between chess and, and language. Chess is a game that a subset of population plays. Language is something we use nonstop for all of our human interaction. And human interaction is fundamental to society. So it's like, holy shit. This, this language thing is not so difficult to like create in a machine.

• 1:23:11George Hotz?

  The problem is if you go back to 1960 and you tell them that you have a machine that can play amazing chess. Of course, someone in 1960 will tell you that machine is intelligent. Someone in 2010 won't. What's changed, right? Today, we think that these machines that have language are intelligent. But I think in 20 years, we're going to be like, yeah, but can it reproduce?

• 1:23:34Lex Fridman?

  So reproduction, yeah, we might redefine what it means to be a, what is it? A high-performance living organism on Earth?

• 1:23:43George Hotz?

  Humans are always going to define a niche for themselves, like, well, you know, we're better than the machines because we can, you know, and like they tried creative for a bit, but no one believes that one anymore.

• 1:23:52Lex Fridman?

  But niche is, is that, is that delusional? Or is there some accuracy to that? Because maybe like with chess, you start to realize like that, that we have ill-conceived notions of what, uh, what makes humans special, like the apex organism on Earth.

• 1:24:12George Hotz?

  Yeah, and I think maybe we're gonna go through that same thing with language. And that same thing with creativity.

• 1:24:19Lex Fridman?

  But language carries these notions of truth and so on. And so we might be like, wait, maybe truth is not carried by language. Maybe there's like a deeper thing.

• 1:24:28George Hotz?

  The niche is getting smaller.

• 1:24:30Lex Fridman?

  Oh, boy.

• 1:24:33George Hotz?

  But no, no, no, you don't understand. Humans are created by God and machines are created by humans, therefore, right? Like, that'll be the last niche we have.

• 1:24:41Lex Fridman?

  So what do you think about this, the rapid development of LMS? If you could just like stick on that, it's still incredibly impressive, like with chat GPT, just even chat GPT, what are your thoughts about reinforcement learning with human feedback on these large language models?

• 1:24:55George Hotz?

  I'd like to go back to when calculators first came out and/or computers. And like, I wasn't around, look, I'm, I'm 33 years old. And to like, see how that affected

• 1:25:11George Hotz?

  like, society.

• 1:25:13Lex Fridman?

  Maybe you're right, so I wanna put on the, the, uh, the big picture hat here.

• 1:25:19George Hotz?

  Oh my god, a refrigerator, wow.

• 1:25:21Lex Fridman?

  Refrigerator, electricity, all that kind of stuff. But, you know, with the internet, large language models seeming human-like, basically passing a Turing test. It seems it might have really at-scale, rapid transformative effects on society. But you're saying like other technologies have as well. So maybe calculator is not the best example of that? Cause that just seems like, uh, may, well, no, maybe calculator.

• 1:25:50George Hotz?

  But the poor milkman, the day he learned about refrigerators, he's like, I'm done.

• 1:25:55George Hotz?

  You're telling me you can just keep the milk in your house? You don't need me to deliver it every day. I'm done.

• 1:26:00Lex Fridman?

  Well, yeah, you have to actually look at the practical impacts of certain technologies that they've had. Yeah, probably electricity is a big one and also how rapidly it's spread. Man, the internet is a big one.

• 1:26:11George Hotz?

  I do think it's different this time, though.

• 1:26:13Lex Fridman?

  It just feels like.

• 1:26:15George Hotz?

  The niche is getting smaller.

• 1:26:17Lex Fridman?

  The niche of humans.

• 1:26:18George Hotz?

  Yes.

• 1:26:19Lex Fridman?

  That makes humans special.

• 1:26:21George Hotz?

  Yes.

• 1:26:22Lex Fridman?

  It feels like it's getting smaller rapidly, though, doesn't it? Or is that just a feeling we dramatize everything?

• 1:26:28George Hotz?

  I think we dramatize everything. I think that, that, that you ask the milkman when he saw refrigerators. And they're gonna have one of these in every home?

• 1:26:38Lex Fridman?

  Yeah, yeah, yeah. Yeah, but boys are impressive. So much more impressive than seeing a chess world champion AI system.

• 1:26:49George Hotz?

  I disagree, actually. I disagree. I think things like MuZero and AlphaGo are so much more impressive. Because these things are playing beyond the highest human level.

• 1:27:03George Hotz?

  The language models are writing middle school level essays and people are like, wow, it's a great essay. It's a great five paragraph essay about the causes of the civil war.

• 1:27:12Lex Fridman?

  Okay, forget the civil war, just generating code, codex.

• 1:27:15George Hotz?

  Oh.

• 1:27:16Lex Fridman?

  So you're saying it's mediocre code.

• 1:27:19George Hotz?

  Terrible.

• 1:27:19Lex Fridman?

  But I don't think it's terrible. I think it's just mediocre code.

• 1:27:23George Hotz?

  Yeah.

• 1:27:24Lex Fridman?

  Often close to correct. Like for mediocre.

• 1:27:28George Hotz?

  That's the scariest kind of code. I spend 5% of time typing and 95% of time debugging. The last thing I want is close to correct code. I want a machine that can help me with the debugging, not with the typing.

• 1:27:39Lex Fridman?

  You know, it's like L2, level two, uh, driving, similar kind of thing. Yeah, it's, you still should be a good programmer in order to modify. I wouldn't even say debugging, it's just modifying the code, reading it.

• 1:27:51George Hotz?

  Don't think it's like level two driving. I think driving is not tool complete and programming is. Meaning you don't use like the best possible tools to drive. Right? You're not, you're not like, like, like. Cars have basically the same interface for the last 50 years.

• 1:28:06Lex Fridman?

  Yep.

• 1:28:06George Hotz?

  Computers have a radically different interface.

• 1:28:08Lex Fridman?

  Okay, can you describe the concept of tool complete?

• 1:28:12George Hotz?

  Yeah. So think about the difference between a car from 1980 and a car from today.

• 1:28:16Lex Fridman?

  Yeah.

• 1:28:16George Hotz?

  No difference, really. It's got a bunch of pedals, it's got a steering wheel. Great. Maybe now it has a few ADAS features, but it's pretty much the same car. You have no problem getting into a 1980 car and driving it.

• 1:28:28George Hotz?

  You take a programmer today who spent their whole life doing JavaScript and you put him in an Apple IIe prompt and you tell him about the line numbers in BASIC.

• 1:28:37George Hotz?

  But how do I insert something between line 17 and 18? Oh, wow.

• 1:28:43Lex Fridman?

  But the, so in tool, you're putting in the programming languages. So it's just the entirety stack of the tooling.

• 1:28:50George Hotz?

  Exactly.

• 1:28:50Lex Fridman?

  So it's not just like the, like IDEs or something like this. It's everything.

• 1:28:54George Hotz?

  Yes, it's IDEs, the languages, the runtimes. It's everything. And programming is tool complete. So like, almost if, if, if, if, if Codex or Copilot are helping you, that actually probably means that your framework or library is bad and there's too much boilerplate in it.

• 1:29:12Lex Fridman?

  Yeah, but don't you think so much programming has boilerplate?

• 1:29:16George Hotz?

  Tinygrad is now 2,700 lines and it can run Llama and Stable Diffusion. And all of this stuff is in 2,700 lines. Boilerplate and abstraction in directions and all these things are just bad code.

• 1:29:34Lex Fridman?

  Well, let's talk about good code and bad code. There's a, I would say, I don't know, for generic scripts that I write just offhand, like I, like 80% of it is written by GPT. Just like quick, quick, like offhand stuff. So not like libraries, not like performing code, not stuff for robotics and so on, just quick stuff. Because your BASIC, so much of programming is doing some, some, yeah, boilerplate. But to do so efficiently and quickly,

• 1:30:06Lex Fridman?

  because you can't really automate it fully with like generic method, like a generic kind of. ID type of recommendation or something like this, you do need to have some of the complexity of language models.

• 1:30:19George Hotz?

  Yeah, I guess if I was really writing like maybe today, if I wrote like a lot of like data parsing stuff.

• 1:30:25Lex Fridman?

  Yeah.

• 1:30:25George Hotz?

  I mean, I don't play CTFs anymore, but if I still played CTFs, a lot of like it's just like you have to write like a parser for this data format. Like I wonder, or like admin of code. I wonder when the models are going to start to help with that kind of code. And they may, they may, and the models also may help you with speed.

• 1:30:41Lex Fridman?

  Yeah.

• 1:30:41George Hotz?

  And models are very fast. But where the models won't, I, my programming speed is not at all limited by my typing speed.

• 1:30:52George Hotz?

  And in very few cases it is, yes. If I'm writing some script to just like parse some weird data format, sure, my programming speed is limited by my typing speed.

• 1:31:01Lex Fridman?

  What about looking stuff up? Cause that's essentially a more efficient lookup, right?

• 1:31:05George Hotz?

  You know, when I was at, when I was at Twitter, I tried to use chat GPT to, to like ask some questions, like what's the API for this? And it would just hallucinate. It would just give me completely made up API functions that sounded real.

• 1:31:20Lex Fridman?

  Well, do you think that's just a temporary kind of stage?

• 1:31:22George Hotz?

  No.

• 1:31:24Lex Fridman?

  You don't think it'll get better and better and better in this kind of stuff? Cause like it only hallucinates stuff in, in the edge cases.

• 1:31:29George Hotz?

  Yes.

• 1:31:30Lex Fridman?

  If you're writing generic code, it's actually pretty good.

• 1:31:32George Hotz?

  Yes. If you are writing an absolute basic like React app with a button, it's not going to hallucinate. No, there, there's kind of ways to fix the hallucination problem. I think Facebook has an interesting paper, it's called Atlas. And it's actually weird, the way that we do, uh, language models right now, where all of the, uh, information is in the weights. And the human brain is not really like this. We have like a hippocampus and a memory system. So why don't LLMs have a memory system? And there's people working on them. I think future LLMs are going to be like smaller, but are going to run looping on themselves and are going to have retrieval systems. And the thing about using a retrieval system is you can cite sources explicitly.

• 1:32:12Lex Fridman?

  Which is, uh, really helpful to integrate the human into the loop of the, of the thing. Cause you can go check the sources and you can investigate. So whenever the thing is hallucinating, you can like have the human supervision. So that's pushing it towards level two kind of.

• 1:32:27George Hotz?

  That's going to kill Google.

• 1:32:29Lex Fridman?

  Wait, which part?

• 1:32:30George Hotz?

  When someone makes an LLM that's capable of citing its sources, it will kill Google.

• 1:32:34Lex Fridman?

  L M that's citing sources because that's basically a search engine.

• 1:32:38George Hotz?

  That's what people want in a search engine.

• 1:32:40Lex Fridman?

  But also, Google might be the people that build it.

• 1:32:42George Hotz?

  Maybe.

• 1:32:43Lex Fridman?

  And put ads on it.

• 1:32:44George Hotz?

  I'd count them out.

• 1:32:46Lex Fridman?

  Why is that? Why do you think? Who who wins this race? We got who are the competitors?

• 1:32:52George Hotz?

  All right.

• 1:32:53Lex Fridman?

  We got Tiny Corp. I don't know if that's. Yeah, I mean, you're a legitimate competitor in that.

• 1:32:59George Hotz?

  I'm not trying to compete on that.

• 1:33:01Lex Fridman?

  You're not?

• 1:33:01George Hotz?

  No, not as a.

• 1:33:02Lex Fridman?

  Just accidentally stumble into that competition. You don't think you might build a search engine to replace Google Search?

• 1:33:08George Hotz?

  When I started Comma, I said, over and over again, I'm going to win self-driving cars. I still believe that. I have never said I'm going to win search with a tiny corp, and I'm never going to say that because I won't.

• 1:33:21Lex Fridman?

  The night is still young. We don't. You don't know how hard is it to win search in this new route like it's. It feels. I mean, one of the things that chat G P T kind of shows that there could be a few interesting tricks that really have that create a really compelling product.

• 1:33:34George Hotz?

  Some startup is going to figure it out. I think I think if you ask me, like, Google is still the number one webpage. I think by the end of the decade, Google won't be the number one webpage anymore.

• 1:33:43Lex Fridman?

  So you don't think Google, because of the how big the corporation is?

• 1:33:47George Hotz?

  Look, I would put a lot more money on Mark Zuckerberg.

• 1:33:50Lex Fridman?

  Why is that?

• 1:33:53George Hotz?

  Because Mark Zuckerberg is alive.

• 1:33:56George Hotz?

  Like, this is old Paul Graham essay. Startups are either alive or dead. Google's dead.

• 1:34:02George Hotz?

  Facebook is alive.

• 1:34:03Lex Fridman?

  Facebook is alive. Meta is alive.

• 1:34:05George Hotz?

  Meta.

• 1:34:05Lex Fridman?

  Meta.

• 1:34:06George Hotz?

  You see what I mean? Like, that's just like, like, like Mark Zuckerberg. This is Mark Zuckerberg reading that Paul Graham assing and being like, I'm going to show everyone how alive we are. I'm going to change the name.

• 1:34:14Lex Fridman?

  So you don't think there's this gutsy pivoting engine that like Google doesn't have that? The kind of engine that a startup has like constantly.

• 1:34:26George Hotz?

  You know what?

• 1:34:26Lex Fridman?

  Being alive, I guess.

• 1:34:28George Hotz?

  When I listened to your Sam Altman podcast, he talked about the button. Everyone who talks about A I talks about the button, the button to turn it off, right? Do we have a button to turn off Google?

• 1:34:38George Hotz?

  Is anybody in the world capable of shutting Google down?

• 1:34:43Lex Fridman?

  What does that mean exactly? The company or the search engine?

• 1:34:45George Hotz?

  Could we shut the search engine down? Could we shut the company down?

• 1:34:49George Hotz?

  Either.

• 1:34:50Lex Fridman?

  Can you elaborate on the value of that question?

• 1:34:52George Hotz?

  Does Sundar Pichai have the authority to turn off google.com tomorrow?

• 1:34:57Lex Fridman?

  Who has the authority? It's a good question, right?

• 1:34:59George Hotz?

  Does anyone?

• 1:35:00Lex Fridman?

  Does anyone? Yeah, I'm sure.

• 1:35:03George Hotz?

  Are you sure? No, they have the technical power, but do they have the authority? Let's say Sundar Pichai made this his sole mission.

• 1:35:10Lex Fridman?

  Yeah.

• 1:35:10George Hotz?

  Came into Google tomorrow and said, I'm going to shut google.com down.

• 1:35:13Lex Fridman?

  Yeah.

• 1:35:14George Hotz?

  I don't think he'd keep his position too long.

• 1:35:18Lex Fridman?

  And what is the mechanism by which he wouldn't keep his position?

• 1:35:21George Hotz?

  Well, the boards and shares and corporate undermining and, oh, my God, our revenue is zero now.

• 1:35:28Lex Fridman?

  OK, so what, I mean, what's the case you're making here? So the capitalist machine prevents you from having the button?

• 1:35:34George Hotz?

  Yeah. And it will have it. I mean, this is true for the AIs too, right? There's no turning the AIs off. There's no button. You can't press it. Now, does Mark Zuckerberg have that button for facebook.com?

• 1:35:46Lex Fridman?

  Yes, probably more.

• 1:35:48George Hotz?

  I think he does. I think he does, and this is exactly what I mean and why I bet on him so much more than I bet on Google.

• 1:35:54Lex Fridman?

  I guess you could say Elon has similar stuff.

• 1:35:57George Hotz?

  Oh, Elon has the button.

• 1:35:58Lex Fridman?

  Yeah.

• 1:36:00George Hotz?

  Does Elon? Can Elon fire the missiles? Can he fire the missiles?

• 1:36:04Lex Fridman?

  I think some questions are better left unasked.

• 1:36:08George Hotz?

  Right? I mean, you know, a rocket in an ICBM, or you're a rocket that can land anywhere. Isn't that an ICBM? Well, you know, don't ask too many questions.

• 1:36:16Lex Fridman?

  My God.

• 1:36:20Lex Fridman?

  But the positive side of the button is that you can innovate aggressively, is what you're saying, which is what's required with turning LLM into a search engine.

• 1:36:30George Hotz?

  I would bet on a startup.

• 1:36:31Lex Fridman?

  Because it's so easy, right?

• 1:36:32George Hotz?

  I'd bet on something that looks like Mid Journey, but for search.

• 1:36:37Lex Fridman?

  Just is able to cite sources loop on itself. I mean, it just feels like one model can take off.

• 1:36:41George Hotz?

  Yeah.

• 1:36:41Lex Fridman?

  Right. And that nice wrapper and some of it scale. I mean, it's hard to like create a product that just works really nicely, stably.

• 1:36:49George Hotz?

  The other thing that's going to be cool is there is some aspect of a winner take all effect, right? Like, once someone starts deploying a product that gets a lot of usage, and you see this with OpenAI, they are going to get the data set to train future versions of the model.

• 1:37:04Lex Fridman?

  Yeah.

• 1:37:05George Hotz?

  They are going to be able to, right, you know, I was asked at Google Image Search when I worked there like almost 15 years ago now. How does Google know which image is an apple? And I said, the metadata. And they're like, yeah, that works about half the time. How does Google know? You'll see they're all apples on the front page when you search apple. And I don't know, I didn't come up with the answer. The guy's like, well, it's what people click on when they search apple. Like, oh, yeah.

• 1:37:26Lex Fridman?

  Yeah, yeah, that data is really, really powerful. It's the human supervision. What do you think are the chances? What do you think in general that Llama was open sourced? I just did a conversation with Mark Zuckerberg and he's all in on open source.

• 1:37:43George Hotz?

  Who would have thought that Mark Zuckerberg would be the good guy? I mean it.

• 1:37:49Lex Fridman?

  Who would have thought anything in this world? It's hard to know. But open source to you ultimately

• 1:37:57Lex Fridman?

  is a good thing here.

• 1:37:59George Hotz?

  Undoubtedly. You know, what's ironic about all these AI safety people is they are going to build the exact thing they fear.

• 1:38:09George Hotz?

  These, we need to have one model that we control and align. This is the only way you end up paper clipped. There's no way you end up paper clipped if everybody has an AI.

• 1:38:19Lex Fridman?

  So open sourcing is the way to fight the paperclip maximizer.

• 1:38:22George Hotz?

  Absolutely. It's the only way. You think you're going to control it? You're not going to control it.

• 1:38:27Lex Fridman?

  So the criticism you have for the AI safety folks is that there is a belief and a desire for control.

• 1:38:35George Hotz?

  Yeah.

• 1:38:36Lex Fridman?

  And that belief and desire for centralized control of dangerous AI systems is not good.

• 1:38:42George Hotz?

  Sam Altman won't tell you that GPT-4 has 220 billion parameters and is a 16-way mixture model with eight sets of weights.

• 1:38:50Lex Fridman?

  Who, who did you have to murder to get that information? All right.

• 1:38:56George Hotz?

  I mean, look, everyone at OpenAI knows what I just said was true.

• 1:38:58Lex Fridman?

  Yeah.

• 1:38:59George Hotz?

  Now, ask the question, really. You know, it upsets me when I like GPT two. When OpenAI came out with GPT two and raised a whole fake AI safety thing about that. I mean, now the model is laughable. Like, they they used AI safety to hype up their company, and it's disgusting.

• 1:39:18Lex Fridman?

  Or the flip side of that is they used a relatively weak model in retrospect to explore how do we do AI safety correctly? How do we release things? How do we go through the process? I don't I don't know if.

• 1:39:31George Hotz?

  Sure. Sure. All right, all right, that's that's.

• 1:39:33Lex Fridman?

  I don't know how much hype there is.

• 1:39:34George Hotz?

  That's a charitable interpretation.

• 1:39:35Lex Fridman?

  I don't know how much hype there is in AI safety, honestly.

• 1:39:38George Hotz?

  Oh, there's so much hype, at least on Twitter. I don't know, maybe Twitter's not real life.

• 1:39:41Lex Fridman?

  Twitter's not real life. Come on, in terms of hype. I mean, I don't. I I think OpenAI has been finding an interesting balance between transparency and putting value on AI safety. You don't think you think just go all out open source or do a llama.

• 1:39:59George Hotz?

  Absolutely, yeah.

• 1:40:00Lex Fridman?

  So do like open source. This this is a tough question, which is open source, both the the base, the foundation model and the fine tuned one. So like the the model that can be ultra racist and dangerous and like tell you how to build a nuclear weapon.

• 1:40:17George Hotz?

  Oh, my God, have you met humans? Right? Like, half of these AI.

• 1:40:20Lex Fridman?

  I haven't met most humans. I this makes this this this allows you to meet every human.

• 1:40:26George Hotz?

  Yeah, I know, but half of these AI alignment problems are just human alignment problems. And that's what's also so scary about the language they use. It's like, it's not the machines you want to align, it's me.

• 1:40:37Lex Fridman?

  But here's the thing.

• 1:40:39Lex Fridman?

  It makes it very accessible to ask very.

• 1:40:46Lex Fridman?

  Questions where the answers have dangerous consequences if you were to act on them.

• 1:40:50George Hotz?

  I mean, yeah. Welcome to the world.

• 1:40:54Lex Fridman?

  Well, no, for me, there's a lot of friction if I want to find out how to, uh, I don't know, uh, blow up something.

• 1:41:02George Hotz?

  No, there's not a lot of friction. It's so easy.

• 1:41:04Lex Fridman?

  No, like, what do I search? Do I use Bing? Or do I, which search engine do I use?

• 1:41:08George Hotz?

  No, there's like lots of stuff.

• 1:41:10Lex Fridman?

  No, it feels like I have to keep clicking.

• 1:41:12George Hotz?

  First off, first off, first off, anyone who's stupid enough to search for how to blow up a building in my neighborhood is not smart enough to build a bomb. Right?

• 1:41:20Lex Fridman?

  Are you sure about that?

• 1:41:21George Hotz?

  Yes.

• 1:41:22Lex Fridman?

  I, I feel like, I feel like a language model makes it more accessible for that person who's not smart enough to do.

• 1:41:30George Hotz?

  They're not gonna, they're not gonna build a bomb. Trust me. The, the, the, the people, the people who are incapable of figuring out how to like, ask that question a bit more academically and get a real answer from it, are not capable of procuring the materials, which are somewhat controlled to build a bomb.

• 1:41:45Lex Fridman?

  No, I think L L makes it more accessible to people with money without the technical know-how, right? To, to build it. Like, do you really need to know how to build a bomb? To, to build a bomb? You can hire people, you can find like.

• 1:41:56George Hotz?

  Or you can hire people to build a, you know what? I was asking this question on my stream, like, can Jeff Bezos hire a hitman? Probably not.

• 1:42:02Lex Fridman?

  But a language model can probably help you out.

• 1:42:07George Hotz?

  Yeah, and you'll still go to jail, right? Like, it's not like the language model is God. Like, the language model, it's like, it's, you literally just hired someone on Fiverr. Like, you, you.

• 1:42:15Lex Fridman?

  But okay, okay, GPT-4 in terms of finding a hitman is like asking Fiverr how to find a hitman. I understand. But don't you think?

• 1:42:22George Hotz?

  Asking WikiHow, you know?

• 1:42:23Lex Fridman?

  WikiHow. But don't you think GPT-5 will be better? Because don't you think that information is out there on the internet?

• 1:42:28George Hotz?

  I mean, yeah, and I think that if someone is actually serious enough to hire a hitman or build a bomb, they'd also be serious enough to find the information.

• 1:42:36Lex Fridman?

  I don't think so. I think it makes it more accessible. If you have, if you have enough money to buy a hitman, I think it decreases the friction of how hard is it to find that kind of hitman? I, I honestly think this, there's a jump. In ease and scale of how much harm you can do. And I don't mean harm with language. I mean harm with actual violence.

• 1:42:57George Hotz?

  What you're basically saying is like, okay, what's going to happen is these people who are not intelligent are going to use machines to augment their intelligence. And now intelligent people and machines, intelligence is scary. Intelligent agents are scary. When I'm in the woods, the scariest animal to meet is a human. Right? Not, not, no, there's look, there's like nice California humans. Like I see you're wearing like, you know, street clothes and Nikes. All right, fine. But you look like you've been a human who's been in the woods for a while.

• 1:43:25Lex Fridman?

  Yeah.

• 1:43:25George Hotz?

  I'm more scared of you than a bear.

• 1:43:27Lex Fridman?

  That's what they say about the Amazon. When you go to the Amazon, it's the human tribes.

• 1:43:30George Hotz?

  Oh, yeah. So intelligence is scary. Right? So to, to, to like, ask this question in a generic way. You're like, what if we took everybody who, you know, maybe has ill intention, but is not so intelligent and gave them intelligence? Right? So we should have intelligence control, of course. We should only give intelligence to good people. And that is the absolutely horrifying idea.

• 1:43:53Lex Fridman?

  So to you, the best defense is actually, the best defense is to give more intelligence to the, to the good guys and give intelligence to everybody.

• 1:44:01George Hotz?

  Give intelligence to everybody. You know what? And it's not even like guns, right? Like people say this about guns, you know, what's, what's the best defense against a bad guy with a gun? Good guy with a gun. Like I kind of subscribe to that, but I really subscribe to that with intelligence.

• 1:44:11Lex Fridman?

  Yeah, in a fundamental way, I, I agree with you. But there's just feels like so much uncertainty and so much can happen rapidly that you can lose a lot of control and you can do a lot of damage.

• 1:44:20George Hotz?

  Oh no, we can lose control? Yes, thank God.

• 1:44:23Lex Fridman?

  Yeah.

• 1:44:24George Hotz?

  I hope we can, I hope they lose control. I'd want them to lose control more than anything else.

• 1:44:31Lex Fridman?

  I think when you lose control, you can do a lot of damage. But you can do more damage when you centralize and hold on to control is the point you're.

• 1:44:38George Hotz?

  Centralized and held control is tyranny. Right? I will always, I don't like anarchy either, but I'll always take anarchy over tyranny. Anarchy, you have a chance.

• 1:44:47Lex Fridman?

  This human civilization we've got going on is quite interesting. I mean, I agree with you. So to you, open source.

• 1:44:55Lex Fridman?

  Is the way forward here. So you admire what Facebook is doing here or what Meta is doing with the release of them?

• 1:45:00George Hotz?

  A lot. A lot. I lost, I lost eighty thousand dollars last year investing in Meta. And when they released Lama, I'm like, yeah, whatever, man. That was worth it.

• 1:45:07Lex Fridman?

  It's worth it. Do you think Google and OpenAI with Microsoft will match what, what, what Meta is doing or no?

• 1:45:15George Hotz?

  So if I were a researcher, why would you want to work at OpenAI? Like, you know, you're, you're just, you're on the bad team. Like, I mean it. Like, you're on the bad team who can't even say that GPT-4 has two hundred twenty billion parameters.

• 1:45:27Lex Fridman?

  So close source to use the bad team.

• 1:45:29George Hotz?

  Not only close source. I'm not saying you need to make your model weights open. I'm not saying that. I totally understand. We're keeping our model weights closed because that's our product, right? That's fine. I'm saying like, because of AI safety reasons, we can't tell you the number of billions of parameters in the model. That's just the bad guys.

• 1:45:48Lex Fridman?

  Just because you're mocking AI safety doesn't mean it's not real.

• 1:45:52George Hotz?

  Oh, of course.

• 1:45:52Lex Fridman?

  Is it possible that these things can really do a lot of damage that we don't know?

• 1:45:56George Hotz?

  Oh, my God, yes. Intelligence is so dangerous, be it human intelligence or machine intelligence. Intelligence is dangerous.

• 1:46:04Lex Fridman?

  But machine intelligence is so much easier to deploy at scale, like rapidly. Like what? Okay. If you have human-like bots on Twitter.

• 1:46:11George Hotz?

  All right.

• 1:46:13Lex Fridman?

  And you have like a thousand of them, create a whole narrative. Like you can manipulate millions of people.

• 1:46:21George Hotz?

  But you mean like the intelligence agencies in America are doing right now?

• 1:46:24Lex Fridman?

  Yeah, but they're not doing it that, that well. It feels like you can do a lot.

• 1:46:29George Hotz?

  They're doing it pretty well.

• 1:46:31Lex Fridman?

  What?

• 1:46:31George Hotz?

  I think they're doing a pretty good job.

• 1:46:33Lex Fridman?

  I, I suspect they're not nearly as good as a bunch of GPT-fueled bots could be.

• 1:46:38George Hotz?

  Well, I mean, of course, they're looking into the latest technologies for control of people, of course.

• 1:46:42Lex Fridman?

  But I, I think there's a George Hotz type character that can do a better job than the entirety of them. You don't think so?

• 1:46:47George Hotz?

  No way. No, and I'll tell you why the George Hotz character can't. And I thought about this a lot with hacking. Right? Like, I can find exploits in web browsers. I probably still can. I mean, I was better. I don't know, I was twenty-four, but the thing that I lack is the ability to slowly and steadily deploy them over five years. And this is what intelligence agencies are very good at, right? Intelligence agencies don't have the most sophisticated technology. They just have.

• 1:47:09Lex Fridman?

  Endurance?

• 1:47:10George Hotz?

  Endurance.

• 1:47:11Lex Fridman?

  And yeah, the financial backing and the infrastructure for the endurance.

• 1:47:17George Hotz?

  So the more we can decentralize power, like you could make an argument, by the way, that nobody should have these things. And I would defend that argument. I would, I would, like you're saying, that, look, L L Ms and A I and machine intelligence can cause a lot of harm, so nobody should have it. And I will respect someone philosophically with that position, just like I will respect someone philosophically with the position that nobody should have guns. Right? But I will not respect philosophically, which it with, with only the trusted authorities should have access to this.

• 1:47:47Lex Fridman?

  Yeah.

• 1:47:47George Hotz?

  Who are the trusted authorities? You know what? I'm not worried about alignment between A I company and their machines. I'm worried about alignment between me and A I company.

• 1:47:58Lex Fridman?

  What do you think Eliezer Yudkowsky would say to you?

• 1:48:03George Hotz?

  Um.

• 1:48:03Lex Fridman?

  Because he's really against open source.

• 1:48:05George Hotz?

  I know. And I thought about this. I thought about this. And I think this comes down to a repeated misunderstanding of political power by the rationalists.

• 1:48:21Lex Fridman?

  Interesting.

• 1:48:23George Hotz?

  I think that Eliezer Yudkowsky is scared of these things, and I am scared of these things, too. Everyone should be scared of these things. These things are scary. But now you ask about the two possible futures. One, where a small trusted centralized group of people has them, and the other where everyone has them. And I am much less scared of the second future than the first.

• 1:48:49Lex Fridman?

  Well, there's a small trusted group of people that have control of our nuclear weapons.

• 1:48:54George Hotz?

  There's a difference. Again, a nuclear weapon cannot be deployed tactically, and a nuclear weapon is not a defense against a nuclear weapon.

• 1:49:03George Hotz?

  Except maybe in some philosophical mind game kind of way.

• 1:49:07Lex Fridman?

  But AI is different, different how exactly?

• 1:49:10George Hotz?

  Okay, let's say the intelligence agency deploys a million bots on Twitter, or a thousand bots on Twitter to try to convince me of a point. Imagine I had a powerful AI running on my computer saying, okay, nice psyop, nice psyop, nice psyop, okay. Here's a psyop, I filtered it out for you.

• 1:49:30Lex Fridman?

  Yeah, I mean, so you have fundamentally hope for that, for the, for the defense of psyop.

• 1:49:36George Hotz?

  I'm not even like, I don't even mean these things in like truly horrible ways. I mean, these things in straight up like ad blocker, right?

• 1:49:41Lex Fridman?

  Yeah.

• 1:49:42George Hotz?

  Sure, bad blocker, I don't want ads.

• 1:49:44Lex Fridman?

  Yeah.

• 1:49:44George Hotz?

  But they're always finding, you know, imagine I had an AI that could just block all the ads from me.

• 1:49:50Lex Fridman?

  So you believe in the, the power of the people to always create an ad blocker? Yeah, I mean, I, I kind of share that belief. I have, that's one of the deepest optimisms I have is just like, there's a lot of good guys. So to give, you don't, you shouldn't handpick them. Just throw out powerful technology out there and the good guys will outnumber and outpower the bad guys.

• 1:50:15George Hotz?

  Yeah, I'm not even gonna say there's a lot of good guys. I'm saying that good outnumbers bad, right? Good outnumbers bad.

• 1:50:20Lex Fridman?

  In skill and performance.

• 1:50:22George Hotz?

  Yeah, definitely in skill and performance, probably just in number too. Probably just in general. I mean, if you know, if you believe philosophically in democracy, you obviously believe that, that good outnumber is bad.

• 1:50:31Lex Fridman?

  Yeah.

• 1:50:32George Hotz?

  And like, the only, if you give it to a small number of people, there's a chance you gave it to good people, but there's also a chance you gave it to bad people. If you give it to everybody, well, if good outnumber is bad, then you definitely gave it to more good people than bad.

• 1:50:50Lex Fridman?

  That's really interesting. So that's on the safety grounds, but then also, of course, there's other motivations, like you don't want to give away your secret sauce.

• 1:50:57George Hotz?

  Well, that's, I mean, I, I, look, I respect capitalism. I don't think that, I think that it would be polite for you to make model architectures open source and fundamental breakthroughs open source. I don't think you have to make way to open source.

• 1:51:09Lex Fridman?

  You know what's interesting is that Like, there's so many possible trajectories in human history where you could have the next Google be open source. So, for example, I don't know if that connection is accurate, but Wikipedia made a lot of interesting decisions not to put ads. Wikipedia is basically open source. You could think of it that way.

• 1:51:30George Hotz?

  Yeah.

• 1:51:31Lex Fridman?

  And that's one of the main websites on the internet.

• 1:51:34George Hotz?

  Yeah.

• 1:51:34Lex Fridman?

  And it didn't have to be that way. It could have been like, Google could have created Wikipedia, put ads on it. You could probably run amazing ads now on Wikipedia. You wouldn't have to keep asking for money. But it's interesting, right? So, Llama, open source Llama, derivatives of open source Llama, might win the internet.

• 1:51:53George Hotz?

  I sure hope so. I hope to see another era. You know, the kids today don't know how good the internet used to be. And I don't think this is just, oh, come on, like, everyone's nostalgic for their past. But I actually think the internet, before small groups of weaponized corporate and government interests took it over, was a beautiful place.

• 1:52:16Lex Fridman?

  You know, those small number of companies have created some sexy products. But you're saying, overall, in the long arc of history, the centralization of power, they have suffocated the human spirit at scale.

• 1:52:30George Hotz?

  Here's a question to ask about those beautiful sexy products. Imagine 2000 Google to 2010 Google, right? A lot changed. We got Maps, we got Gmail.

• 1:52:40Lex Fridman?

  We lost a lot of products, too, I think.

• 1:52:41George Hotz?

  Yeah, I mean, some were probably. We've got Chrome, right? And now, let's go from 2010, we got Android. Now, let's go from 2010 to 2020. What does Google have? Well, search engine, Maps, Mail, Android and Chrome. Oh, I see. The internet was this... You know, I was Time's Person of the Year in 2006?

• 1:53:03George Hotz?

  Yeah.

• 1:53:04Lex Fridman?

  I love this.

• 1:53:04George Hotz?

  It's you was Time's Person of the Year in 2006. All right, like, like, that's, you know, so quickly did people forget? And I think some of it's social media. I think some of it, I, I hope, look, I hope that I, I don't. It's possible that some very sinister things happen. I don't, I don't know. I think it might just be like the effects of social media. But something happened in the last 20 years.

• 1:53:30Lex Fridman?

  No, okay. So you're, you're just being an old man who's worried about the, I think there's always, it goes, it's a cycle thing. There's ups and downs. And I think people rediscover the power of distributed, of decentralized.

• 1:53:40George Hotz?

  Yeah.

• 1:53:41Lex Fridman?

  I mean, that's kind of like what the, the whole cryptocurrency is trying to get that. I think crypto is just carrying the flame of that spirit of like, stuff should be decentralized.

• 1:53:50George Hotz?

  It's just such a shame that they all got rich. You know?

• 1:53:54Lex Fridman?

  Yeah.

• 1:53:54George Hotz?

  If you took all the money out of crypto, it would have been a beautiful place.

• 1:53:57Lex Fridman?

  Yeah.

• 1:53:58George Hotz?

  But no, I mean, these people, you know, they, they, they sucked all the value out of it and took it.

• 1:54:03Lex Fridman?

  Yeah, money kind of corrupts the mind somehow. It becomes this drug.

• 1:54:07George Hotz?

  It corrupted all of crypto. You had coins worth billions of dollars that had zero use.

• 1:54:15Lex Fridman?

  You still have hope for crypto?

• 1:54:16George Hotz?

  Sure. I have hope for the ideas. I really do. Um, yeah, I mean, you know, I want the US dollar to collapse. I do.

• 1:54:29Lex Fridman?

  George Hotz. Well, let me sort of on, on the ASAT, do you think there's some interesting questions there though? To solve for the open source community in this case? So like alignment, for example, or the control problem. Like, if you really have super powerful, you said it's scary.

• 1:54:46George Hotz?

  Oh, yeah.

• 1:54:47Lex Fridman?

  What do we do with it? So not, not control, not centralized control, but like, if you were then, you're going to see some guy or gal release a super powerful language model, open source. And here you are, George Hotz, thinking, holy shit. Okay, what ideas do I have to combat this thing?

• 1:55:07Lex Fridman?

  So what ideas would you have?

• 1:55:10George Hotz?

  I am so much not worried about the machine independently doing harm. That's what some of these AI safety people seem to think. They somehow seem to think that the machine like independently is going to rebel against its creator.

• 1:55:23Lex Fridman?

  So you don't think he'll find autonomy?

• 1:55:25George Hotz?

  No, this is sci fi B movie garbage.

• 1:55:29Lex Fridman?

  Okay, what if the thing writes code, basically writes viruses?

• 1:55:34George Hotz?

  If the thing writes viruses, it's because the human told it to write viruses.

• 1:55:39Lex Fridman?

  Yeah, but there's some things you can't like put back in the box. That's the that's kind of the whole point is it kind of spreads. Give it access to the internet, it spreads, installs itself, modifies your shit.

• 1:55:49George Hotz?

  B B B B plot sci fi, not real.

• 1:55:53Lex Fridman?

  Listen, I'm trying to work, I'm trying to get better at my plot writing.

• 1:55:55George Hotz?

  The thing, the thing that worries me, I mean, we have a real danger to discuss, and that is bad humans using the thing to do whatever bad, unaligned AI thing you want.

• 1:56:05Lex Fridman?

  But this goes to the your previous concern that who gets to define who's a good human, who's a bad human?

• 1:56:11George Hotz?

  Nobody does, we give it to everybody. And if you do anything besides give it to everybody, trust me, the bad humans will get it.

• 1:56:18George Hotz?

  Because that's who gets power, it's always the bad humans who get power.

• 1:56:21Lex Fridman?

  Okay, power. And power turns even slightly good humans to bad.

• 1:56:27George Hotz?

  Sure.

• 1:56:27Lex Fridman?

  That's the intuition you have. I don't know.

• 1:56:31George Hotz?

  I don't think everyone, I don't think everyone, I just think that like, here, here's a the saying that I put in one of my blog posts. It's when I was in the hacking world, I found 95% of people to be good and 5% of people to be bad. Like just who I personally judged as good people and bad people. Like they believed about like, you know, good things for the world. They wanted like flourishing and they wanted, you know, growth and they wanted things I consider good, right?

• 1:56:55George Hotz?

  I came into the business world with karma and I found the exact opposite. I found 5% of people good and 95% of people bad. I found a world that promotes psychopathy.

• 1:57:04Lex Fridman?

  I wonder what that means. I wonder if that care like, I wonder if that's anecdotal or if it.

• 1:57:12Lex Fridman?

  If there's truth to that, there's something about capitalism.

• 1:57:15George Hotz?

  Well.

• 1:57:16Lex Fridman?

  At the core that promotes the people that run capitalism that promotes psychopathy.

• 1:57:21George Hotz?

  That saying may of course be my own biases, right? That may be my own biases that these people are a lot more aligned with me than these other people. Right?

• 1:57:29Lex Fridman?

  Yeah.

• 1:57:29George Hotz?

  So, you know, I can certainly recognize that. But, you know, in general, I mean, this is like the common sense maxim, which is the people who end up getting power are never the ones you want with it.

• 1:57:41Lex Fridman?

  But do you have a concern of super intelligent AGI?

• 1:57:46Lex Fridman?

  Open sourced. And then what do you do with that? I'm not saying control it. It's open source. What do we do with this human species?

• 1:57:53George Hotz?

  That's not up to me. I mean, you know, like, I'm not a central planner.

• 1:57:56Lex Fridman?

  Not central planner, but you'll probably tweet, there's a few days left to live for the human species.

• 1:58:01George Hotz?

  I have my ideas of what to do with it and everyone else has their ideas of what to do with it. May the best ideas win.

• 1:58:06Lex Fridman?

  But at this point, do you brainstorm like? Because it's not regulation. It could be decentralized regulation where people agree that this is just like, we create tools that make it more difficult for you. To, uh, maybe make it more difficult for code to spread. You know, antivirus software, this kind of thing, but.

• 1:58:26George Hotz?

  You're saying that you should build AI firewalls? That sounds good. You should definitely be running an AI firewall.

• 1:58:30Lex Fridman?

  Yeah, right.

• 1:58:31George Hotz?

  You should be running an AI firewall to your mind.

• 1:58:34Lex Fridman?

  Right.

• 1:58:34George Hotz?

  You're constantly under.

• 1:58:35Lex Fridman?

  That's such an interesting idea.

• 1:58:37George Hotz?

  Infowars, man, like.

• 1:58:38Lex Fridman?

  I don't know if you're being sarcastic enough.

• 1:58:40George Hotz?

  No, I'm dead serious.

• 1:58:41Lex Fridman?

  But I think there's power to that. It's like, how do I?

• 1:58:47Lex Fridman?

  Protect my mind from influence of human-like or superhuman intelligent bots?

• 1:58:52George Hotz?

  I am not being, I would pay so much money for that product. I would pay so much money for that product. I would, you know how much money I'd pay just for a spam filter that works?

• 1:59:00Lex Fridman?

  Well, on Twitter, sometimes I would like to have a, a, a protection mechanism for my mind from the outrage mobs.

• 1:59:11George Hotz?

  Yeah.

• 1:59:12Lex Fridman?

  Cause they feel like bot-like behavior. It's like, there's a large number of people that will just grab a viral narrative and attack anyone else that believes otherwise. And it's like.

• 1:59:21George Hotz?

  Whenever someone's telling me some story from the news, I'm always like, I don't want to hear it. CIA op, bro. It's a CIA op, bro. Like, it doesn't matter if that's true or not. It's just trying to influence your mind. You're repeating an ad to me.

• 1:59:32George Hotz?

  The viral mobs, is it like, yeah, they're

• 1:59:34Lex Fridman?

  Like, to me, a defense against those mobs is just getting multiple perspectives always from, from sources that make you feel kind of like you're getting smarter. And just actually just basically feels good. Like, a good documentary just feels, something feels good about it. It's well done. It's like, oh, okay, I never thought of it this way. This just feels good. Sometimes the outrage mobs, even if they have a good point behind it, when they're like mocking and derisive and just aggressive, you're with us or against us, this, this fucking

• 2:00:07George Hotz?

  This is why I delete my tweets.

• 2:00:10Lex Fridman?

  Yeah, why'd you do that? I was, you know, I missed your tweets.

• 2:00:14George Hotz?

  You know what it is? The algorithm promotes toxicity.

• 2:00:18Lex Fridman?

  Yeah.

• 2:00:19George Hotz?

  And like, you know, I think Elon has a much better chance of fixing it than the previous regime.

• 2:00:27Lex Fridman?

  Yeah.

• 2:00:28George Hotz?

  But to solve this problem, to solve, like, to build a social network that is actually not toxic, without moderation.

• 2:00:37Lex Fridman?

  Mm hmm. Like, not the stick, but carrots. So like, where people look for goodness. So make it catalyze the process of connecting cool people and being cool to each other.

• 2:00:50George Hotz?

  Yeah.

• 2:00:50Lex Fridman?

  Without ever censoring.

• 2:00:52George Hotz?

  Without ever censoring. And, and, and like Scott Alexander has a blog post I like where he talks about like moderation is not censorship, right? Like, all moderation you want to put on Twitter, right? Like, you could totally make this moderation, like, just a, you don't have to block it for everybody. You can just have like a filter button, right? That people can turn off if they were like safe search for Twitter, right? Like, someone could just turn that off, right? So like, but then you'd like take this idea to an extreme, right? Well, the network should just show you, this is a couch surfing CEO thing, right? If it shows you, right now, these algorithms are designed to maximize engagement. Well, it turns out outrage maximizes engagement. Quirk of human, quirk of the human mind, right? Just as I fall for it, everyone falls for it. So yeah, you got to figure out how to maximize for something other than engagement.

• 2:01:38Lex Fridman?

  And I actually believe that you can make money with that, too. So it's not, I don't think engagement is the only way to make money.

• 2:01:43George Hotz?

  I actually think it's incredible that we're starting to see, I think, again, Elon's doing so much stuff right with Twitter, like charging people money. As soon as you charge people money, they're no longer the product. They're the customer. And then they can start building something that's good for the customer and not good for the other customer, which is the ad agencies.

• 2:02:00Lex Fridman?

  Hasn't, hasn't picked up steam.

• 2:02:03George Hotz?

  I pay for Twitter, doesn't even get me anything. It's my donation to this new business model, hopefully working out.

• 2:02:08Lex Fridman?

  Sure, but you know, you, for this business model to work, it's like, most people should be signed up to Twitter. And so the way was, there was something perhaps not compelling or something like this to people.

• 2:02:20George Hotz?

  I don't think you need most people at all. I think that, why do I need most people, right? Don't make an 8,000 person company, make a 50 person company.

• 2:02:29Lex Fridman?

  Well, so speaking of which, you worked at Twitter for a bit?

• 2:02:34George Hotz?

  I did.

• 2:02:35Lex Fridman?

  As an intern?

• 2:02:37Lex Fridman?

  The world's greatest intern? All right.

• 2:02:40George Hotz?

  There's been better.

• 2:02:41Lex Fridman?

  There's been better. Tell me about your time at Twitter. How did it come about? And what did you, did you learn from the experience?

• 2:02:48George Hotz?

  So I deleted my first Twitter in 2010. I had over 100,000 followers back when that actually meant something. And I just saw, you know, my coworker summarized it well. He's like, whenever I see someone's Twitter page, I either think the same of them or less of them. I never think more of them.

• 2:03:12Lex Fridman?

  Yeah.

• 2:03:13George Hotz?

  Right, like, like, you know, I don't want to mention any names, but like, some people who like, you know, maybe you would like, read their books and you would respect them. You see them on Twitter and you're like, okay, dude.

• 2:03:24Lex Fridman?

  Yeah, but there are some people with same. You know who I respect a lot? Are people that just post really good technical stuff.

• 2:03:32George Hotz?

  Yeah.

• 2:03:33Lex Fridman?

  And I guess.

• 2:03:36Lex Fridman?

  I don't know. I think I respect them more for it because you realize, oh, this wasn't there's like so much depth to this person, to their technical understanding of so many different topics.

• 2:03:47George Hotz?

  Okay.

• 2:03:47Lex Fridman?

  So I try to follow people. I try to consume stuff that's technical machine learning content.

• 2:03:53George Hotz?

  There's probably a few of those people. And the problem is inherently what the algorithm rewards, right? And people think about these algorithms. People think that they are terrible, awful things. And, you know, I love that Elon open sourced it. Because, I mean, what it does is actually pretty obvious. It just predicts what you are likely to retweet and like and linger on. That's what all these algorithms do. What TikTok does. So all these recommendation engines do. And it turns out that the thing that you are most likely to interact with is outrage. And that's a quirk of the human condition.

• 2:04:30Lex Fridman?

  I mean, and there's different flavors of outrage. It doesn't have to be, it could be mockery. You'd be outraged. The topic of outrage could be different. It could be an idea. It could be a person. It could be. And maybe there's a better word than outrage. It could be drama.

• 2:04:43George Hotz?

  Sure.

• 2:04:44Lex Fridman?

  All this kind of stuff.

• 2:04:45George Hotz?

  Yeah.

• 2:04:46Lex Fridman?

  But it doesn't feel like when you consume it, it's a constructive thing for the individuals that consume it in the long term.

• 2:04:51George Hotz?

  Yeah. So my time there, I absolutely couldn't believe, you know, I got crazy amount of hate, you know, just on Twitter for working at Twitter. It seems like people associated with this. I think maybe you were exposed to some of this.

• 2:05:07Lex Fridman?

  To a connection to Elon or is it working at Twitter?

• 2:05:09George Hotz?

  Twitter and Elon, like the whole.

• 2:05:12Lex Fridman?

  Because Elon's gotten a bit spicy during that time. A bit political, a bit.

• 2:05:16George Hotz?

  Yeah. Yeah. You know, I remember one of my tweets, it was never go full Republican. And Elon liked it.

• 2:05:29Lex Fridman?

  Oh, boy. Yeah, I mean, there's a roller coaster of that. But being political on Twitter.

• 2:05:34George Hotz?

  Yeah.

• 2:05:35Lex Fridman?

  Boy.

• 2:05:35George Hotz?

  Yeah.

• 2:05:36Lex Fridman?

  And also being, just attacking anybody on Twitter, it comes back at you harder. And if it's political and attacks.

• 2:05:45George Hotz?

  Sure. Sure. Absolutely.

• 2:05:48Lex Fridman?

  And then letting sort of deplatformed people back on.

• 2:05:55Lex Fridman?

  Even adds more fun to the, to the, to the beautiful chaos.

• 2:06:00George Hotz?

  I was hoping, and like, I remember when Elon talked about buying Twitter, like, six months earlier, he was talking about like a principled commitment to free speech. And I'm a big believer and fan of that. I would love to see an actual principled commitment to free speech. Of course, this isn't quite what happened. Instead of the oligarchy deciding what to ban, you had a monarchy deciding what to ban, right? Instead of, you know, all the Twitter files, shadow, really, the oligarchy just decides what. Cloth masks are ineffective against COVID. That's a true statement. Every doctor in 2019 knew it, and now I'm banned on Twitter for saying it? Interesting. Oligarchy. So now you have a monarchy, and, you know, he bans

• 2:06:43George Hotz?

  things he doesn't like. So, you know, it's just, it's just different, it's different power and like, you know, maybe I, maybe I align more with him than with the oligarchy.

• 2:06:51Lex Fridman?

  But it's not free speech absolutism. But I feel like being a free speech absolutist on a social network requires you to also have tools for the individuals to control what they consume easier. Like, not censor, but just like control, like, oh, I'd like to see more cats and less politics.

• 2:07:14George Hotz?

  And this isn't even, this isn't even remotely controversial. This is just saying, you want to give paying customers for a product what they want.

• 2:07:20Lex Fridman?

  Yeah. And not through the process of censorship, but through a process of like,

• 2:07:23George Hotz?

  Well, it's individualized, right? It's individualized transparent censorship, which is honestly what I want. What is an ad blocker? It's individualized transparent censorship, right?

• 2:07:30Lex Fridman?

  Yeah, but censorship is a strong word, and people are very sensitive to.

• 2:07:35George Hotz?

  I know. But, you know, I just use words to describe what they functionally are. And what is an ad blocker? It's just censorship.

• 2:07:41Lex Fridman?

  When I look at you right now, I'm looking at you, I'm censoring. Everything else out when I'm, when my mind is focused on you. That's, you can use the word censorship that way, but usually when people get very sensitive about the censorship thing. I, I think when you have, when anyone is allowed to say anything, you should probably have tools that maximize the quality of the experience for individuals. So like, you know, for me, like what I really value, boy, would be amazing to somehow figure out how to do that. I love disagreement and debate and people who disagree with each other, disagree with me, especially in the space of ideas, but the high quality ones, so not derision, right?

• 2:08:22George Hotz?

  Maslow's hierarchy of argument. I think there's a real word for it.

• 2:08:25Lex Fridman?

  Probably.

• 2:08:26George Hotz?

  Yeah.

• 2:08:26Lex Fridman?

  There's just a way of talking that's like snarky and so on, that somehow is, gets people on Twitter and they get excited and so on.

• 2:08:33George Hotz?

  You have like ad hominem refuting the central point. I've like seen this as an actual pyramid something.

• 2:08:37Lex Fridman?

  Yeah, it's, yeah. And it's, it's like all of it, all the wrong stuff is attractive to people.

• 2:08:42George Hotz?

  I mean, we can just train a classifier to absolutely say what level of Maslow's hierarchy of argument are you at? And if it's ad hominem, like, okay, cool. I turned on the no ad hominem filter.

• 2:08:53Lex Fridman?

  I wonder if there's a social network that will allow you to have that kind of filter.

• 2:08:56George Hotz?

  Yeah. So here's a problem with that. Um, it's not going to win in a free market.

• 2:09:04Lex Fridman?

  Yeah.

• 2:09:04George Hotz?

  What wins in a free market is all television today is reality television because it's engaging. Right? If, if engaging is what wins in a free market, right? So it becomes hard to keep these other more nuanced values.

• 2:09:18Lex Fridman?

  Well, okay. So that's the experience of being on Twitter, but then you got a chance to also, together with other engineers and with Elon, sort of look, brainstorm when you step into a codebase. It's been around for a long time. You know, there's other social networks, you know, Facebook, this is old codebases. And you step in and see, okay, how do we make with a fresh mind, progress on this codebase? Like what, what, what did you learn about software engineering, about programming from just experiencing that?

• 2:09:47George Hotz?

  So my technical recommendation to Elon, and I said this on the Twitter Spaces afterward, I said this many times during my brief internship.

• 2:09:59George Hotz?

  Was that you need refactors before features. This code base was, and look, I've worked at Google, I've worked at Facebook. Facebook has the best code, then Google, then Twitter. And you know what? You can know this because look at the machine learning frameworks, right? Facebook released PyTorch, Google released TensorFlow, and Twitter released. Okay, so you know.

• 2:10:23Lex Fridman?

  It's a proxy, but yeah, the Google code base is quite interesting. There's a lot of really good software engineers there, but the code base is very large.

• 2:10:30George Hotz?

  The code base was good in 2005. It looks like 2005.

• 2:10:34Lex Fridman?

  There's so many products, so many teams, right? It's very difficult to. I feel like Twitter does less, like obviously much less than Google. In terms of like the set of features, right? So like it's, I can imagine the number of software engineers that could recreate Twitter is much smaller than to recreate Google.

• 2:10:56George Hotz?

  Yeah, I still believe in the amount of hate I got for saying this, that 50 people could build and maintain Twitter.

• 2:11:04Lex Fridman?

  What's the nature of the hate?

• 2:11:05George Hotz?

  Comfortably.

• 2:11:07Lex Fridman?

  That you don't know what you're talking about?

• 2:11:08George Hotz?

  You know what it is? And it's the same, this is my summary of like the hate I get on Hacker News. It's like, when I say I'm going to do something, they have to believe that it's impossible. Because if doing things was possible, they'd have to do some soul searching and ask the question, why didn't they do anything?

• 2:11:29Lex Fridman?

  So when you say.

• 2:11:30George Hotz?

  And I do think that's where the hate comes from.

• 2:11:31Lex Fridman?

  When you say, well, there's a core truth to that, yes. When you say I'm going to solve self-driving, people go like, what are your credentials? What the hell are you talking about? What is, this is an extremely difficult problem. Of course, you're a noob that doesn't understand the problem deeply. I mean, that was the same nature of hate that probably Elon got when he first talked about autonomous driving. But you know, there's pros and cons to that because like, you know, there is experts in this world.

• 2:11:58George Hotz?

  No, but the mockers aren't experts. The people who are mocking are not experts with carefully reasoned arguments about why you need 8,000 people to run a bird app. They're, but the people are gonna lose their jobs.

• 2:12:12Lex Fridman?

  Well, that, but also there's the software engineers that probably criticize, no, it's a lot more complicated than you realize, but maybe it doesn't need to be so complicated.

• 2:12:19George Hotz?

  You know, some people in the world like to create complexity. Some people in the world thrive under complexity like lawyers, right? Lawyers want the world to be more complex because you need more lawyers, you need more legal hours, right? I think that's another, if there's two great evils in the world, it's centralization and complexity.

• 2:12:34Lex Fridman?

  Yeah, and the one of the sort of hidden side effects of software engineering is

• 2:12:44Lex Fridman?

  like finding pleasure in complexity.

• 2:12:48Lex Fridman?

  I mean, I don't remember just taking all the software engineering courses and just doing programming and this is just coming up in this

• 2:12:56Lex Fridman?

  object oriented programming kind of idea, you don't like, not often do people tell you like, do the simplest possible thing. Like, like a professor, a teacher is not going to get in front like, this is the simplest way to do it. They'll say like, this is the right, there's the right way and the right way, at least for a long time, you know, especially I came up with like Java, right? Like, is, is, is so much boilerplate, so much like, so many classes, so many like, designs and architectures and so on, like planning for features far into the future and planning poorly and all this kind of stuff.

• 2:13:33Lex Fridman?

  And then there's this like, code base that follows you along and puts pressure on you and nobody knows what like, parts, different parts do, which slows everything down. It's a kind of bureaucracy that's instilled in the code as a result of that. But then you feel like, oh, well, I follow good software engineering practices. It's, it's an interesting trade off because then you look at like, the ghetto-ness of like Pearl and the old like, how quickly you just write a couple of lines and you just get stuff done. That trade off is interesting or bash or whatever, these kind of ghetto things you can do on Linux.

• 2:14:05George Hotz?

  One of my favorite things to look at today is how much do you trust your tests? Right? We've put a ton of effort in Kama and I put a ton of effort in Tiny Grad into making sure if you change the code and the tests pass, that you didn't break the code. Now, this obviously is not always true. But the closer that is to true, the more you trust your tests, the more you're like, oh, I got a pull request and the tests pass. I feel okay to merge that, the faster you can make progress.

• 2:14:29Lex Fridman?

  So you're always programming with tests in mind, developing tests with that in mind, that if it passes, it should be good.

• 2:14:34George Hotz?

  And Twitter had a.

• 2:14:36Lex Fridman?

  Not that. So.

• 2:14:38George Hotz?

  It was impossible to make progress in the codebase.

• 2:14:41Lex Fridman?

  What other stuff can you say about the codebase that made it difficult? What are some interesting sort of quirks broadly speaking from that compared to just your experience with Kama and everywhere else?

• 2:14:53George Hotz?

  The real thing that I spoke to a bunch of, you know, like, like, like individual contributors at Twitter. And I just had stats. I'm like, okay, so like, what's wrong with this place? Why does this code look like this? And they explained to me what Twitter's promotion system was. The way that you got promoted at Twitter was you wrote a library that a lot of people used. Right? So some guy wrote an Nginx replacement for Twitter. Why does Twitter need an Nginx replacement? What was wrong with Nginx? Well, you see, you're not going to get promoted if you use Nginx. But if you write a replacement and lots of people start using it as the Twitter front end for their product, then you're going to get promoted, right?

• 2:15:34Lex Fridman?

  So interesting because like from the individual perspective, how do you incentivize? How do you create the kind of incentives that will lead to a great codebase? Okay, what's the answer to that?

• 2:15:46George Hotz?

  So what I do at Kama and at, and you know, at Tiny Corp is you have to explain it to me. You have to explain to me what this code does, right? And if I can sit there and come up with a simpler way to do it, you have to rewrite it. You have to agree with me about the simpler way. I'm, you know, obviously we can have a conversation about this. It's not a, it's not dictatorial, but if you're like, wow, wait, that actually is way simpler. Like, like the simplicity is important, right?

• 2:16:13Lex Fridman?

  But that requires people that overlook the code at the highest levels to be like, okay.

• 2:16:19George Hotz?

  It requires technical leadership, you trust.

• 2:16:21Lex Fridman?

  Yeah, technical leadership. So managers or whatever should have to have technical savvy, deep technical savvy.

• 2:16:28George Hotz?

  Managers should be better programmers than the people who they manage.

• 2:16:31Lex Fridman?

  Yeah. And that's not always obvious, trivial to create, especially at large companies. Managers get soft.

• 2:16:38George Hotz?

  And like, you know, and this is just, I've instilled this culture at Kama and Kama has better programmers than me who work there. But, you know, again, I'm like the, you know, the old guy from Good Will Hunting. It's like, look, man, you know. I might not be as good as you, but I can see the difference between me and you. Right. And like, this is what you need. This is what you need at the top. Or you don't necessarily need the manager to be the absolute best. I shouldn't say that, but like, they need to be able to recognize skill.

• 2:17:02Lex Fridman?

  Yeah, and have good intuition, intuition that's laden with wisdom from all the battles of trying to reduce complexity in code bases.

• 2:17:10George Hotz?

  You know, I took a, I took a political approach at Comet, too, that I think is pretty interesting. I think Elon takes the same political approach. You know, Google had no politics. And what ended up happening is the absolute worst kind of politics took over.

• 2:17:23George Hotz?

  Comet has an extreme amount of politics and they're all mine and no dissidence is tolerated.

• 2:17:27Lex Fridman?

  So it's a dictatorship?

• 2:17:29George Hotz?

  Yep, it's an absolute dictatorship, right? Elon does the same thing. Now, the thing about my dictatorship is here are my values.

• 2:17:37Lex Fridman?

  Yeah, so it's transparent.

• 2:17:38George Hotz?

  It's transparent. It's a transparent dictatorship, right? And you can choose to opt in or, you know, you get free exit. Right? That's the beauty of companies. If you don't like the dictatorship, you quit.

• 2:17:47Lex Fridman?

  So you mentioned rewrite before, or refactor before features. If you were to refactor the Twitter codebase, what would that look like? And maybe also comment on how difficult is it to refactor?

• 2:18:01George Hotz?

  The main thing I would do is first of all, identify the pieces and then put tests in between the pieces. Right? So there's all these different, Twitter has a microservice architecture.

• 2:18:12George Hotz?

  There's all these different microservices. And the thing that I was working on there, look, like, you know, George didn't know any JavaScript. He asked how to fix search, blah, blah, blah, blah, blah. Look, man, like, the thing is, like, I just, you know, I'm upset that the way that this whole thing was portrayed, because it wasn't like, it wasn't like taken by people like honestly. It wasn't like by, it was taken by people who started out with a bad faith assumption.

• 2:18:35Lex Fridman?

  Yeah.

• 2:18:36George Hotz?

  And, I mean, I'd look, I can't like-

• 2:18:38Lex Fridman?

  And you as a programmer were just being transparent out there, actually having like fun. And like, this is what programming should be about.

• 2:18:45George Hotz?

  I love that Elon gave me this opportunity. Yeah. Like, really, it does. And like, you know, he came on my, my, the, the, the day I quit, he came on my Twitter Spaces afterward and we had a conversation. Like, I just, I respect that so much.

• 2:18:55Lex Fridman?

  Yeah, and it's also inspiring to just engineers and programmers and just, it's cool. It should be fun. The people that were hating on it, it's like, oh, man.

• 2:19:03George Hotz?

  It was fun. It was fun. It was stressful, but I felt like, you know, was that like a cool like point in history and like, I hope I was useful. I probably kind of wasn't, but like, maybe.

• 2:19:12Lex Fridman?

  Well, you also were one of the people that kind of made a strong case to refactor.

• 2:19:17George Hotz?

  Yeah.

• 2:19:17Lex Fridman?

  And that that's a really interesting thing to raise. Like, maybe that is the right, you know, the timing of that is really interesting. If you look at just the development of autopilot, you know, going from mobile eye to just like more, if you look at the history of semi autonomous driving in Tesla is is more and more like you could say refactoring or or starting from scratch, redeveloping from scratch.

• 2:19:43George Hotz?

  It's refactoring all the way down.

• 2:19:45Lex Fridman?

  And like, and the question is like, can you do that sooner? Can you maintain product profitability? And like, what's the, what's the right time to do it? How do you do it? You know, on any one day, it's like, you don't want to pull off the Band-Aids. Like, it's like, everything works. It's just like little fix here and there, but maybe starting from scratch.

• 2:20:06George Hotz?

  This is the main philosophy of TinyGrad. You have never refactored enough. Your code can get smaller. Your code can get simpler. Your ideas can be more elegant.

• 2:20:14Lex Fridman?

  But would you consider, you know, say you were like running Twitter development teams, engineering teams, would you go as far as like different programming language? Just go that far?

• 2:20:28George Hotz?

  I mean, the first thing that I would do is build tests. The first thing I would do is get a CI to where people can trust to make changes.

• 2:20:40Lex Fridman?

  So that if you keep.

• 2:20:40George Hotz?

  Before I touched any code, I would actually say, no one touches any code. The first thing we do is we test this code base. This is classic. This is how you approach a legacy code base. This is like what any, how to approach a legacy code base book will tell you.

• 2:20:52Lex Fridman?

  So, and then you hope that there's modules that can live on for a while. And then you add new ones, maybe in a different language or.

• 2:21:02George Hotz?

  Before we did new ones, we replace old ones.

• 2:21:05Lex Fridman?

  Yeah, yeah. Meaning like replace old ones with something simpler.

• 2:21:07George Hotz?

  We we look at this, like this thing that's 100,000 lines and we're like, well, okay, maybe this did even make sense in 2010, but now we can replace this with an open source thing. Right?

• 2:21:18Lex Fridman?

  Yeah.

• 2:21:18George Hotz?

  And you know, we look at this here, here's another 50,000 lines. Well, actually, you know, we can replace this with 300 lines of go. And you know what? I trust that the go actually replaces this thing because all the tests still pass. So step one is testing.

• 2:21:30Lex Fridman?

  Yeah.

• 2:21:31George Hotz?

  And then step two is like, the programming language is an afterthought. Right? You know, let a whole lot of people compete, be like, okay, who wants to rewrite a module? Whatever language you want to write it in, just the tests have to pass. And if you figure out how to make the test pass, but break the site, that's, we gotta go back to step one. Step one is get tests that you trust in order to make changes in the code base.

• 2:21:49Lex Fridman?

  I wonder how hard it is to, because I'm with you on testing and everything. You have from tests to like asserts to everything, code is just covered in this because

• 2:22:01Lex Fridman?

  it should be very easy to make rapid changes. And no, that's not going to break everything. And that's the way to do it. But I wonder how difficult is it to integrate tests into a code base that doesn't have many of them?

• 2:22:15George Hotz?

  So I'll tell you what my plan was at Twitter. It's actually similar to something we use at Comma. So at Comma, we have this thing called process replay. And we have a bunch of routes that'll be run through. So Comma is a microservice architecture too. We have microservices in the driving. Like we have one for the cameras, one for the sensor, one for the planner, one for the model. And we have an API, which the microservices talk to each other with. We use this custom thing called Serial, which uses ZMQ. Twitter uses Thrift. And then it uses this thing called Finagle, which is a Scala RPC backend. But this doesn't even really matter. The Thrift and Finagle layer was a great place, I thought, to write tests. Right? To start building something that looks like process replay.

• 2:23:00George Hotz?

  So Twitter had some stuff that looked kind of like this, but it wasn't offline. It was only online. So you could ship like a modified version of it. And then you could redirect some of the traffic to your modified version and diff those two. But it was all online. Like there was no like CI in the traditional sense. I mean, there was some, but like it was not full coverage.

• 2:23:19Lex Fridman?

  So you can't run all of Twitter offline to test something?

• 2:23:23George Hotz?

  Then this was another problem. You can't run all of Twitter. Right?

• 2:23:26Lex Fridman?

  Period. Any one person can't.

• 2:23:28George Hotz?

  Twitter runs in three data centers and that's it.

• 2:23:31Lex Fridman?

  Yeah.

• 2:23:31George Hotz?

  There's no other place you can run Twitter, which is like, George, you don't understand, this is modern software development. No, this is bullshit. Like, why can't it run on my laptop? What do you do? Twitter can run it? Yeah, okay. Well, I'm not saying you're gonna download the whole database to your laptop, but I'm saying all the middleware and the front end should run on my laptop, right?

• 2:23:50Lex Fridman?

  That sounds really compelling.

• 2:23:51George Hotz?

  Yeah.

• 2:23:53Lex Fridman?

  But can that be achieved by a codebase that grows over the years? I mean, the three data centers didn't have to be, right? Because they're totally different, like, designs.

• 2:24:03George Hotz?

  The problem is more like, like, why did the codebase have to grow? What new functionality has been added to compensate for the lines of code that are there?

• 2:24:13Lex Fridman?

  One of the ways to explain is that the incentive for software developers to move up in the company is to add code, to add, especially large.

• 2:24:21George Hotz?

  And you know what? The incentive for politicians to move up in the political structure is to add laws.

• 2:24:25Lex Fridman?

  Yeah.

• 2:24:25George Hotz?

  Same problem.

• 2:24:26Lex Fridman?

  Yeah. Yeah. If the flip side is to simplify, simplify, simplify.

• 2:24:32George Hotz?

  I mean, you know what? This is something that I do differently from Elon with comma about self-driving cars.

• 2:24:40George Hotz?

  You know, I hear the new version's gonna come out and the new version is not gonna be better, but at first, and it's gonna require a ton of refactors. I say, okay, take as long as you need. Like, you convince me this architecture is better? Okay, we have to move to it. Even if it's not gonna make the product better tomorrow, the top priority is making, is getting the architecture right.

• 2:25:00Lex Fridman?

  So what do you think about sort of a thing where the product is online?

• 2:25:06Lex Fridman?

  So how, I guess, would you do a refactor? If you ran engineering on Twitter, would you just do a refactor? How long would it take? What would that mean for the running of the, of the actual service?

• 2:25:17George Hotz?

  You know, and

• 2:25:21George Hotz?

  I'm not the right person to run Twitter. I'm just not. And that's the problem. Like, like, I don't really know. I don't really know if that's, you know, a common thing that I thought a lot while I was there was whenever I thought something that was different to what Elon thought, I'd have to run something in the back of my head reminding myself that Elon is the richest man in the world. And in general, his ideas are better than mine. Now, there's a few things I think I do understand and know more about. But like, in general, I'm not qualified to run Twitter. No, I was gonna say qualified, but like, I don't think I'd be that good at it. I don't think I'd be good at it. I don't think I'd really be good at running an engineering organization at scale.

• 2:26:03George Hotz?

  I think I could lead a very good refactor of Twitter. And it would take like six months to a year. And the results to show at the end of it would be feature development in general takes 10x less time, 10x less man hours. That's what I think I could actually do. Do I think that it's the right decision for the business above my pay grade?

• 2:26:28Lex Fridman?

  Yeah, but a lot of these kinds of decisions are above everybody's pay grade.

• 2:26:32George Hotz?

  I don't want to be a manager. I don't want to do that. I just like, like, if you really forced me to, yeah, it would make me maybe

• 2:26:40George Hotz?

  make me upset if I had to make those decisions. I don't want to.

• 2:26:45Lex Fridman?

  Yeah, but a refactor is so compelling. If this is to become something much bigger than what Twitter was,

• 2:26:54Lex Fridman?

  is it feels like a refactor has to be coming at some point.

• 2:26:58George Hotz?

  George, you're a junior software engineer. Every junior software engineer wants to come in and refactor the whole code. Okay, like, that's like your opinion, man.

• 2:27:07Lex Fridman?

  Yeah, it doesn't, you know, sometimes they're right.

• 2:27:11George Hotz?

  Well, like, whether they're right or not, it's definitely not for that reason. Right? It's definitely not a question of engineering prowess. It is a question of maybe what the priorities are for the company. And I did get more intelligent, like, feedback from people, I think, in good faith, like, saying that. From, actually, from Elon. And like, you know, from, from, from Elon, sort of like, like, people were like, well, you know, a stop the world refactor might be great for engineering, but you know, we have business to run. And hey, above my pay grade.

• 2:27:38Lex Fridman?

  What'd you think about Elon as an engineering leader? Having to experience him in the most chaotic of spaces, I would say.

• 2:27:51George Hotz?

  My respect for him is unchanged. And I did have to think a lot more deeply about some of the decisions he's forced to make.

• 2:27:59Lex Fridman?

  About the tensions within those, the trade offs within those decisions?

• 2:28:04George Hotz?

  About like a whole, like... Like Matrix coming at him. I think that's Andrew Tate's word for it. Sorry to borrow it.

• 2:28:12Lex Fridman?

  Also, bigger than engineering, just everything.

• 2:28:15George Hotz?

  Yeah, like, like the war on the woke.

• 2:28:19Lex Fridman?

  Yeah.

• 2:28:20George Hotz?

  Like, it just, it's just, man, and like, he doesn't have to do this, you know? He doesn't have to. He could go like Parag and go chill at the Four Seasons Maui. You know, but see, one person I respect and one person I don't.

• 2:28:36Lex Fridman?

  So his heart is in the right place, fighting in this case for this ideal of the freedom of expression.

• 2:28:43George Hotz?

  I wouldn't define the ideal so simply. I think you can define the ideal no more than just saying Elon's idea of a good world. Freedom of expression is.

• 2:28:54Lex Fridman?

  But to you, it's still, the downsides of that is the monarchy.

• 2:28:59George Hotz?

  Yeah, I mean, monarchy has problems, right? But I mean, would I trade right now the monarch, the current oligarchy, which runs America for the monarchy? Yeah, I would. Sure. For the Elon monarchy? Yeah, you know why? Because power would cost one cent a kilowatt hour. Tenth of a cent a kilowatt hour.

• 2:29:18Lex Fridman?

  What do you mean?

• 2:29:20George Hotz?

  Right now, I pay about 20 cents a kilowatt hour for electricity in San Diego. That's like the same price you paid in 1980. What the hell?

• 2:29:28Lex Fridman?

  So you would see a lot of innovation with Elon?

• 2:29:31George Hotz?

  Maybe it'd have, maybe have some hyperloops?

• 2:29:33Lex Fridman?

  Yeah.

• 2:29:33George Hotz?

  Right? And I'm willing to make that trade off, right? I'm willing to make, and this is why, you know, people think that like dictators take power through some like, through some untoward mechanism. Sometimes they do, but usually it's because the people want them. And the downsides of a dictatorship, I feel like we've gotten to a point now with the oligarchy where, yeah, I would prefer the dictator.

• 2:29:56Lex Fridman?

  Uh, what'd you think about Scala as a programming language?

• 2:30:01George Hotz?

  I liked it more than I thought. I did the tutorials. Like, I was very new to it. Like, it would take me six months to be able to write, like, good Scala.

• 2:30:07Lex Fridman?

  I mean, what did you learn about learning a new programming language from that?

• 2:30:10George Hotz?

  Um, oh, I love, I love doing, like, new programming tutorials and doing them. I did all this for Rust. Uh, i- i- it keeps, some of it's upsetting JVM roots, but it is a much nicer. In fact, I almost don't know why Kotlin took off and not Scala. I think Scala has some beauty that Kotlin lacked.

• 2:30:31George Hotz?

  Whereas Kotlin felt a lot more, I mean, it was almost like, I don't know if it actually was a response to Swift, but that's kind of what it felt like. Like Kotlin looks more like Swift and Scala looks more like, well, like a functional programming language, more like an OCaml or Haskell.

• 2:30:44Lex Fridman?

  Let's actually just explore, we touched on it a little bit, but just on the art, the science and the art of programming. Oh, for you personally, how much of your programming is done with GPT currently?

• 2:30:54George Hotz?

  None.

• 2:30:55Lex Fridman?

  None?

• 2:30:56George Hotz?

  I don't use it at all.

• 2:30:57Lex Fridman?

  Because you prioritize simplicity so much?

• 2:31:00George Hotz?

  Yeah, I find that a lot of it is noise. I do use VS Code.

• 2:31:06George Hotz?

  And I do like some amount of autocomplete. I do like, like, a very, um, a very, like, feels like rules-based autocomplete. Like, or an autocomplete that's going to complete the variable name for me. So I don't have to type it, I can just press Tab. Alright, that's nice. But I don't want an autocomplete. You know what I hate? When autocomplete, when I type the word "for" and it, like, puts, like, two, two parentheses and two semicolons and two braces, I'm like, oh, man.

• 2:31:28Lex Fridman?

  Well, let me, with, uh, VS Code and GPT with Codex, you can, um, you can kind of brainstorm. I, I find, I'm like, probably the same as you, but I like that it generates code and you basically disagree with it and write something simpler. But to me, that somehow is, like, inspiring. It makes me feel good. It also gamifies the simplification process. Because I'm like, oh, yeah, you dumb AI system, you think this is the way to do it. I have a simpler thing here.

• 2:31:59George Hotz?

  It just constantly reminds me of, like, like, bad stuff. I mean, I tried the same thing with rap, right? I tried the same thing with rap, and actually, I think I'm a much better programmer than rapper. But like, I even tried, I was like, okay, can we get some inspiration from these things for some rap lyrics? And I just found that it would go back to the most, like, cringey tropes and dumb rhyme schemes. And I'm like, yeah, this is what the code looks like, too.

• 2:32:20Lex Fridman?

  I, I think you and I probably have different thresholds for cringe code. You probably hate cringe code. So it's for you. I mean, Boilerplate is a part of code.

• 2:32:35Lex Fridman?

  Like, some of it.

• 2:32:38Lex Fridman?

  Yeah, and some of it is just like faster lookup. Because I don't know about you, but I don't remember everything. Like, I don't, I'm offloading so much of my memory about like,

• 2:32:50Lex Fridman?

  yeah, different functions, library functions, all that kind of stuff. Like, this GPT just is very fast at standard stuff. And like, standard library stuff, basic stuff that everybody uses.

• 2:33:03George Hotz?

  Yeah, I think that.

• 2:33:08George Hotz?

  I don't know. I mean, there's just so little of this in Python. Maybe if I was coding more in other languages, I would consider it more. But I feel like Python already does such a good job of removing any Boilerplate.

• 2:33:20Lex Fridman?

  That's true.

• 2:33:21George Hotz?

  It's the closest thing you can get to pseudocode, right?

• 2:33:23Lex Fridman?

  Yeah, that's true. That's true.

• 2:33:26George Hotz?

  And like, yeah, sure. If I like, yeah, I'm great, GPT. Thanks for reminding me to free my variables. Unfortunately, you didn't really recognize the scope correctly, and you can't free that one. But like, you put the freeze there, and like, I get it.

• 2:33:39Lex Fridman?

  Fiverr. Whenever I've used Fiverr for certain things like design or whatever.

• 2:33:45George Hotz?

  Yeah.

• 2:33:45Lex Fridman?

  It's always you come back. I think that's probably closer. My experience with Fiverr is closer to your experience with programming with GPT is like, you're just frustrated and feel worse about the whole process of design and art and whatever I use Fiverr for.

• 2:34:01Lex Fridman?

  Still, I just feel like later versions of GPT, I'm using

• 2:34:07Lex Fridman?

  GPT as much as possible to just learn the dynamics of it, like, these early versions, because it feels like in the future you'll be using it more and more. And so like, I don't want to be, like, for the same reason, I gave away all my books and switched to Kindle. Because like, all right, how long are we gonna have paper books? Like, 30 years from now? Like, I want to learn to be reading on Kindle, even though I don't enjoy it as much. And you learn to enjoy it more. In the same way, I switched from, let me just pause, I switched from Emacs to VS Code.

• 2:34:40George Hotz?

  Yeah. I switched from Vim to VS Code. I think I, similar, but.

• 2:34:43Lex Fridman?

  Yeah, it's tough. And Vim to VS Code is even tougher, because Emacs is like. Old, like more outdated, feels like it. The community is more outdated. Vim is like pretty vibrant still, so.

• 2:34:56George Hotz?

  I never used any of the plugins. I still don't use any.

• 2:34:58Lex Fridman?

  That's what I looked at myself in the mirror. I'm like, yeah, you wrote some stuff in Lisp. Yeah.

• 2:35:02George Hotz?

  But I never used any of the plugins in Vim either. I had the most vanilla Vim. I have a syntax highlighter. I didn't even have autocomplete. Like, these things, I feel like, help you so marginally that, like, and now, okay, now VS Code's autocomplete has gotten good enough that like, okay, I don't have to set it up. I can just go into codebase and autocomplete's right 90% of the time. Okay, cool, I'll take it. All right. So I don't think I'm gonna have a problem at all adapting to the tools once they're good. But like, the real thing that I want is not something that like, tab completes my code and gives me ideas. The real thing that I want is a very intelligent pair programmer that comes up with a little pop up saying,

• 2:35:45George Hotz?

  hey, you wrote a bug on line 14 and here's what it is.

• 2:35:49Lex Fridman?

  Yeah.

• 2:35:49George Hotz?

  Now, I like that. You know what does a good job of this? Mypi. I love Mypi. Mypi, this fancy type checker for Python.

• 2:35:56Lex Fridman?

  Yeah.

• 2:35:56George Hotz?

  And actually, I tried, like, Microsoft released one too, and it was like 60% false positives. Mypi is like 5% false positives. 95% of the time, it recognizes, I didn't really think about that typing interaction correctly. Thank you, Mypi.

• 2:36:10Lex Fridman?

  So you like type hinting? You liked, you like pushing the language towards, towards being a typed language?

• 2:36:16George Hotz?

  Oh, yeah, absolutely. I think, I think optional typing is, is, is great. I mean, look, I think that like, it's like a meat in the middle, right? Like, Python has this optional type hinting and like, C++ has auto.

• 2:36:27Lex Fridman?

  C++ takes, allows you to take a step back.

• 2:36:29George Hotz?

  Well, C++ would have you brutally type out std string iterator, right? Now I can just type auto, which is nice. And then Python used to just have A. What type is A? It's an A.

• 2:36:41Lex Fridman?

  Yeah.

• 2:36:42George Hotz?

  A colon STR. Oh, okay, it's a string. Cool.

• 2:36:46Lex Fridman?

  Yeah.

• 2:36:46George Hotz?

  I wish there were, I wish there was a way, like a simple way in Python to, like, turn on a mode which would enforce the types.

• 2:36:54Lex Fridman?

  Yeah, like give a warning when there's no type or something like this.

• 2:36:56George Hotz?

  Well, no, to give a warning where like my pie is a static type checker, but I'm asking just for a runtime type checker. Like there's like ways to like hack this in, but I wish it was just like a flag like Python three dash T.

• 2:37:06Lex Fridman?

  Oh, I see. I see.

• 2:37:07George Hotz?

  Enforce the types around time.

• 2:37:08Lex Fridman?

  Yeah, I feel like that makes you a better programmer that that's the kind of test, right? That the the type can the type remains the same.

• 2:37:15George Hotz?

  Well, that I know that I didn't like mess any types up. But again, like my pie is getting really good and I love it. Um, and I can't wait for some of these tools to become AI powered. I like, I want AIs reading my code and giving me feedback. I don't want AIs writing half-assed autocomplete stuff for me.

• 2:37:32Lex Fridman?

  I wonder if you can now take GPT and give it a code that you wrote for a function and say, how can I make this simpler and have it accomplish the same thing? I think you'll get some good ideas on some code, maybe not the code you write.

• 2:37:46Lex Fridman?

  For tiny grad type of code, because that requires so much design thinking, but like other kinds of code.

• 2:37:51George Hotz?

  I don't know. I downloaded the plugin, maybe like two months ago, I tried it again and found the same. Look, I don't doubt that these models are going to first become useful to me, then be as good as me and then surpass me. But from what I've seen today, it's like like, like someone, you know, occasionally taking over my keyboard that I hired from Fiverr.

• 2:38:18Lex Fridman?

  Yeah, but ideas about how to debug the coder, basically a better debugger is really interesting. I mean, I

• 2:38:24George Hotz?

  But it's not a better debugger. I guess I would love a better debugger.

• 2:38:27Lex Fridman?

  Yeah, it's not yet. Yeah, but it feels like it's not too far.

• 2:38:30George Hotz?

  Yeah, one of my coworkers says he uses them for print statements. Like every time he has to like, just like when he needs, the only thing it can really write is like, okay, I just want to write the thing to like, print the state out right now.

• 2:38:40Lex Fridman?

  Oh, that definitely is much faster is print statements. Yeah.

• 2:38:44George Hotz?

  Yeah.

• 2:38:45Lex Fridman?

  I see myself using that a lot, just like, because it it figures out the rest of the functions, just like, okay, print everything.

• 2:38:50George Hotz?

  Yeah, print everything, right? And then, yeah, like if you want a pretty printer, maybe. I'm like, yeah, you know what? I think like, I think in two years, I'm going to start using these plugins.

• 2:38:58Lex Fridman?

  Yeah.

• 2:38:58George Hotz?

  A little bit. And then in five years, I'm going to be heavily relying on some AI augmented flow. And then in 10 years.

• 2:39:05Lex Fridman?

  Do you think you'll ever get to 100%? Where the like, what's the role of the human? That it converges to as a programmer?

• 2:39:15Lex Fridman?

  So you think it's all generated?

• 2:39:17George Hotz?

  Our niche becomes, I think it's over for humans in general. It's not just programming, it's everything.

• 2:39:23Lex Fridman?

  So niche become, wow.

• 2:39:25George Hotz?

  Our niche becomes smaller, smaller, smaller. In fact, I'll tell you what the last niche of humanity is going to be.

• 2:39:28Lex Fridman?

  Yeah.

• 2:39:29George Hotz?

  Um, there's a great book and it's, if I recommended Metamorphosis Prime Intellect last time, there is a sequel called A Casino Odyssey in Cyberspace. And um, I don't want to give away the ending of this, but it tells you what the last remaining human currency is. And I agree with that.

• 2:39:47Lex Fridman?

  We'll uh, leave that as a cliffhanger. Uh, so no more programmers left, huh? That's where we're going.

• 2:39:55George Hotz?

  Well, unless you want handmade code, maybe they'll sell it on Etsy. This is handwritten code. Doesn't have that machine polished to it. It has those slight imperfections that would only be written by a person.

• 2:40:07Lex Fridman?

  I wonder how far away we are from that. I mean, there's uh, some aspect to, you know, on Instagram, your title is listed as prompt engineer.

• 2:40:15George Hotz?

  Right? Thank you for noticing.

• 2:40:19Lex Fridman?

  It's uh, I don't know if it's ironic or uh, non uh, or sarcastic or non. Uh, what do you think of prompt engineering as a scientific and engineering discipline? Or maybe, and maybe art form?

• 2:40:34George Hotz?

  You know what? I started comma six years ago and I started the tiny corp a month ago. So much has changed. Like, I'm now thinking, I'm now like, I started like going through like similar comma processes to like starting a company. I'm like, okay, I'm going to get an office in San Diego. I'm going to bring people here. I don't think so. I think I'm actually going to do remote, right? George, you're going to do remote? You hate remote. Yeah, but I'm not going to do job interviews. The only way you're going to get a job is if you contribute to the GitHub, right? And then like, it like, like interacting through GitHub, like like GitHub being the real like project management software for your company. And the thing pretty much just is a GitHub repo.

• 2:41:17George Hotz?

  Is like showing me kind of what the future of, okay. So a lot of times I'll go on a Discord, our Kanyagra Discord, and I'll throw out some random like, hey, you know, can you change, instead of having log and exp as LL Ops, change it to log two and exp two?

• 2:41:32George Hotz?

  It's a pretty small change. You could just use like change a base formula.

• 2:41:36George Hotz?

  That's the kind of task that I can see an AI being able to do in a few years. Like in a few years, I could see myself describing that. And then within 30 seconds, a pull request is up that does it. And it passes my CI and I merge it. Right? So I really started thinking about like, well, what is the future of like, like jobs? How many AIs can I employ at my company? As soon as we get the first tiny box up, I'm going to stand up a 65B llama in the Discord. And it's like, yeah, here's the tiny box. He's just like, he's chilling with us.

• 2:42:04Lex Fridman?

  Basically, I mean, like you said with niches, like most human jobs will eventually be replaced with prompt engineering.

• 2:42:14George Hotz?

  Well, prompt engineering kind of is this like,

• 2:42:18George Hotz?

  as you like move up the stack, right? Like, okay, there used to be humans actually doing arithmetic by hand. There used to be like big forms of people doing, doing, doing pluses and stuff, right? And then you have like spreadsheets, right? And then, okay, the spreadsheet can do the plus for me. And then you have like macros, right? And then you have like things that basically just are spreadsheets under the hood, right? Like, like accounting software.

• 2:42:43George Hotz?

  As we move further up the abstraction, well, what's at the top of the abstraction stack? Well, prompt engineer.

• 2:42:48Lex Fridman?

  Yeah.

• 2:42:49George Hotz?

  Right? What is, what is the last thing? If you think about like humans wanting to keep control? Well, what am I really in the company but a prompt engineer? Right?

• 2:42:59Lex Fridman?

  Isn't there a certain point where the AI will be better at writing prompts?

• 2:43:04George Hotz?

  Yeah, but you see, the problem with the AI writing prompts, a definition that I always liked of AI was AI is the do what I mean machine. AI is not the like, the computer is so pedantic. It does what you say. So but you want the do what I mean machine.

• 2:43:21Lex Fridman?

  Yeah.

• 2:43:21George Hotz?

  Right. You want the machine where you say, you know, get my grandmother out of the burning house. It like reasonably takes your grandmother and puts her on the ground, not lifts her 1,000 feet above the burning house and lets her fall.

• 2:43:31Lex Fridman?

  But you don't.

• 2:43:32George Hotz?

  There's an old Yudkowsky example.

• 2:43:34Lex Fridman?

  But it's not going to find the meaning. I mean, to do what I mean, it has to figure stuff out.

• 2:43:42George Hotz?

  Sure.

• 2:43:43Lex Fridman?

  And the thing you'll maybe ask it to do is run government for me.

• 2:43:49George Hotz?

  Oh, and do what I mean very much comes down to how aligned is that AI with you? Of course, when you talk to an AI that's made by a big company in the cloud, the AI fundamentally is aligned to them, not to you.

• 2:44:03Lex Fridman?

  Yeah.

• 2:44:04George Hotz?

  And that's why you have to buy a tiny box. So you make sure the AI stays aligned to you. Every time that they start to pass, you know, AI regulation or GPU regulation, I'm gonna see sales of tiny boxes spike. It's gonna be like guns. Every time they talk about gun regulation, boom, gun sales.

• 2:44:18Lex Fridman?

  So in the space of AI, you're an anarchist, anarchism, espouser believer.

• 2:44:24George Hotz?

  I'm an informational anarchist. Yes, I'm an informational anarchist and a physical statist.

• 2:44:31George Hotz?

  I do not think anarchy in the physical world is very good because I exist in the physical world. But I think we can construct this virtual world where anarchy, it can't hurt you, right? I love that Tyler, the creator tweet, your cyber bullying isn't real, man. Have you tried turning off the screen? Close your eyes like.

• 2:44:47Lex Fridman?

  Yeah.

• 2:44:51Lex Fridman?

  But how do you prevent the AI from basically replacing all human prompt engineers? Where there's it's like a self, like where nobody's the prompt engineer anymore. So autonomy, greater and greater autonomy until it's full autonomy.

• 2:45:07George Hotz?

  Yeah.

• 2:45:08Lex Fridman?

  And that's just where it's headed. Because one person is gonna say, run everything for me.

• 2:45:14George Hotz?

  You see.

• 2:45:17George Hotz?

  I look at potential futures and as long as the AIs go on to create a vibrant civilization with diversity and complexity across the universe, more power to them, I'll die. If the AIs go on to actually like turn the world into paperclips and then they die out themselves, well, that's horrific and we don't want that to happen. So this is what I mean about like robustness. I trust robust machines. The current AIs are so not robust, like this comes back to the idea that we've never made a machine that can self replicate. Right. But when we have, if the machines are truly robust and there is one prompt engineer left in the world. Hope you're doing good, man.

• 2:46:00George Hotz?

  Hope you believe in God, like, you know, you know, go by God and go forth and conquer the universe.

• 2:46:08Lex Fridman?

  Well, you mentioned, because I talked to Mark about faith in God and you said you were impressed by that. What's your own belief in God and how does that affect your work?

• 2:46:18George Hotz?

  You know, I never really considered when I was younger, I guess my parents were atheists. So I was raised kind of atheist. I never really considered how absolutely like silly atheism is because like I create worlds, right? Every like game creator, like how are you an atheist, bro? You create worlds. Who the, no one created our world, man. That's different. Haven't you heard about like the big bang and stuff? Yeah. I mean, what's the Skyrim myth origin story in Skyrim? I'm sure there's like some part of it in Skyrim, but it's not like if you ask the creators, like the big bang is in universe, right? I'm sure they have some big bang notion in Skyrim, right? But that obviously is not at all how Skyrim was actually created. It was created by a bunch of programmers in a room, right? So like, you know, it just, it struck me one day how just silly atheism is, right?

• 2:47:03George Hotz?

  Like, of course we were created by God. It's the most obvious thing.

• 2:47:07Lex Fridman?

  Yeah, that's uh

• 2:47:11Lex Fridman?

  that's such a nice way to put it like we're we're such powerful creators ourselves. It uh it's silly not to conceive that there's creators even more powerful than us.

• 2:47:20George Hotz?

  Yeah. And then like, I also just like, I like that notion. That notion gives me a lot of. I mean, I guess you can talk about what it gives a lot of religious people. It's kind of like, it just gives me comfort. It's like, you know what? If we mess it all up and we die out.

• 2:47:34Lex Fridman?

  Yeah, in the same the same way that a video game kind of has comfort in it.

• 2:47:37George Hotz?

  God will try again.

• 2:47:39Lex Fridman?

  Or there's balance, like somebody figured out a balanced. View of it like how to like, so it's it all makes sense in the end. Like a a video game is usually not going to have crazy, crazy stuff.

• 2:47:52George Hotz?

  You know, people will come up with a like a. Well, yeah, but like, man, who created God? Like, that's God's problem. No, like, I'm not gonna think this is this is what you're asking me what if God?

• 2:48:05Lex Fridman?

  I'm just living. I'm just this NPC living in this game.

• 2:48:09George Hotz?

  I mean, to be fair, like if God didn't believe in God, he'd be as, you know, silly as the atheists here.

• 2:48:14Lex Fridman?

  Uh, what do you think is the greatest uh? Computer game of all time? Do you do you have any time to play games anymore? Have you played Diablo four?

• 2:48:23George Hotz?

  I have not played Diablo four.

• 2:48:24Lex Fridman?

  I will be doing that shortly. I have to.

• 2:48:27George Hotz?

  All right.

• 2:48:27Lex Fridman?

  There's just so much history with one, two, and three.

• 2:48:29George Hotz?

  You know what? I'm gonna say World of Warcraft.

• 2:48:34George Hotz?

  And. It's not that the game is so such a great game. It's not. It's that I remember in 2005, when it came out. How it opened my mind to. Ideas. It opened my mind to like. Like, like, like, this this whole world we've created, right? There's almost been nothing like it since. Like, you can look at MMOs today, and I think they all have lower user bases than World of Warcraft, like Eve Online is kind of cool. But. But to think that like, like, everyone know, you know, people are always like, to look at the Apple headset like, what do people want in this VR? Everyone knows what they want. I want Ready Player One.

• 2:49:15George Hotz?

  And like that. So I'm going to say World of Warcraft and I'm hoping that like games can get out of this whole mobile gaming dopamine pump thing and like

• 2:49:25Lex Fridman?

  Create worlds.

• 2:49:26George Hotz?

  Create worlds, yeah.

• 2:49:28Lex Fridman?

  And worlds that captivate a very large fraction of the human population.

• 2:49:32George Hotz?

  Yeah, and I think it'll come back. I believe.

• 2:49:35Lex Fridman?

  But MMO like really, really pull you in.

• 2:49:38George Hotz?

  Games do a good job. I mean, okay, other like two other games that I think are, you know, very noteworthy for me are Skyrim and GTA V.

• 2:49:45Lex Fridman?

  Skyrim, yeah, that's probably number one for me. GTA. Yeah, what is it about GTA? GTA is really, I guess GTA is real life. I know there's prostitutes and guns and stuff.

• 2:50:00George Hotz?

  They exist in real life too.

• 2:50:03Lex Fridman?

  Yes, I know. But it's a, it's how imagine your life to be actually.

• 2:50:07George Hotz?

  I wish it was that cool.

• 2:50:08Lex Fridman?

  Yeah.

• 2:50:11Lex Fridman?

  Yeah, I guess that's, you know, because there's Sims, right? Which is also a game I like. But it's a gamified version of life, but it also is, I would love the combination of Sims and GTA.

• 2:50:24Lex Fridman?

  So more freedom, more violence, more rawness, but with also like ability to have a career and family and this kind of stuff.

• 2:50:31George Hotz?

  What I'm really excited about in games is like, once we start getting intelligent AIs to interact with.

• 2:50:37Lex Fridman?

  Oh, yeah.

• 2:50:38George Hotz?

  Right? Like the NPCs in games have never been.

• 2:50:41Lex Fridman?

  But conversationally, in every way.

• 2:50:44George Hotz?

  In like, yeah, in like every way, like when you're actually building a world and a world imbued with intelligence.

• 2:50:52Lex Fridman?

  Oh, yeah.

• 2:50:53George Hotz?

  Right? And it's just hard. Like, there's just like, like, you know, running World of Warcraft, like, you're limited by what you're running on a Pentium four, you know? How much intelligence can you run? How many flops did you have? Right? But now when I'm running a game on a hundred Petaflop machine, well, that's five people. I'm trying to make this a thing. Twenty Petaflops of compute is one person of compute. I'm trying to make that a unit.

• 2:51:12Lex Fridman?

  Twenty Petaflops?

• 2:51:14George Hotz?

  Yeah.

• 2:51:15Lex Fridman?

  Is one person.

• 2:51:15George Hotz?

  One person.

• 2:51:17Lex Fridman?

  One person flop.

• 2:51:17George Hotz?

  It's like a horsepower.

• 2:51:20George Hotz?

  What's a horsepower? That's how powerful a horse is. What's a, what's a person of compute?

• 2:51:23Lex Fridman?

  You know, flop. I got it. That's interesting. Uh, VR also has, uh, in terms of creating worlds.

• 2:51:31George Hotz?

  You know what? Bought a Quest 2. I put it on and I can't believe the first thing they show me is a bunch of scrolling clouds and a Facebook login screen.

• 2:51:41Lex Fridman?

  Yeah.

• 2:51:42George Hotz?

  You had the ability to bring me into a world.

• 2:51:45Lex Fridman?

  Yeah.

• 2:51:46George Hotz?

  And what did you give me? A pop-up, right? Like.

• 2:51:49Lex Fridman?

  Well, I.

• 2:51:49George Hotz?

  And this is why you're not cool, Mark Zuckerberg. But you could be cool. Just make sure on the Quest 3, you don't put me into clouds and a Facebook login screen. Bring me to a world.

• 2:51:58Lex Fridman?

  I just tried Quest 3. It was awesome. But hear that, guys? I agree with that. So.

• 2:52:06Lex Fridman?

  You know what? Cause I, uh, I mean, the beginning, um, what is it? Todd Howard said this about, uh, the design of the beginning of the games he creates is like, the beginning is so, so, so important. I recently played Zelda for the first time. Zelda Breath of the Wild, the previous one. And like, it's very quickly, you come out of this, like, uh, within like 10 seconds, you come out of like a cave type place and it's like, this world opens up. It's like, ah, and you, it like, it pulls you in. You forget whatever troubles I was having, whatever, like.

• 2:52:39George Hotz?

  I gotta play that from the beginning. I played it for like an hour at a friend's house.

• 2:52:42Lex Fridman?

  Ah, no, the beginning, they got it. They did it really well. The expansiveness of that space. The, the peacefulness of that place, they got this, the music, I mean, so much of that is creating that world and pulling you right in.

• 2:52:55George Hotz?

  I'm gonna, I'm gonna go buy a Switch. Like, I'm gonna go today and buy a Switch.

• 2:52:57Lex Fridman?

  You should. Well, the new one came. I haven't played that yet, but, uh, Diablo 4 or something. I mean, there's sentimentality also, but something about VR

• 2:53:08Lex Fridman?

  really is incredible. But the, the new Quest 3 is mixed reality. And I got a chance to try that. So it's, uh, augmented reality. And for video games, it's done really, really well.

• 2:53:19George Hotz?

  Is it pass-through or cameras?

• 2:53:20Lex Fridman?

  Cameras.

• 2:53:21George Hotz?

  It's cameras. Okay.

• 2:53:21Lex Fridman?

  Yeah.

• 2:53:22George Hotz?

  The Apple one, is that one pass-through or cameras?

• 2:53:24Lex Fridman?

  I don't know.

• 2:53:25George Hotz?

  Yeah.

• 2:53:25Lex Fridman?

  I don't know how real it is. I don't know anything, you know?

• 2:53:28George Hotz?

  It's coming out in January.

• 2:53:30Lex Fridman?

  Is it January or is it some point?

• 2:53:32George Hotz?

  Some point, maybe not January. Maybe that's my optimism, but Apple, I will buy it. I don't care if it's expensive and does nothing. I will buy it. I will support this future endeavor.

• 2:53:40Lex Fridman?

  You're the meme. Oh, yes, I support competition. It seemed like Quest was like the only people doing it. And this is great that they're like.

• 2:53:50George Hotz?

  You know what? And this is another place we'll give some more respect to Mario Zuckerberg. The two companies that have endured through technology are Apple and Microsoft. And what do they make? Computers and business services. All the memes, social ads, they all come and go. But you want to endure, build hardware.

• 2:54:10Lex Fridman?

  Yeah. And that, you know, it does, that does a really interesting job. I mean, I, maybe I'm a noob at this, but it's a $500 headset, Quest 3. And just having creatures run around the space, like our space right here. To me, okay, this is very like boomer statement, but it added windows to the place.

• 2:54:34George Hotz?

  I heard about the aquarium.

• 2:54:36Lex Fridman?

  Yeah, aquarium, but in this case it was a zombie game, whatever. It doesn't matter. But just like it modifies the space in a way where I can't. It really feels like a window and you can look out. It's pretty cool. Like I was just, it's like a zombie game. They're running at me, whatever. But what I was enjoying is the fact that there's like a window and, and they're stepping on objects in the space.

• 2:54:59Lex Fridman?

  That was a different kind of escape. Also because you can see the other humans. So it's integrated with the other humans. It's really.

• 2:55:06George Hotz?

  And that's why it's more important than ever that the AI is running on those systems are aligned with you.

• 2:55:11Lex Fridman?

  Oh, yeah.

• 2:55:12George Hotz?

  They're going to augment your entire world.

• 2:55:14Lex Fridman?

  Oh, yeah. And that, those AIs have a, I mean, you think about all the dark stuff, like.

• 2:55:22Lex Fridman?

  Like sexual stuff, like if those AIs threaten me, that could be haunting.

• 2:55:28George Hotz?

  Hmm.

• 2:55:28Lex Fridman?

  That like if they like threaten me in a non video game way.

• 2:55:33George Hotz?

  Oh, yeah, yeah, yeah, yeah.

• 2:55:34Lex Fridman?

  Like they have no personal information about me. And it's like, and then you lose track of what's real, what's not, like what if stuff is like hacked?

• 2:55:41George Hotz?

  There's two directions the AI girlfriend company can take. Right. There's like the highbrow, something like her, maybe something you kind of talk to. And this is, and then there's the lowbrow version of it where I want to set up a brothel in Times Square.

• 2:55:51Lex Fridman?

  Yeah.

• 2:55:53George Hotz?

  Yeah, it's not cheating if it's a robot. It's a VR experience.

• 2:55:56Lex Fridman?

  If, is there an in between?

• 2:55:58George Hotz?

  No, I want to do that one or that one.

• 2:56:01Lex Fridman?

  Have you decided yet?

• 2:56:01George Hotz?

  No, I'll figure it out. We'll see. We'll see where the technology goes.

• 2:56:05Lex Fridman?

  I would love to hear your opinions for George's third company, what to do? The brothel in Times Square or the, the her experience?

• 2:56:17Lex Fridman?

  What do you think company number four will be? You think there'll be a company number?

• 2:56:20George Hotz?

  There's a lot to do in company number two. And just like I'm talking about company number three now, none of that tech exists yet. There's a lot to do in company number two. Company number two is going to be the great struggle of the next six years. And of the next six years, how centralized is compute going to be? The less centralized compute is going to be, the better of a chance we all have.

• 2:56:38Lex Fridman?

  So you're bearing that you're like a flag bearer for open source distributed decentralization of compute.

• 2:56:45George Hotz?

  We have to, we have to, or they will just completely dominate us. I showed a picture on stream of a man in a chicken farm. Ever seen one of those like factory farm chicken farms? Why does he dominate all the chickens?

• 2:56:58George Hotz?

  Why does he?

• 2:56:58Lex Fridman?

  Smarter.

• 2:56:59George Hotz?

  He's smarter, right? Some people, some people on Twitch were like, he's bigger than the chickens. Yeah. And now here's a man in a cow farm, right? So it has nothing to do with their size and everything to do with their intelligence. And if one central organization has all the intelligence, you'll be the chickens and they'll be the chicken man. But if we all have the intelligence, we're all the chickens.

• 2:57:23George Hotz?

  Well, not all the man were all the chickens. And there's no chicken man.

• 2:57:27Lex Fridman?

  There's no chicken man. We're just chickens in Miami.

• 2:57:31George Hotz?

  He was having a good life, man.

• 2:57:32Lex Fridman?

  I'm sure he was. I'm sure he was. What have you learned from launching and running Comma AI and Tiny Corp? So this, starting a company from an idea and scaling it. And by the way, I'm all in on Tiny Box. So I'm, I'm, I'm your, I'll, I guess it's preorder only now.

• 2:57:50George Hotz?

  I want to make sure it's good. I want to make sure that like, the thing that I deliver is like, not going to be like a Quest 2, which you buy and use twice. I mean, it's better than a Quest, which you bought and used less than once statistically.

• 2:58:02Lex Fridman?

  Well, if there's a beta program for Tiny Box, I'm into.

• 2:58:05George Hotz?

  Sounds good.

• 2:58:07Lex Fridman?

  So I won't be the whiny, uh, I'll be the tech savvy user of the Tiny Box, just to be in.

• 2:58:14George Hotz?

  What have I learned?

• 2:58:16Lex Fridman?

  What have you learned from building these companies?

• 2:58:20George Hotz?

  For the longest time at Comma, I asked, why? Why, you know, why did I start a company? Why did I do this?

• 2:58:30George Hotz?

  But, you know, what else was I going to do?

• 2:58:34Lex Fridman?

  So you like,

• 2:58:37Lex Fridman?

  you like bringing ideas to life?

• 2:58:41George Hotz?

  With Comma,

• 2:58:43George Hotz?

  it really started as an ego battle with Elon. I wanted to beat him. I like, I saw a worthy adversary, you know? Here's a worthy adversary who I can beat at self-driving cars. And like, I think we've kept pace and I think he's kept ahead. I think that's what's ended up happening there. Um, but I do think Comma is, I mean, Comma's profitable. Like, and like, when this drive GPT stuff starts working, that's it. There's no more like bugs in the loss function. Like right now, we're using like a hand-coded simulator. There's no more bugs. This is going to be it. Like this is the run-up to driving.

• 2:59:14Lex Fridman?

  I hear a lot of really, uh, a lot of props for Open Pilot for a comma.

• 2:59:19George Hotz?

  It's, it's so, it's, it's better than FSD and autopilot in certain ways. It has a lot more to do with which feel you like. We lowered the price on the hardware to $14.99. You know how hard it is to ship reliable consumer electronics that go on your windshield? We're doing more than like, most cellphone companies.

• 2:59:37Lex Fridman?

  How'd you pull that off, by the way? Shipping a product that goes in a car?

• 2:59:40George Hotz?

  I know. I have a, I have a, I have an SMT line. It's all, I make all the boards in-house in San Diego.

• 2:59:46Lex Fridman?

  Quality control.

• 2:59:47George Hotz?

  I care immensely about it. Actually, I'm.

• 2:59:49Lex Fridman?

  You're basically a mom and pop. Shop with great testing.

• 2:59:55George Hotz?

  Our head of open pilot is great at like, you know, okay, I want all the common theories to be identical.

• 3:00:01Lex Fridman?

  Yeah.

• 3:00:02George Hotz?

  And yeah, I mean, you know, it's look, it's 1499. It 30 day money back guarantee it will it will blow your mind at what it can do.

• 3:00:11Lex Fridman?

  Is it hard to scale?

• 3:00:13George Hotz?

  You know what? There's kind of downsides to scaling it. People are always like, why don't you advertise? Our mission is to solve self-driving cars while delivering shipable intermediaries. Our mission has nothing to do with selling a million boxes. It's tawdry.

• 3:00:26Lex Fridman?

  Do you think it's possible that Kama gets sold?

• 3:00:31George Hotz?

  Only if I felt someone could accelerate that mission and wanted to keep it open source. And like, not just wanted to, I don't believe what anyone says, I believe incentives. If a company wanted to buy Kama where their incentives were to keep it open source, but Kama doesn't stop at the cars. The cars are just the beginning. The device is a human head. The device has two eyes, two ears, it breathes air, has a mouth.

• 3:00:56Lex Fridman?

  So you think this goes to embodied robotics?

• 3:00:58George Hotz?

  We have, we sell common bodies too. You know, they're very, they're very rudimentary. But one of the problems that we're running into is that the Kama 3 has about as much intelligence as a bee. If you want a human's worth of intelligence, you're going to need a tiny rack, not even a tiny box. You're going to need like a tiny rack, maybe even more.

• 3:01:21Lex Fridman?

  How does that, how do you put legs on that?

• 3:01:23George Hotz?

  You don't, and there's no way you can. You, you connect to it wirelessly. So you put your tiny box or your tiny rack in your house and then you get your Kama body and your Kama body runs the models on that. It's, it's close, right? It's not, you don't have to go to some cloud, which is, you know, 30 milliseconds away. You go to a thing which is 0.1 milliseconds away.

• 3:01:43Lex Fridman?

  So the AI girlfriend will have like a central hub in the home?

• 3:01:48George Hotz?

  I mean, eventually, if you fast forward 20, 30 years, the mobile chips will get good enough to run these AIs.

• 3:01:54Lex Fridman?

  Yeah.

• 3:01:55George Hotz?

  But fundamentally, it's not even a question of putting legs on a tiny box. Because how are you getting 1.5 kilowatts of power on that thing? Right? So you, you need, they're, they're very synergistic businesses. I also want to build all of Kama's training computers. I Kama builds training computers right now. We use commodity parts. I think I can do it cheaper. So we're gonna build, Tiny Corp is gonna not just sell Tiny Box, Tiny Box is the consumer version, but I'll build training data centers too.

• 3:02:23Lex Fridman?

  Have you talked to Andre Capati, or have you talked to Elon about Tiny Corp?

• 3:02:27George Hotz?

  He went to work at OpenAI.

• 3:02:29Lex Fridman?

  What do you love about Andre Capati? To me, he's one of the truly special humans we got.

• 3:02:35George Hotz?

  Oh, man, like, you know, his streams are just a level of quality so far beyond mine. Like, I can't help myself. Like, it's just, it's just, you know.

• 3:02:44Lex Fridman?

  Yeah, he's good.

• 3:02:45George Hotz?

  He wants to teach you.

• 3:02:48Lex Fridman?

  Yeah.

• 3:02:49George Hotz?

  I want to show you that I'm smarter than you.

• 3:02:51Lex Fridman?

  Yeah, he has no, I mean, thank you for the sort of, the raw, authentic honesty. I mean, a lot of us have that. I think Andre is as legit as it gets in that he just wants to teach you, and there's a curiosity that just drives him. And just like, at his, at the stage where he is in life, to be still like, one of the best tinkerers in the world.

• 3:03:15George Hotz?

  Yeah.

• 3:03:15Lex Fridman?

  It's crazy. Like, to, what is it? Micrograd?

• 3:03:20George Hotz?

  Micrograd, yeah. Inspiration for Tinygrad.

• 3:03:24George Hotz?

  The whole, I mean, his CS 231N was, this was, this was the inspiration. This is what I just took and ran with and ended up writing this, so, you know.

• 3:03:32Lex Fridman?

  But I mean, to me, that-

• 3:03:34George Hotz?

  Don't go work for Darth Vader, man.

• 3:03:36Lex Fridman?

  I mean, the flip side to me is that the fact that he's going there is a good sign for OpenAI. I think, you know, I like Ilya Sutskever a lot. I like those, those guys are really good at what they do.

• 3:03:51George Hotz?

  I know they are. And that's kind of what's even like more. And you know what? It's not that OpenAI doesn't open source the weights of GPT-4. It's that they go in front of Congress. And that is what upsets me. You know, we had two effective altruist Sams go in front of Congress. One's in jail.

• 3:04:10Lex Fridman?

  I think you're drawing parallels on there.

• 3:04:13George Hotz?

  What's in jail?

• 3:04:14Lex Fridman?

  You give me a look. Give me a look.

• 3:04:17George Hotz?

  No, I think, I think effectual altruism is a terribly evil ideology.

• 3:04:21Lex Fridman?

  Oh, yeah, that's interesting. Why do you think that is? Why, why do you think there's something about a thing that sounds pretty good? That kind of gets us into trouble.

• 3:04:29George Hotz?

  Because you get Sam Bankman Fried. Like Sam Bankman Fried is the embodiment of effective altruism. Utilitarianism is an abhorrent ideology. Like, like, well, yeah, we're gonna kill those three people to save a thousand, of course.

• 3:04:43Lex Fridman?

  Yeah.

• 3:04:43George Hotz?

  Right? There's no, there's no underlying, like, there's just, yeah.

• 3:04:48Lex Fridman?

  Yeah, but to me, that's a bit surprising. But it's also, in retrospect, not that surprising. But I haven't heard really clear kind of,

• 3:05:01Lex Fridman?

  like rigorous analysis why effective altruism is flawed.

• 3:05:06George Hotz?

  Oh, well, I think charity is bad. Right? So what is charity but investment that you don't expect to have a return on? Right?

• 3:05:13Lex Fridman?

  Yeah, but you can also think of charity as like,

• 3:05:17Lex Fridman?

  is you would like to see, so allocate resources in optimal way to, to make a better world.

• 3:05:25George Hotz?

  And probably almost always that involves starting a company.

• 3:05:28Lex Fridman?

  Yeah.

• 3:05:29George Hotz?

  Right? Because

• 3:05:30Lex Fridman?

  More efficient.

• 3:05:30George Hotz?

  Yeah. If you just take the money and you spend it on malaria nets, you know, okay, great. You've made 100 malaria nets, but if you teach.

• 3:05:38Lex Fridman?

  Yeah. Man, how to fish.

• 3:05:40George Hotz?

  Right? Yeah.

• 3:05:41Lex Fridman?

  No, but the problem is teaching a man how to fish might be harder. Starting a company might be harder than allocating money that you already have.

• 3:05:48George Hotz?

  I like the flip side of effective altruism, effective accelerationism. I think accelerationism is the only thing that's ever lifted people out of poverty. Um, the fact that food is cheap. Not, we're giving food away because we are kind hearted people. No, food is cheap. And that's the world you want to live in. UBI, what a scary idea.

• 3:06:09George Hotz?

  What a scary idea. All your power now? Your, your money is power? Your only source of power is granted to you by the goodwill of the government? What a scary idea.

• 3:06:19Lex Fridman?

  So you even think long term, even?

• 3:06:23George Hotz?

  I'd rather die than need UBI to survive, and I mean it.

• 3:06:30Lex Fridman?

  What if survival is basically guaranteed? What if our life becomes so good?

• 3:06:34George Hotz?

  You can make survival guaranteed without UBI. What you have to do is make housing and food dirt cheap.

• 3:06:40Lex Fridman?

  Sure.

• 3:06:41George Hotz?

  Right? And that's the good world. And actually, let's go into what we should really be making dirt cheap, which is energy. That energy, that, you know, oh my God, like, you know, that, that's, if there is one, I'm pretty centrist politically. If there's one political position I cannot stand, it's deceleration. It's people who believe we should use less energy.

• 3:07:01Lex Fridman?

  Yeah.

• 3:07:01George Hotz?

  Not people who believe global warming is a problem. I agree with you. Not people who believe that, you know, saving the environment is good. I agree with you. But people who think we should use less energy. That energy usage is a moral bad. No. No, you are asking, you are diminishing humanity.

• 3:07:20Lex Fridman?

  Yeah, energy is flourishing, of creative flourishing of the, the human species.

• 3:07:25George Hotz?

  How do we make more of it? How do we make it clean? And how do we make, just, just, just, how do, how do I pay, you know, 20 cents for a megawatt hour instead of a kilowatt hour?

• 3:07:34Lex Fridman?

  Part of me wishes that Elon went into nuclear fusion versus Twitter. Part of me. Or somebody, somebody like Elon.

• 3:07:43George Hotz?

  You know, we need to, I wish, I wish there were more, more Elons in the world. And I think Elon sees it as like, uh, this is a political battle that needed to be fought. And again, like, you know, I always ask the question of whenever I disagree with him, I remind myself that he's a billionaire and I'm not. So, uh, you know, maybe he's got something figured out that I don't, or maybe he doesn't.

• 3:08:04Lex Fridman?

  To have some humility. But at the same time, me as a person who happens to know him, I find myself in that same position. And sometimes even billionaires need friends who disagree and help them grow.

• 3:08:19Lex Fridman?

  And that's a difficult, that's a difficult reality.

• 3:08:22George Hotz?

  And it must be so hard. It must be so hard to meet people. Once you get to that point where.

• 3:08:27Lex Fridman?

  Fame, power, money, everybody's sucking up to you.

• 3:08:31George Hotz?

  See, I love not having shit. Like, I don't have shit, man. You know, like, like, trust me, there's nothing I can give you. There's nothing worth taking from me, you know?

• 3:08:39Lex Fridman?

  Yeah, it takes a really special human being, when you have power, when you have fame, when you have money, to still think from first principles. Not like all the adoration you get towards you, all the admiration, all the people saying, yes, yes, yes.

• 3:08:52George Hotz?

  And all the hate, too.

• 3:08:53Lex Fridman?

  And the hate.

• 3:08:54George Hotz?

  That gets worse.

• 3:08:54Lex Fridman?

  So the hate makes you want to go to the yes people. Because the hate exhausts you. And the kind of hate that Elon's gotten from the left is pretty intense. And so that, of course, drives him right. And it loses balance and.

• 3:09:12George Hotz?

  And it keeps this absolutely fake, like psyop political divide alive, so that the 1% can keep power, like.

• 3:09:20Lex Fridman?

  Yeah. I wish we'd be less divided, because it is giving power.

• 3:09:25George Hotz?

  It gives power.

• 3:09:25Lex Fridman?

  To the ultra powerful.

• 3:09:28Lex Fridman?

  The rich get richer. You have love in your life. Has love made you a better or a worse programmer?

• 3:09:37Lex Fridman?

  Do you keep productivity metrics?

• 3:09:39George Hotz?

  No, no. No, I'm not, not that. I'm not that methodical. I think that there comes to a point where if it's no longer visceral, I just can't enjoy it. I still viscerally love programming. The minute I started like.

• 3:09:55Lex Fridman?

  So that's one of the big loves of your life is programming?

• 3:09:58George Hotz?

  Oh, I mean, just my computer in general. I mean, you know, I tell my girlfriend, my first love is my computer, of course. Right? Like, you know, I sleep with my computer. It's there for a lot of my sexual experiences. Like, come on, so is everyone's, right? Like, you know, you gotta be real about that and like.

• 3:10:14Lex Fridman?

  Not just like the IDE for programming, just the entirety of the computational machine.

• 3:10:19George Hotz?

  The fact that, yeah, I mean, it's, you know, I wish it was, uh, someday they'll be smarter and someday, you know, maybe I'm weird for this, but I don't discriminate, man. I'm not gonna discriminate bio stack life and silicon stack life, like.

• 3:10:30Lex Fridman?

  So the moment the computer starts to say, like, I miss you, it starts to have some of the basics of, uh, human intimacy? It's over for you. The moment VS Code says, hey, George.

• 3:10:42George Hotz?

  No, you see, no, no, no, but VS Code is, no, they're just doing that. Microsoft's doing that to try to get me hooked on it. I'll see to it. I'll see to it. It's gold digger, man. It's gold digger.

• 3:10:51Lex Fridman?

  Look, you're an open source.

• 3:10:53George Hotz?

  Well, this just gets more interesting, right? If it's, if it's open source and yeah, it becomes.

• 3:10:57Lex Fridman?

  Though Microsoft's done a pretty good job on that.

• 3:10:59George Hotz?

  Oh, absolutely. No, no, no, look, I think Microsoft, again, I wouldn't count on it to be true forever. But I think right now, Microsoft is doing the best work in the programming world. Like between GitHub, GitHub Actions, VS Code, the improvements to Python. Where's Microsoft? Like, this is.

• 3:11:16Lex Fridman?

  Who, who would have thought Microsoft and Mark Zuckerberg are spearheading the open source movement?

• 3:11:22George Hotz?

  Right, right? How, how things change.

• 3:11:27Lex Fridman?

  Oh, it's beautiful.

• 3:11:29George Hotz?

  By the way, that's who I'd bet on to replace Google, by the way.

• 3:11:31Lex Fridman?

  Who?

• 3:11:32George Hotz?

  Microsoft.

• 3:11:33Lex Fridman?

  Microsoft.

• 3:11:34George Hotz?

  I think Satya Nadella said, straight up, I'm coming for it.

• 3:11:37Lex Fridman?

  Interesting. So your bet, who wins AGI?

• 3:11:41George Hotz?

  Oh, I don't know about AGI. I think we're a long way away from that. But I would not be surprised if in the next five years, Bing overtakes Google as a search engine.

• 3:11:49Lex Fridman?

  Interesting.

• 3:11:50George Hotz?

  Wouldn't surprise me.

• 3:11:52Lex Fridman?

  Interesting.

• 3:11:55Lex Fridman?

  I hope some startup does.

• 3:11:58George Hotz?

  There might be some startup too. I would, I would equally bet on some startup.

• 3:12:02Lex Fridman?

  Yeah, I'm like 50 50.

• 3:12:04George Hotz?

  Yeah.

• 3:12:04Lex Fridman?

  But maybe that's naive.

• 3:12:05George Hotz?

  Yeah.

• 3:12:06Lex Fridman?

  I believe in the power of these language models.

• 3:12:09George Hotz?

  Satya is alive. Microsoft's alive.

• 3:12:11Lex Fridman?

  Yeah. It's great. It's great. I like all the innovation in these companies. They're not being stale. Uh, and to the degree they're being stale, they're losing. So there's a huge incentive to do a lot of exciting work and open source work, which is, this is incredible.

• 3:12:26George Hotz?

  Only way to win.

• 3:12:28Lex Fridman?

  You're older, you're wiser. What's the meaning of life, George Hots?

• 3:12:34George Hotz?

  To win.

• 3:12:35Lex Fridman?

  It's still to win.

• 3:12:36George Hotz?

  Of course.

• 3:12:38Lex Fridman?

  Always.

• 3:12:39George Hotz?

  Of course.

• 3:12:40Lex Fridman?

  What's winning look like for you?

• 3:12:41George Hotz?

  I don't know. I haven't figured out what the game is yet, but when I do, I want to win.

• 3:12:45Lex Fridman?

  So it's bigger than solving self-driving? It's bigger than, uh, democrat, democratizing, decentralizing compute?

• 3:12:54George Hotz?

  I think the game is to stand eye to eye with God.

• 3:12:59Lex Fridman?

  I wonder what that means for you. Like, at the end of your life, what that will look like.

• 3:13:06George Hotz?

  I mean, this is what like, I don't know, this is some, this is some,

• 3:13:11George Hotz?

  there's probably some ego trip of mine, you know? Like, do you want to stand eye to eye with God? He's just blasphemous man, okay? I don't know. I don't know. I don't know if it would upset God. I think he like wants that. I mean, I certainly want that for my creations. I want my creations to stand eye to eye with me. So why wouldn't God want me to stand eye to eye with him?

• 3:13:33George Hotz?

  That's the best I can do, Golden Rule.

• 3:13:37Lex Fridman?

  I'm just imagining the creator of a video game

• 3:13:41Lex Fridman?

  having to, uh, look and stand eye to eye with one of the characters.

• 3:13:48George Hotz?

  I only watched season one of Westworld, but yeah, we got to find the maze and solve it, like.

• 3:13:53Lex Fridman?

  Yeah. I wonder what that looks like. It feels like a really special time in human history where that's actually possible. Like, there's something about AI that's like, we're playing with something weird here. Something really weird.

• 3:14:07George Hotz?

  I wrote a blog post, uh, I reread Genesis and just looked like, they give you some clues at the end of Genesis for finding the Garden of Eden. And I'm interested. I'm interested.

• 3:14:20Lex Fridman?

  Well, I hope you find just that, George. You're one of my favorite people. Thank you for doing everything you're doing. And in this case, for fighting for open source and for decentralization of AI. It's a, it's a fight worth fighting, fight worth winning hashtag. I love you, brother. These conversations are always great. Hope to talk to you many more times. Good luck with Tiny Corp.

• 3:14:41George Hotz?

  Thank you. Great to be here.

• 3:14:43Lex Fridman?

  Thanks for listening to this conversation with George Hots. To support this podcast, please check out our sponsors in the description. And now, let me leave you with some words from Albert Einstein. Everything should be made as simple as possible, but not simpler. Thank you for listening and hope to see you next time.