E22: Helping Businesses Use AI with Rachel Woods of The AI Exchange

Nathan Labenz sits down with Rachel Woods, founder of The AI Exchange, an AI education platform with over 130,000 subscribers on TikTok. They discuss Rachel’s experience consulting for public and private companies on AI strategy, how teams should think about their AI strategy, and how Rachel uses AI in her own business. Rachel was previously a data scientist at Meta, and a founder of a venture-backed e-commerce startup, where she began to use chatGPT in her startup operations. This episode is the second in a series centered on talking to rising voices in AI media: people who are now only working overtime to understand everything going on in AI, but also creating thought leadership and educational content meant to help others get up to speed as well. -- We're hiring across the board at Turpentine and for Erik's personal team on other projects he's incubating. He's hiring a Chief of Staff, EA, Head of Special Projects, Investment Associate, and more. For a list of JDs, check out: eriktorenberg.com. RECOMMENDED PODCAST: The HR industry is at a crossroads. What will it take to construct the next generation of incredible businesses – and where can people leaders have the most business impact? Hosts Nolan Church and Kelli Dragovich have been through it all, the highs and the lows – IPOs, layoffs, executive turnover, board meetings, culture changes, and more. With a lineup of industry vets and experts, Nolan and Kelli break down the nitty-gritty details, trade offs, and dynamics of constructing high performing companies. Through unfiltered conversations that can only happen between seasoned practitioners, Kelli and Nolan dive deep into the kind of leadership-level strategy that often happens behind closed doors. Check out the first episode with the architect of Netflix’s culture deck Patty McCord. https://link.chtbl.com/hrheretics TIMESTAMPS: (00:00) Episode preview (02:59) Rachel Woods’ story (04:30) State of the public who comes across videos (9:49): How do you frame AI to business owners and how should they use it? (15:29) Sponsor: Omneky (18:58) How should teams set up their AI strategy for success? (26:41) The leap from GPT 3.5 to GPT-4 (30:42) Step changes of AI and public perception of capabilities (35:17) Common misconceptions people have about AI (36:22) How Rachel uses AI in her business (44:09) What’s your style of prompt engineering for content creation? (46:01) How AI will change the way we use a computer in the next year? (49:34) Bing launch and how companies launch AI products (54:15) AI safety and restraint (1:00:11) Alpaca and LLaMA (1:04:16) Rachel’s favorite products in AI (1:05:24) Would Rachel get a Neuralink implant? (1:07:55) AI hopes and fears TWITTER: @CogRev_Podcast @labenz (Nathan) @eriktorenberg (Erik) @rachel_l_woods (Rachel) SPONSORS: Shopify is the global commerce platform that helps you sell at every stage of your business. Shopify powers 10% of ALL eCommerce in the US. And Shopify's the global force behind Allbirds, Rothy's, and Brooklinen, and 1,000,000s of other entrepreneurs across 175 countries. From their all-in-one e-commerce platform, to their in-person POS system – wherever and whatever you're selling, Shopify's got you covered. With free Shopify Magic, sell more with less effort by whipping up captivating content that converts – from blog posts to product descriptions using AI. Sign up for $1/month trial period: https://shopify.com/cognitive Thank you Omneky for sponsoring The Cognitive Revolution. Omneky is an omnichannel creative generation platform that lets you launch hundreds of thousands of ad iterations that actually work, customized across all platforms, with a click of a button. Omneky combines generative AI and real-time advertising data. Mention "Cog Rev" for 10% off. More show notes and reading material released in our Substack: https://cognitiverevolution.substack.com/

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

• 0:00Speaker 00

  The most absurd conversation that I have is like literally from TikTok. People reached out and they wanted me to join their two hour long board meeting. Public pharma company, like what they hired me to do is to sit there and just like be a sounding board for as they're thinking through what does an AI strategy mean? If everybody can post on Twitter and everybody has an AI that just like takes their raw thoughts and puts them into like high performing tweets and everybody is using these agents. Okay, is anybody actually on Twitter anymore? If you're paying attention, you can feel at least like you're making progress. The number one thing I recommend to people is start using ChatGPT with an automation tool like Zapier or Make.

• 0:42Speaker 00

  We are using AI for like so many things that are like so serious. And honestly, sometimes it's just really fun to use AI to make much better memes. And like, I've never been a person who's good at making memes before. And now I feel like I can make ones that make me laugh at least.

• 0:56Nathan Labenz?

  Hello and welcome to The Cognitive Revolution, where we interview visionary researchers, entrepreneurs and builders working on the frontier of artificial intelligence. Each week we'll explore their revolutionary ideas and together we'll build a picture of how AI technology will transform work, life and society in the coming years. I'm Nathan Labenz, joined by my co-host Eric Torenberg. Hi, everyone, and welcome back to The Cognitive Revolution. Today, we continue our AI in Media series with special guest Rachel Woods. Rachel was previously a data scientist at Meta, where she worked on AI systems that support the small business advertising platform. And she was also the founder of Vinebase, an e-commerce and marketing platform focused on small vineyards,

• 1:40Nathan Labenz?

  which she sold to Corksi in 2022.

• 1:44Nathan Labenz?

  Today, Rachel is the founder of the AI Exchange and best known for her Twitter and especially her TikTok profile, where she creates roughly one video per day to help people keep up with and understand all the latest AI news. She also dispenses practical advice for both individuals and companies.

• 2:03Nathan Labenz?

  The AI Exchange publishes a regular newsletter which you can subscribe to at news.theaixchange.com and she provides additional resources, community, and even optional consulting services for subscribers. Rachel and I talked about the kind of audience she is reaching, just how quickly people are waking up to the opportunity of AI, how many companies that didn't have AI in their 2023 plans are now scrambling to adapt, how she tries to consult with and guide those companies forward, and lots, lots more. I hope you enjoy this delightful conversation with Rachel Woods. Rachel Woods, welcome to the cognitive revolution.

• 2:42Speaker 00

  Thank you. Thanks so much for having me.

• 2:44Nathan Labenz?

  Yeah, I'm excited to talk to you. I think you have staked out a really interesting position for yourself as a sort of very smart and plugged in AI influencer, popularizer, educator. So I'd love to start off by just asking you a little bit about like, how you conceive of the role that you're playing? And, you know, who your audience is? And how you decided to do this in such a, at such an opportune time?

• 3:12Speaker 00

  Yeah, so a little bit of luck and foresight, I think, is kind of the, the moral of my story in the space. So I actually started my TikTok account November 1st, which if you think back to the timing of that, before the whole world cared about ChatGPT, it's almost comical that that was like, you know, the month before everything imploded or exploded, however you want to think about it. But yeah, I mean, so my story is I was a data scientist before. And then I founded a venture-backed startup in the e-commerce space where we were actually using GPT-3 a bunch in that product.

• 3:56Speaker 00

  And after kind of that experience and seeing the impact that large language models were having on what we were building, I was like, okay, the world like maybe doesn't care yet, but the world needs to know that AI can do this crazy thing where it writes for you. And that ended up being like that, plus a little bit of a joke of like, I'm gonna go be a TikTok influencer was the impetus to go start a TikTok account. Which, yeah, I mean, it's just been a crazy journey since then.

• 4:21Nathan Labenz?

  So now you're reaching regularly. Something like a hundred thousand people a day and obviously there's, you know, spikes on top of that on TikTok. What is the, what is the state of the public, you know, the broad public right now who just like comes across your videos? I'm so deep down the rabbit hole myself and I think probably most of our listeners are that I, you know, sometimes find it's like easy to lose touch with what somebody who's cruising TikTok and like sees one of your videos for the first time, you know, might be pretty supposed to think.

• 4:52Speaker 00

  Yeah, I mean, a lot of people are surprised when I tell them my stats. So myself included, like I feel like I originally thought that TikTok was a lot of young people doing dances or like funny memes. But 90% of my audience is over the age of 25 and 50% is over the age of 35. And they're predominantly US-based. And then I have the really fun privilege of people reach out to me and like book time with me and, and wanna talk. And so I've found out that they're mostly business leaders, professionals, founders and engineers, literally like on TikTok nights and weekends, like listening and learning about AI.

• 5:33Speaker 00

  And then funny enough, sharing my TikToks in their company's Slack channels, which I just like sit back for a minute and like think about like being at work and just seeing like a stream of TikToks about AI in your Slack is pretty funny. But yeah, I mean, like, I know we're in the AI bubble. I mean, I'm very active on Twitter too, but one thing that's been truly like amazing over the last three, four, five months has just been like the pure like mainstream, especially business explosion and interest in, in AI. You know, I'm sure you've seen that as well with, with what you're doing, but it's really there, which has been pretty incredible.

• 6:17Nathan Labenz?

  Yeah, it seems like we're, I feel like we're in this kind of probably pretty short in-between moment where, you know, my joke right now is that I'm, I restarted the calendar on the release of GPT-4. So we're now at GPT-4 plus one month to the day, I believe. And.

• 6:37Nathan Labenz?

  It's, you know, this huge step change where all of a sudden you're seeing like pretty sophisticated reasoning, not quite breakthrough insight, but you know, like professional grade reasoning on just a super wide range of things. And yet that's not really deployed almost anywhere at this point. And I have no idea like to what degree people are, again, like outside of the bubble.

• 7:01Speaker 00

  Like aware.

• 7:02Nathan Labenz?

  Yeah, just have any concept of whether that's happening or not. Cause the messages are so confused, right? Like at the same time, you'll see people saying like, it's all hype and you know, this technology is never going to work. We need a totally different paradigm. And then, you know, at the same time, like other people were reporting that it's, you know, matching human doctors. So yeah, I just, I don't know. It's like hard to get a read on, but it sounds like you are mostly reaching people that are much more plugged in and kind of shepherding them through education as opposed to, you know, you don't see too many randoms on your page.

• 7:36Speaker 00

  Yeah, I mean, I would say like the audience that I really speak to is the early adopters across almost every industry. So kind of how I decided to go from Tik Tok being a fun side thing to it being my full time thing and building a whole company around it was I put a link in my bio, January 1st. And within 24 hours, I had six people purchase time with me, so talk to me about not like the latest prompts or even kind of like how to log into the mid journey discord or discord. But it was like, okay, I see that there's a huge shift in AI. What should my business even be doing in this next wave?

• 8:21Speaker 00

  And I feel like that is actually like a side of the AI conversation we like, don't see on Twitter, we don't see in the doomsday conversation is like, there are just so many businesses across like every single industry that are actually sitting down with their board, with their team and thinking, okay, should we be having like rethinking how AI is brought into like our product or, or our offering? So then from that lens, like some of the things that. I think the the AI bubble people are like paying attention to around like these emerging capabilities of reasoning. Like, other than a lot of ways, the leaders across these other industries are paying attention to those types of trends. Because they're trying to figure out what does an AI strategy even mean, right?

• 9:05Speaker 00

  Is an AI strategy that I'm going to have like prompts in my offering to my customers? Or is an AI strategy that I'm going to have to figure out how to deploy agents? Right? Um, and I think that's actually the conversation that's going on, even if you're a lawyer, a dentist, a uh, CEO of a software company that's based out of, you know, uh, maybe not Silicon Valley, like it really is like the conversation people are having, which is fascinating.

• 9:31Nathan Labenz?

  I uh, think those conversations sound pretty fascinating and it sounds like you've had a wide range of them. That probably literally does span like, you know, local, small business owner, all the way up to kind of CEO of a, you know, significant company. How do you frame things for them when somebody comes to you and is like, okay, you know, here's my deal. Um, and maybe the answer is like, it's highly contextual, which would be interesting in and of itself, but I'm sure you have some kind of high level frame that you offer, you know, make a couple of conversions to people who are like, I've, you know, I've used JATGPT a little bit, I've seen it on the news, but I'm really kind of at a loss for what this means for my business.

• 10:14Speaker 00

  Yeah, I mean, so uh, Nathan, the funniest or like the most absurd conversation that I've had is like literally from TikTok. People reached out and they wanted me to join their two hour long board meeting, public company, public pharma company. And my, like what they hired me to do is to sit there and just like be a sounding board for as they're thinking through. Like, what does an AI strategy mean? Like, that's the level at which people are actually having conversations.

• 10:42Speaker 00

  Yeah, I mean, there's definitely some commonalities, you know. I have done over a hundred of these calls since the beginning of the year. And so I've started to like distill down things into resources, which is like a lot of what we put the business around. But, you know, one question is like, are you going to use AI to improve your internal operations? Or are you going to use AI to offer a new service or product or experience to your customers? And like, those are, those are pretty distinct conversations a lot of times. And I often talk to people about how to think about risk, right? There's a lot more risk in offering an AI solution to your customers today, just on like where the technology and the capabilities and human loop and these other like questions.

• 11:25Speaker 00

  Limitations we have are versus starting to figure out, you know, how you can use it in your internal operations. I usually recommend companies start there because they get more of a feedback loop and a sense of the technology. And like, you're going to learn stuff that's going to help you figure out how to go and maybe deliver something like different to your customer. But, yeah, like, I think the thing that would surprise most people is how serious these conversations are and like how like legit. It's not just like the AI bubble, people saying the words like human in the loop, you know? It is, it is a much broader audience.

• 12:06Nathan Labenz?

  Yeah, fascinating. I agree with your advice, for what it's worth, to start with something in your own operations. That's certainly how I got comfortable and, you know, to the point where I felt qualified. Speaking publicly about this technology was really just breaking. And I have, you know, probably sounds like a pretty similar riff. I always talk about identifying discrete tasks that AI can do. And then embedding that into like a broader process, you know, that takes advantage of the ability to delegate to AI and. A lot of times that can be pretty analogous to tasks that are currently done. Sometimes it can be like new tasks or, you know, it can enable scale in new ways. But just kind of starting with those discrete things of like, is there a sort of input output

• 12:49Nathan Labenz?

  where it's like hard now? And if I had, you know, infinite interns or whatever, it could in theory be easy. But if you can identify those places, you get off to a pretty good start and then you feel real value. You definitely learn things. You see some failure modes and hopefully, you know, develop a healthy respect for the technology. But yeah, I agree, that's a very good place for most to start.

• 13:15Speaker 00

  You know, like when I think about like what to create content on, like I think there are people in the space who are creating content for the objective of getting views, growing an account, growing an audience, growing distribution. And, you know, early on, I like, I think you have to make a decision of like whether that's going to be your objective in this space because you go down a certain path. Or if your objective is going to be more playing a little bit of the long game, right? And creating like really a relationship more as a creator, I guess, than an influencer where you're creating value for people and helping them actually like navigate what's going on, right?

• 13:56Speaker 00

  I think like what, what I see you talk about online is like very much in these like same, same objectives, which is, hey, we sitting here, look at how AI is going to change the way we operate business. And we also see that if people don't really have kind of the right way to think about it, or not the full information, or they're not really paying attention, or they're not inspired to get like small wins, then like this space is going to keep moving and people are not going to be, you know, prepared to ride that wave. And I think that that's like. It's just so important and it's like where I create content from is like if someone can not feel

• 14:40Speaker 00

  concerned about the future of their business because like they watched a few videos, right? Or because they like read my newsletter than like that's, that's the win. Like, how do you arm people to, to actually feel confident navigating this next wave? Because I think you and I both sitting here like see there's just gonna be a huge change.

• 15:00Speaker 02

  Hey, we'll continue our interview in a moment after a word from our sponsors. I want to tell you about my new interview show, Upstream. Upstream is where I go deeper with some of the world's most interesting thinkers to map the constellation of ideas that matter. On the first season of Upstream, you'll hear from Marc Andreessen, David Sacks, Balaji, Ezra Klein, Joe Lonsdale, and more. Make sure to subscribe and check out the first episode with A16Z's Marc Andreessen. The link is in the description.

• 15:29Nathan Labenz?

  Do you know anyone who feels if you could, point to anyone who's confident in navigating this, I'd be very interested. I don't feel very confident at all. Do you feel confident?

• 15:38Speaker 00

  I mean, I feel confident in, in like, it's just a step by step, like paying attention and trying to make smart decisions along the way. Right? Like, I'm not, I'm not sitting here saying like, okay, like people have it totally solved. People know how it's gonna play out. But I do think there's a level of like, if you're paying attention, you can feel at least like you're making progress. Right? And then, you know, like repurposing content, the thing that used to take you like 30% of your time as a content creator, okay, now takes you 2%, right? And like, that's progress. And I think like confidence can come out of like those, you know, small wins.

• 16:15Nathan Labenz?

  Yeah, it's fascinating. I, I love the space in many ways. And one of the biggest ones is just how much I am constantly learning. I try to. Target like 50% of my, you know, working time just on kind of understanding what's going on and probably do fall a bit short of that, but you know, have to maintain a pretty good amount or you know, things get stale so quickly. I've, the number of times that I've kind of gone back and looked at writing from two weeks ago and felt like, well, it needs at a minimum, like a significant update is just crazy.

• 16:48Speaker 00

  Uh, yeah, I mean, it's interesting that you say you don't feel like confident in that. Cause like, I guess I think, um, I think looking back and seeing how much the space is changing and then realizing that it's changing, like that's, that's actually just like what's going on right now. Right? It's like, we're all having to do that.

• 17:06Nathan Labenz?

  Yeah, no doubt. Uh, and I like it, but I do also at the same time feel like my ability, I think like I can see maybe six months out, you know, with some amount of clarity. And when I, you know, talk to people on the podcast, I'm like, I like apologize for even asking them to envision 2030, or occasionally I have asked and they're like, 2030, what are you nuts? You know, like that's, uh, that's beyond anyone's event horizon. So yeah, I mean, I guess there's different meanings of confidence. I would, you know, I'm with you in that. I do feel like I have enough command of the current situation to, you know, at least help people start to get up to speed. But I, I'm very reluctant

• 17:49Nathan Labenz?

  to make, you know, any predictions that kind of get beyond 2023 at this point.

• 17:55Speaker 00

  I totally agree, man.

• 17:56Nathan Labenz?

  So what is going back to the, the kind of board level conversations or, you know, where you're interacting with companies as a strategic advisor. I'm kind of keeping in mind this, you know, large pharma company type example. I have a fair degree of uncertainty around how quickly things are going to happen over, you know, the course of like this, you know, I've called it the great implementation of kind of, we have GPT four, it's, you can see a pretty clear path of being everywhere. You know, I am confident that there is a wave of transformation coming based on that deployment, but the speed of the wave, you know, and exact kind of timing dynamics, how it plays out, you know, a lot more uncertainty

• 18:39Nathan Labenz?

  there. So what have you seen in terms of the ambition of these companies? You could talk about that in a couple of ways, like timelines, but also, you know, like a pharma company, I could see.

• 18:52Nathan Labenz?

  You know, wisely, I would say, you know, starting with something small and internal and, you know, like expense receipts, you know, you could, you can reform all these little things that are kind of annoying today. But then you could also be like, you could imagine a world where it's instead of talk to your doctor about our drug, it's like, talk to our language model on our website as kind of the first thing you do before, you know, we then encourage you to go talk to a doctor. So like, how ambitious are these companies thinking right now? And, and when do you think we see like the small and then the bigger changes?

• 19:26Speaker 00

  The first thing that I see is most people, in fact, like I would almost say all with an asterix, didn't have AI in their 2023 roadmap or budget. And so most of the conversations that I was helping companies have either directly or kind of supporting through content was how to think about like having that conversation and what investments might look like. I recommend to a lot of companies that they do, they kind of take a page out of Silicon Valley's book and think about doing an internal hackathon with their team. Like, I think we're starting to see some of those start to play out over the next like couple of months, where companies are at least just giving their team the space to

• 20:10Speaker 00

  experiment and see where value could come from within anything, especially internal operations. You know, when you ask about like the, the really big broad strokes of like, wow, we don't recognize like what it means to be an e-commerce company, or what it means to be a pharma company, or what it means to do X, Y, and Z. Like, I do think those things are much further out and like much fuzzier, but I guess this is my bias. Like, I, I sit in the day to day of like, your team does so much manual work that they know is manual work, they don't love doing, they're sitting there on the slog and like, they want to use these tools. How do you enable them to use some of these tools and kind of this like productivity unlock?

• 20:53Speaker 00

  And I think that's happening, like, you know, now and exceedingly so in the next like three to six months.

• 21:00Nathan Labenz?

  How do you see that shaping up? Because this is something I also have, you know, again, in terms of my limited confidence. If you'd asked me a couple of months ago, I was very like,

• 21:11Nathan Labenz?

  We got to curate the best tools. And in particular, I'm advising a friend's company, which you know about, which is called Athena, which is in the executive assistant space. And as of January, I was like, we need to test a ton of tools. We need to curate the best tools. And then we'll have like this Swiss Army knife of like proven things. It's going to be awesome. And, you know, we'll train all the EAs on how to use them. And that'll be like ultimately kind of a part of how we can position the company. And I don't think that's wrong, but I do think the number of tools that I expected to be in that set has dropped precipitously

• 21:49Nathan Labenz?

  to where I'm now like, it might be like three, you know, or, or even maybe just kind of one core one plus like other things that are naturally built into, you know, spreadsheets and whatnot, where, you know, just the product itself changes. So what's your kind of expectation for that productivity suite? Over, you know, whatever, three to six months, like as that kind of matures, what do you think people are going to be using most?

• 22:18Speaker 00

  So I also think about this question in two ways, which is when I tell people, hey, maybe think about giving your team space to do like an internal hackathon type experience. Most of that is for building up what I think of as like AI literacy. Cause again, I go back to like, most companies did not have AI in their vocabulary before ChatGPT. And so I think we're just still in this space of, you know, people have like tried ChatGPT, but they're not using it on a regular basis because they can't really find like ways that it's useful.

• 22:52Nathan Labenz?

  Also, they're probably still on the free version a lot of the time, which is a huge difference in value.

• 22:58Speaker 00

  Yeah. And there's open, like, people don't really understand data privacy and the ownership of what happens to the prompts and the outputs that goes into that. So like, I think there's just a lot of adoption that needs to happen just from a like learning perspective. And then from the tool side, I'm actually very much in the same boat as you. Like, I encourage a lot of companies and teams to be a little patient with the tools that they test out. My hot take is I say an annual plan is one of the worst things that you can

• 23:34Speaker 00

  like commit to in this market because it's just changing so fast and you never know, like, one month, the AI sales assistant tool could be the thing that you think is going to really like enable your team and the next month it could be something totally different. Right? But I think that like some of the best success stories I've seen are when companies and people building have like really close relationships and they're kind of like iterating through that together. So I think we're still so early on the the tools themselves in a lot of regards as well.

• 24:06Nathan Labenz?

  But here's a hot take back. I think it's shaping up to be ChatGPT as kind of the Google, you know, the canonical name brand, you know, iPhone equivalent of using an AI chatbot. And it seems like there will be certainly alternatives, but it seems like OpenAI has a pretty substantial lead and for those that are really taking advantage of this stuff, it seems like that is going to be the hub. That's kind of my most likely projection for like later this year. It's like plugins come online, they work well and everybody's like ChatGPT plus like it's the thing, that's what we're using.

• 24:49Speaker 00

  I think that could happen. I still feel like it's a little early. I think there are some dynamics that haven't like played out totally yet, just from as I've like talked to people. I mean, one thing

• 25:01Speaker 00

  that a lot of people overestimate, I think, is like how many people are using ChatGPT on a regular basis. A lot of people have tried it, a lot of people haven't had success with it yet because maybe their prompt wasn't very good, right? Or they didn't figure out the right workflow. So I think we're still really early, but that could happen for sure.

• 25:21Nathan Labenz?

  How much, so for for me, I think the core thing is just. The leap from three point five to four is so big. I wonder if you would, you know, agree with this general assessment, but in my experience, like so many things have gone from at three point five, you know, still this kind of like art of eliciting decent performance, you know, really got to tinker and drill in on the the instructions and often, you know, a couple of examples and you're, you know, you got, I typically had to think still pretty hard to get good results from, you know, anything up to three point five.

• 25:59Speaker 00

  Yeah.

• 26:00Nathan Labenz?

  I find with four, it's honestly pretty easy most of the time. And, you know, I haven't like, so this company Athena is like a thousand plus people. I have a standing open invitation. Anyone can just send me whatever they want help with. And I usually find that like within a half an hour, I can kind of take their inputs and get to like a decent working output. How would you characterize that leap and kind of, you know, to me, that's like the thing. If you went to chat GPT and you use three point five and you like didn't get great results, like I would bet four out of five of those people would get good results if they had just used GPT four.

• 26:37Speaker 00

  I mean, I'm curious, when you have this open standing invitation, like how much of that do you feel like is because you're a good prompt engineer now?

• 26:46Nathan Labenz?

  Not that much, honestly, anymore. Some, definitely, I use like a number of kind of pretty standard techniques. Like we do, my go to obviously is like some clear instructions, but I usually don't find I have to like overly tinker with or refine those. Usually use like a role, you know, you are a copywriting expert or you are a, you know, a seasoned recruiting, you know, executive or what have you.

• 27:13Nathan Labenz?

  I usually don't even have to use. Examples. I do use like a little bit of kind of know-how when it comes to kind of segmenting the prompt with a little markup to make it, you know, extra clear to the model like which, you know, this is this and that is that. And that's been a huge advance from 3.5 to four like that stuff just works now. Like it can, it can understand that document A and document B are like distinct things at a conceptual level and like synthesize in a way where it used to just kind of confuse even with, you know, decent markup a lot of the time. So honestly, I don't know. I think like, maybe I underestimate how hard it is to catch up, but I feel like I can communicate this stuff in like,

• 27:55Nathan Labenz?

  not a long time and usually do like usually when I get done with a, you know, half hour hour call, like people are, they have a working thing and they, you know, feel like they can you know, continue to elaborate it from there if they want to.

• 28:08Speaker 00

  That's where I feel like we're just still so early because like the techniques that you're describing, I think we're still at a very, very small percentage of like the workforce or companies that even know about those and not for a lack of like interest, but just it's, you know, we're in, I think there's a bubble in the fact that like we're in AI all the time. We're thinking about it all the time. Right? When I talk to businesses that are definitely paying attention to AI, but like they run an e-commerce business, like their day to day is not saying, thinking about like, oh, does GPT four like recognize markdown better now?

• 28:51Speaker 00

  And I think we're going to get there, but yeah, it just, it still feels extremely, extremely early and kind of the widespread adoption. And while I agree for like my use cases, I feel like 3.5 and four, even by myself like thinking like, how hard is this task? Okay. I should use four for it versus 3.5. I think those are also still like very new behaviors. Cause all the stuff is still only like a couple of months old. So yeah, like that's just, I maybe that's just where I, I sit and like who I talk to on a day to day basis. But seeing like, we're still so early in a lot of stuff. And I feel like then that's where like I get excited because

• 29:33Speaker 00

  while we feel like on a week to week basis, we have this huge step changes with things like AutoGPT. A lot of companies are still just wondering what AI can do. And so it's not like from three point five to four, four to AutoGPT feels like these step changes. Instead, it's more just like AI in general feels like this huge step change. And we're all just figuring out what, what are the capabilities, what can it be used for? Does that make sense?

• 30:05Nathan Labenz?

  Yeah, as you're describing that, it kind of is analogous to the grotting process. You know, that it underlies AI improvement, right? There we see these kind of step changes at different scales. Where there's like the three point five to four is like a very significant one. There's all these like little micro ones in between. And then for the public, you're kind of saying like, it's basically zero to one. Like, we didn't have this before and now we have it. And that's kind of the, the step change that people are having to contend with.

• 30:33Speaker 00

  Yeah, I mean, like, uh, just another example, like even in early January, when I was talking to companies, they were describing things like agents, right? Like, oh, can I build it like a chat GPT for my company that goes and finds leads on LinkedIn? And then figures out how to personalize those leads and then we'll send emails and then we'll like follow up to those emails. And I just give it like the goal of like, close this lead. And I was like, okay, like, that's not quite how it works yet, but like, let me show you how to like break stuff down, maybe into prompts. You can start playing. Okay, well, now we have agents, right? But like to that person who was thinking about what AI could do for them, the problem it could solve for their business. They've more seen the step change of AI as opposed to the step change of where we're in the weeds of the actual capabilities.

• 31:18Speaker 00

  You know, and the, um, the technical feasibility of each of those, which, yeah, I mean, I, I find fascinating. Like sitting between the two worlds, they're, uh, sometimes, um, uh, very different and I find that really fun.

• 31:31Nathan Labenz?

  Yeah, I agree. I think it's fascinating perspective that you have and kind of, uh, nexus point that you sit at. I didn't realize how much, um, Kind of higher level corporate consulting you've been doing. I think that is a fascinating space right now. Do you run into like, I mean, it's interesting, right? You're, as I understand you, an individual, and I know you have the business as well, but like, you show up for me, you know, with your face and like your personality on TikTok and, and Twitter as well. But I do love the TikTok feed. In fact, you're one of maybe 10 or so TikTok AI creators of various sorts that we have curated a list for the EAs at Athena, because we're just

• 32:13Nathan Labenz?

  like, you know, you guys are collectively and even individually, but certainly collectively, like, creating much more content and keeping folks much more up to date than we possibly can internally. So, you know, we just say like, follow these accounts. And, you know, next time you're on TikTok, we kind of, you know, siphon off a little bit of your entertainment budget for some AI education. I think that's honestly a strategy I'd recommend to a lot of companies as well. It's just kind of sharing like good sources.

• 32:42Speaker 00

  Yeah, I mean, like, literally hundreds of short videos of digestible AI content on my TikTok, not even to mention like, yeah, there are a lot of really talented creators on there. So it is a huge resource.

• 32:56Nathan Labenz?

  Do you run into like Bain? Are you, do you get toward, I mean, to public companies, board meetings, sounds like you'd be to some degree overlapping with like, a much different kind of consultant that I wonder what you've seen or heard about how they are approaching the market?

• 33:12Speaker 00

  So I guess taking a step back, I was a research data scientist at Meta. So like, I worked on like some stuff with embeddings there, like ranking infra, a lot of stuff that now I'm like, wow, those were like fun projects for like having kind of a lens and a take on like what's going on. And then between then and now, I built a like e-commerce like venture backed startup. And so I had the whole experience of like, okay, when you're trying to get something off the ground, like how much, you know, a labor of love and a ton of operations that frankly, like AI can really help with.

• 33:50Speaker 00

  Uh, like that is. And so, um, I've really found my sweet spot and who I love talking to and helping are those like, uh, you know, startups, um, mid-sized companies who are the ones who didn't have AI in their vocabulary very strongly last year. And now are thinking through like, what do I, I do next? Um, I, I think like there's excellent coverage at the, the very top of the market, you know, the Fortune 500s, the Fortune 100s. There's gonna be, you know, a lot of, a lot of big projects, a lot of changes, uh, that I, I do

• 34:30Speaker 00

  hear and have, uh, kind of exposure to, but, um, you know, I, I think back and I mean, even just like the companies that, um, we've worked with and helped. There is so much opportunity if you're a non-AI, let's say marketplace, series A company, to look at the bottlenecks in your marketplace and be able to start thinking about some of these problems in a new way. Um, and so, yeah, that's like the, the space that I've had a lot of fun working with people.

• 34:59Nathan Labenz?

  You know, what are the common misconceptions that people have? And, you know, do you find that people, as they're new to this, like are generally optimistic or pessimistic? But I was kind of thinking of more of like a Main Street audience there and it, um, you know, gathering that it's more of a still like a Silicon Valley audience, but it's like a non-AI Silicon Valley type audience that you're mostly supporting.

• 35:22Speaker 00

  One of the interesting misconceptions is when I say to those like, you know, metrics earlier of like, who's actually watching? Uh, at least my TikTok, like I think there are other AI TikTok creators who are more talking to the mass professional audience or even the mass consumer audience. Uh, you know, those people are exposed to different things. Um, I sit here and I talk about data privacy because I get asked that question like. Twenty times a week, right? And people have like real questions that they're trying to navigate. And so that's a lot of the stuff that I spend my time on. So kind of a different type of like influencer, I guess, in the space.

• 36:04Nathan Labenz?

  Tell me a little bit about how you use AI in your business. Like, what are you finding to be impactful? You know, and how do you anticipate that continuing to evolve?

• 36:14Speaker 00

  Yeah, so one of the things I tell people that I think makes them feel a little bit better is like, I've tried a ton of stuff and a lot of stuff doesn't work. Like, it's not just you, which I think some people like kind of think that they're the problem. It's like, no, no, we're just really early on the tech. We use AI in a bunch of like micro ways. So like things that you you think like, oh, it'd be so nice for this to be automated. But like this is a little bit too complex to automate in the past. Like, throw GPT in there. It's a great, simple like classifier. It's a great, simple, kind of like connecting between like messy processes. And so we have a lot of, a lot of places we're using it there.

• 36:56Speaker 00

  So one is like when, so the way our business works is we have a free newsletter and then people can subscribe to like premium content. They get a community. That's kind of our more like scaled offering for this kind of like consulting or business advice. And so a really big part of that is making sure that we know what are people struggling with? And like what types of businesses, business problems they have? And so we have a form that pulls all that information in. But then, like, we're basically, we have a set of scripts that we run that do kind of like better natural language, understanding of the common themes to inform what type of like content we're creating. Or like, let's say we're creating a certain type of content on a topic.

• 37:40Speaker 00

  What are some of the most common like questions? All that stuff would have taken me like so long in the past to do with old school NLP. Now, super, super easy with a simple script. And then like we'll also use it to like kind of personalize some of the outreach and like, hey, you mentioned this business problem in the past. Like, here's some content we just made. So it's a lot of just like. Yeah, feeling like you spent a hundred hours on something and it was actually like fifteen minutes, right?

• 38:09Nathan Labenz?

  Yeah, that's huge. I sometimes find myself talking about two modes of using AI, one being the kind of co-pilot paradigm where you are prosecuting a task and you have this sort of sidecar helper that can autocomplete for you or maybe answer a question or what have you, right? But you're kind of like owning where are we going right now? And then the other one, which you're speaking to there, where there's even more unlock and I suspect kind of like a tipping point coming where kind of everybody goes to this mode or like else is what I call delegation mode.

• 38:50Nathan Labenz?

  And that is where you're like, I am now going to make this discreet enough that I can actually have the AI do the task. And then I will come in and look at outputs, but I'm not going to sit there with it in real time doing it, right? And it's not going to be kind of assisting me. It's instead actually going to be performing the task. And, you know, again, hopefully, hopefully, the people are still reviewing the outputs. I think that remains important for now at least. But how do you think about, does that dichotomy resonate with you? Or do you have a different way of framing it for people?

• 39:25Speaker 00

  No, I mean, I think that that's, I think it's a great way to think about it. You mentioned the intern analogy earlier, which I also really like. Someone suggested at one point that it was maybe a little bit more PC to use the alien analogy, which is like, you have a thousand aliens coming to help you do stuff. What would you do? We use AI in both types of ways in our business. Like just another one on the co-pilot is like repurposing content, right? Is, I mentioned this earlier, a huge part of creating content that like goes further. And if you have any type of like marketing background or have done that stuff before, like you know that you can just like copy paste the content between platforms.

• 40:06Speaker 00

  But frankly, your mindset is different when you're scrolling through TikTok versus when you're scrolling through LinkedIn feed. And so content is going to perform better if you have a slightly different approach or different framing, different call to action, et cetera. And if you can get to a place with a prompt that really like nails that for you or gets close, and it's like, Oh my gosh, like I used to take so much time where I used not even repurpose with kind of that intelligence at all. And like now it's it's so easy. Like there are just a lot of those like smaller things that I I think um we use. And then a lot of businesses like I encourage them to find those for them themselves.

• 40:49Nathan Labenz?

  You do that in kind of co-pilot mode just via chat GPT?

• 40:54Speaker 00

  Yes, um, we also just like write scripts because it's easier. Uh, so I just have like a Python notebook that basically just sends that to chat GPT. Yeah, but I'll do it all uh uh all just directly to the model, yeah.

• 41:09Nathan Labenz?

  So what's the what's the rationale for that? Like I I'm personally interested in this because I honestly have not mastered it myself. I find myself much more uh inclined toward delegation mode versus co-pilot mode, interestingly. Except in coding where I I feel like the i- in that context, the autocomplete of like of the original GitHub co-pilot is amazing and I I definitely benefit from that. But so, yeah, tell me a little bit more about like, how are you doing this kind of repurposing? You start with what? And then you have these scripts and then like, are you reviewing? Or are you like trusted enough to like translate tweets to LinkedIn posts and go? I really want to understand the details of that.

• 41:47Speaker 00

  Yeah, so like one of the most common ones is like I'll take a TikTok transcript. So like something already posted on TikTok. Um, and then I have a a script which I've or like really it's a prompt that I've uh honed in over time that will take that and like create a LinkedIn post draft. Um, for example, and uh to your question, no, I don't just set it on uh autopilot of like, yeah, go forth and and say kind of like uh whatever's going to come out of chat GPT for me. And I really encourage like people don't do that. Um, but I've found it can get pretty close. And especially if you, I mean, this is where the co-pilot and delegation model breaks a little bit.

• 42:31Speaker 00

  But it's like, if you look at your prompt of like, how would I coach someone else to repurpose this content in the way that I think is going to perform best on this channel? Um, then you can kind of start to get to some of these prompts that work pretty well. Um, but yeah, I mean, that being said, like, I would say we're still at maybe.

• 42:50Speaker 00

  Anywhere between two to five percent of the like capability that we could be getting out of the technology that's just like literally readily available today, even though we're like full time trying things, experimenting. As I said, we try the last if it doesn't work, which I think is also great. Yeah, I mean, to me, I just like, I like, we're just so early and the adoption and like figuring out how this technology is like really going to help us and yeah, and how businesses work.

• 43:21Nathan Labenz?

  Two to five percent, that's not a lot. There's a long way to go.

• 43:24Speaker 00

  I mean, we don't have any agents going out and, you know, automatically pulling your LinkedIn and emailing Nathan saying, you know, hey, your stuff on Twitter is pretty cool. And let me tell you, you know, like, like, we don't have that stuff implemented, nor do I think that's like the place to start for a lot of people. But yeah, I mean, I, I think we're on the very, very early side, like as a society of like how much AI is going to change how we operate.

• 43:51Nathan Labenz?

  Do you use, as you're trying to like get it to write like you in different formats, how much of that is pure like descriptive prompt versus few shot based on like, you know, prior conversions that you've done?

• 44:03Speaker 00

  So we've tried both. What we have working right now is a combination, like I'll pull in similar LinkedIn posts for the certain type of content. So kind of like a, I mean, it's not fully automated yet, but like these are the things we think through, which is like, hey, like, if you could use embeddings to pull in a similar type of content, what would that look like? Right? And then how could you use that to do few shot learning to get the post to feel more like what you want? All of it comes down to like, do you like the end output or not? And did it save you time? And like if it didn't, then like try something new, right?

• 44:43Nathan Labenz?

  There seems like there's this, you know, maturing of all these tools that's going to take us from the sort of two to five percent to, you know, some significantly higher percentage, even if we don't, you know, get toward a hundred percent of potential. I wonder if you have a vision for like what the computing experience looks like?

• 45:03Nathan Labenz?

  As that matures, like, you know, my venture was kind of, we all said, using ChatGPT plus with plugins. But how would you begin to, it seems like it's not that far away. People are gonna build, they're building this stuff now. You know, we don't, like, all the, all the concepts seem to be there, the embeddings, the ability to like index your Gmail history, whatever. Like, conceptually, it's easy to say that. In practice, you know, I, I would not fork over my Gmail history to baby AGI just yet. So a lot of it seems like it's down to kind of engineering, but also like coalescing around certain paradigms or UIs that are just so nascent.

• 45:42Nathan Labenz?

  Do you have a sense for like what it's gonna be like to sit at a computer and do stuff in, you know, whatever, six, nine, 12 months?

• 45:49Speaker 00

  I mean, I feel like it could be pretty different than what we do today. I mean, a lot of me like hopes that it's different. Like I have this other hot take, which is like, I think as a society, we all like work way too much and do way too much of the same stuff every day. I mean, that's not like necessarily a hot take, but I'm like, I think it's like really not a bad thing if we are all working maybe like an actual 40 hours a week instead of what we like pretend is 40 hours.

• 46:19Speaker 00

  But in terms of like what that looks like, I mean, I don't know, like, I think I stand it like we're still so early and, and trying to predict what things are gonna be. I think is maybe like not the right mindset, the, or not to like, uh, challenge like your way of thinking too much, but, um, it's more like, how do you just like pay attention to like what is changing and what is happening? So that you can see like how certain trends are progressing. For example, one I'm paying attention to a lot right now is like, um, all the experiments of GPT four being able to code an entire application from someone's like voice command. Like that is huge, right?

• 46:59Speaker 00

  Um, and could have a profound impact on like us having highly personalized software or, uh, a business who's been, who has sticky notes next to their like. Laptop trying to say, okay, like my software doesn't do X, Y, and Z, but like if we do this, this, this, this, this, like hacky thing, it can do this. Okay. Well, if the cost of creating like more personalized software or your software in general is like dropping a lot. Okay. Maybe that stuff starts to like, you know, not be problems anymore. Um, yeah, maybe like long way of saying like I, I wouldn't plant a flag in the ground and say like, we're all going to have chat GPT plugin marketplace like on our phone and it's going to be the only mega app, right? But I do think things are going to change a lot.

• 47:40Nathan Labenz?

  And it seems like it's very clear that the agent paradigm will get you to the point where these sort of user interface tasks that are kind of tedious and time consuming can largely be delegated. Do you want to like come back and watch like sped up, you know, uh, videos of what it did and like find yourself like on the checkout page, like ready to, you know, hit confirm or whatever, almost like a rewind, except, you know, the AI is doing the thing and you're like rewinding the AI's activity instead of your own?

• 48:16Speaker 00

  Yeah. I mean, and then there's like the question of like, if everybody can post on Twitter and everybody has an AI that just like takes their raw thoughts and puts them into like high performing tweets and everybody is using these agents. Okay. Is anybody actually on Twitter anymore? Right? Um, or like, what's the value of being on like being on these platforms? If you have then a summarizing agent that pulls the most interesting tweets of the day, like, I think you can get a little bit too, and they're fun, they're fun thought experiments. Like you can get walk through step by step by step and saying, well, then this will happen, then this will happen, then this might happen. But yeah, that's then where it becomes really like, I think messy and hard to predict like this is going to be how this all plays out.

• 48:59Nathan Labenz?

  Yeah, totally. I, I mean, radical uncertainty is my kind of starting point for any, uh,

• 49:05Speaker 00

  That would've been a good name of the, of a podcast right now. Yeah.

• 49:09Nathan Labenz?

  The other one that, uh, we had considered was AI Summer, which, um, I think would have been a good one as well.

• 49:15Speaker 00

  Yeah.

• 49:16Nathan Labenz?

  The Bing launch. I don't know if you, uh, have a good story to tell there, but that must've been quite a bizarre experience in multiple ways.

• 49:25Speaker 00

  I feel really positive about companies that are like launching,

• 49:32Speaker 00

  you know, what they're building in this market. I think frankly, if you zoom out to some degree of a time horizon, the learnings you get from launching early, I think do like I fall on the side of like it does outweigh the potential risk of just like waiting way too long to like start putting this stuff out into the market. But on the launch itself, I mean, it was really, so I do have a kind of a funny story on this, which is that like, I was not an influencer before or a creator before, like that's like not my background. And so I was very new to a lot of this stuff. But I literally got a cold email from Microsoft just saying,

• 50:16Speaker 00

  like, we want to fly you out for this thing, but we can't tell you what it is. And at the time, like, you can't tell anybody that you even got this email. And I was like, and that was kind of one of my first experiences of getting something like that. I mean, things have happened since that have been, I realize that's more like the MO when you're an influencer, but like, at the time I was like, well, this is either gonna be like the the story, you know, ended pretty badly, right? Like flown out, you don't know where you're going. It says microsoft.com in the email address, but like, I don't know if that could be spoofed.

• 50:57Speaker 00

  But I was like, yeah, like, I thought I'll just go for it, right? It's kind of like my personality. And so then I said, like, yeah, sure, I'll come. And yeah, it ended up being really fun because it was, you know, I got to meet other, it was like a lot of the like AI reporters were there, so I got to make some friends.

• 51:15Nathan Labenz?

  I generally agree, and I think there has been a pretty strong demonstration of the wisdom of Making some contact with the broader public, right? Like, looking back on the timeline of GPT-4. And I didn't know this at the time, but I was in this like weird position where I was participating in the red team program and, you know, had access to the model and was like exploring that, you know, full-time obsessively, like writing all these reports for them, et cetera, et cetera. They did not give us, though, any other information. So I was like totally in the dark as to like what their plans were. And the version we were working with was the raw form that, you know, did not have any of the safety mitigations and was like, pretty crazy.

• 51:57Nathan Labenz?

  So I was genuinely nervous coming out of that period, like, what exactly are they gonna do next? And, you know, are we entering into like a period of total insanity? And then what they actually did shortly thereafter was launch ChatGPT with 3.5 instead of 4 and, you know, kind of allowed all those jailbreaks to happen. And, you know, I, nobody's ever like confirmed this to me, but it seems quite clear that that was a strategic plan to like, get the jailbreaks out, you know, largely on 3.5 and then, you know, be able to roll that whole data set into extended training. So I do broadly agree that like, it does not, you know, as Sam Altman kind of says,

• 52:40Nathan Labenz?

  like, it does not make sense to go develop a godlike AI, maybe not at all, but if you're going to, probably still doesn't make sense to like develop it all in secret and then just, you know, drop it on the world all in one, you know, zero to one moment. So I do agree with your perspective that like, it does, it is important to launch stuff, not only for business, but even just for like, society. At the same time, that felt rushed. And like, to me, like, kind of extremely so. I've got, I'm putting a piece together actually where I'm like documenting this, like going back in time and like, you know, kind of Twitter archaeology-ing, like, who knew what, when, you know, and how did this happen? But it seems like they definitely were rushing it and like kind of at an executive level,

• 53:24Nathan Labenz?

  I feel like they should do better. You know, you should not release your bot in a stage where it's like actively hostile to users. And, you know, some of those transcripts were like,

• 53:39Nathan Labenz?

  Just a simple disagreement about the date, right? We were not, this was not like jailbreak type phenomenon. I feel like that's, you know, when people try to break the model, I put that in a very different classroom, like when the model breaks on someone who was just acting earnestly and trying to interact with it. So it felt like super rushed to me. Did you, did you feel like they had things like buttoned up at the time? Or did you see a sort of like, because the big worry obviously is the race condition, right? For the, for the people that worry about AI safety long-term, I'm definitely one of them, as enthusiastic as I am. You know, the work, the thing that people worry about most is like, if we get into a race condition, everybody's just cutting corners or whatever, like that's really bad. It, to me, it feels like that's kind of what happened there.

• 54:23Nathan Labenz?

  But I wonder what your perspective is, if that like resonates or feels off, you know, given that you flew out and sat in person with them when they did it?

• 54:31Speaker 00

  Yeah, I just go back to like, I am definitely a proponent of like, you need, you need feedback to learn stuff. I also, I feel like working on, like, I feel like we're so early and a lot of this like technology, what it even means to like red team these products, I would wager we know way more now than we knew three, four months ago on some of these teams. Yeah, I completely agree. Like you should not have a bot that is aggressive to users or OpenAI has a moderation endpoint. Like that should be a given, especially if you're launching something high profile.

• 55:11Speaker 00

  And like, I know there were some launches recently that didn't use that endpoint. Like that stuff is really important. That being said, I think that there's a little bit of like, you learn from your failures too. And I think that sometimes some of these, like I mean, you zoom out, they're not that big of issues and we learned a heck of a lot and like, you know, people are going to be way more cautious in a lot of these, avoiding a lot of these like problems in the future.

• 55:42Nathan Labenz?

  Yeah, largely I agree. I would like to see them. Publicly own it a little bit more and like kind of publicly learn a lesson. That may be, you know, wishful thinking on my part, but it is striking to me that we're entering this era where it's like, these technologies are not super easy to control. You know, one of my biggest, just two by, one of my investors used to say, two by four to the head, obvious. And definitely a two by four to the head, obvious point that I've, you know, deeply understood through the red teaming and everything else is just the good behavior of this class of large language model definitely does not come, you know, by default.

• 56:25Nathan Labenz?

  It is not like easy to create. They have like a hell of a time, you know, with false positives and false negatives on refusal. And, you know, it's just very far from being a solved problem. And so, yeah, it's not cra, it's not, you know, somebody was going to do something like Sydney at some point and make that mistake. It doesn't sit great with me that it was like the biggest, you know, one of the very biggest companies in the world, who I would think should have probably known a little bit better. And it definitely still doesn't sit that well with me that like they haven't kind of come back around and been like, yeah, we kind of fucked up, you know? We rushed this thing out and like, what's the standard going forward? You know, should I expect this type of behavior from Word in the near future?

• 57:07Nathan Labenz?

  Or like, is Excel going to be like, you know, intruding into my marriage? I mean, I, I just think that there's, somebody should probably like, at least articulate a standard that they, you know, plan to uphold for themselves.

• 57:20Speaker 00

  I should also say too, like, you know, I, I went to that event. I've worked with Google on some stuff. I've worked with Adobe, like I, I'm working like across the space, which I think is another like fun perspective I get to have, right? Which is just being able to see a lot of the stuff, you know, OpenAI that's going on. Like, I love this because I'm like, the, the challenge that I would throw back to you is like, Okay, yeah, but like, imagine if these types of things were hap- like, were happening with much more powerful models. Does having bad, uh, or like, not great experiences today, like, help us decre- like, decrease the risk of that happening with other models in the, in the future?

• 58:02Nathan Labenz?

  I think it, I think it does. I think we're actually, I'm very pleased with how things have gone relative to how, you know, I feared that they might. And, you know, it seems like there is kind of a, I mean, just today, right? Last night at MIT, Sam Altman kind of said, and now there's been this like wave of Twitter messaging from OpenAI employees. It's like, we're not training GPT-5. I think that's a very key part of the kind of plan that you're, you know, describing, that we do need some time to actually work these things out. So, yeah, I agree. I, I, I view it very differently, I think, in a world where GPT-5 is currently training versus what they're saying now, which is that it's not. I think that makes a huge difference to how you kind of understand the like, the early release decision-making.

• 58:48Nathan Labenz?

  I'm grateful that there is like some, you know, restraint, uh, at the top of this trend because, you know, it is so powerful that, uh, you know, I do kind of, it, it's not hard for me to imagine that it could really get entirely out of control.

• 59:04Speaker 00

  Think about, um, the mass adoption of this technology in different companies, right? If a mid-sized company has some bad prompt injection, uh, exploits happen, that doesn't educate the market, potentially as well as a bigger company, uh, having things like, I don't buy, uh, that Snapchat's My AI is the only, uh, chat GPT bot that has talked about stuff that told the bot not to talk about, right? Uh, but it is the one that's like highest profile. And so I think that does a good job of also like educating the market of like,

• 59:45Speaker 00

  hey, anybody who's building anything that looks like this now knows of that story and is, uh, protecting against it.

• 59:53Nathan Labenz?

  What, are you hearing things from people about like the alpaca, you know, llama moment? Um, and what do you advise people? When they ask you like, hey, this llama thing's out there and I heard it's just as good as ChatGPT, like, what's your response to that kind of thinking?

• 1:00:13Speaker 00

  I mean, I think one thing is, I always remind people, I'm not a lawyer and you need to talk to a lawyer about like the commercial use or just general like use of these different models in your business, not just like limited to llama and alpaca, like I tell them that for everything. But I mean, I think if you like, look at the writing on the wall, we're gonna have powerful open source models as well in this large language model space. And so, yeah, like, there's, I think there's just an element of people getting comfortable with the technology. And yeah, I guess like, I'm more afraid of mistakes made later than us learning now.

• 1:00:57Speaker 00

  It's just my like, general take. Yeah. But I would not advise somebody to go implement something that's not commercially licensed into their core product.

• 1:01:07Nathan Labenz?

  You know, let's imagine that they were like, okay, cool, it's released. But it's still just this sort of pre-trained, totally unrefined, you know, and therefore more unwieldy, more, you know, whatever, alien-like thing. I don't have a mature framework for thinking about this yet, but it does seem like the, one of the things we talked about, you know, much earlier in the conversation around internal like task automation versus a sort of element of your service is probably a really critical distinction there, because if you did have a llama and, you know, then it was commercially licensed and all that, and you then like, applied your, you know, 10,000 example instruction fine tuning, and then you are like in control

• 1:01:52Nathan Labenz?

  of how it's being used in the context of some task, I would expect you'd probably get, you know, could get pretty good results and might be happy that you did it. Whereas if you try to like, take a shortcut to cheap ChatGPT, and then like, put a chatbot on your website that's based on llama, I would expect that you would be quickly embarrassed.

• 1:02:15Speaker 00

  Definitely. I mean, yeah, like the risk profile of launching something poorly thought out externally is like infinitely higher than.

• 1:02:27Speaker 00

  Like experimenting internally. Um, I mean, a lot of teams who go back to like, I talked to a lot of teams who did not have AI in their 2023 roadmap. When they asked me like, where do I start? I really try to encourage them to think about the internal use cases because like the risk is just so much lower. Um, if you don't have that experience already, like you're gonna get your feet wet. Uh, and you're gonna learn really quickly that uh, like how these models work and what their limitations are. Right? Um, I think a good example would be like if somebody had launched a uh, chat bot on their website that goes in like, um, can summarize webpages and it was 3.5.

• 1:03:09Speaker 00

  Um, they might think that it actually does that, but then, and they might offer that service to their customers. But if they instead had started with like, okay, well, let's use this internally first. We're using GPT 3.5 to summarize webpages. They would, they would eventually figure out or learn or just uh, like have the experience to say, oh, this actually isn't doing that, right? Um, and that's just a very, I don't think most people are confused on that point, but like, um, this is like one example of like, when you're using something internally, uh, you do have more of a tight feedback loop of like, how's this actually working?

• 1:03:44Nathan Labenz?

  Yeah, it's critical. Education, it all starts with education. Uh, which is, I think, uh, why you're in such an interesting spot right now because you've been uh, doing a great job at educating so many people. Couple of quick closing questions for you. Just cool tools that you would recommend to the audience that are not like the obvious chat GPT.

• 1:04:04Speaker 00

  The number one thing I recommend to people is start using chat GPT with an automation tool like Zapier or Make. A, it's fun and B, I think that tends to be the big unlock for a lot of people and actually starting to use stuff. I'm gonna give a random shout out, which is supermeme.ai because we are using AI for like so many things that are like so serious, so like business productivity oriented. And honestly, sometimes it's just really fun to use AI to make much better memes. And like, I've never been a person who's good at making memes before. And now I feel like I can make ones that make me laugh at least. So I would uh, definitely go check that out.

• 1:04:42Nathan Labenz?

  Yeah, that's cool. We made the art for this show with Playground AI and.

• 1:04:49Nathan Labenz?

  I mean, that's would have been completely impossible for me to do that previously. So I do love those kinds of things that just unlock a totally new, just like, you know, capability for me as an individual. Okay, second one. You kind of alluded to this earlier as well, but hypothetical scenario, a million people have a Neuralink implant. Whenever that, however long that takes, we're at a million implants. If you get one, you can control your device straight from your thoughts. So you could like transmit, you know, thoughts to text or thoughts to like UI control and be hands-free using a computer.

• 1:05:32Nathan Labenz?

  Would that be enough to have you interested in potentially getting one?

• 1:05:36Speaker 00

  Oh, I like, put me on the shortlist for the early experiments. No, I, my husband is a, is a doctor, so I, he would read all, I trust him to read all of the stuff that I would need to think through to make that decision, but then assuming it's a, it's a go, put me on the list.

• 1:05:55Nathan Labenz?

  You're in good company with that answer. That's a pretty polarizing question, we've found, more so than even the tools, where people are more often just like, I don't know, I pretty much just use ChatGPT.

• 1:06:06Speaker 00

  If I can, what's your answer? I'm just curious.

• 1:06:08Nathan Labenz?

  Oh, I would definitely be open to it. I mean, I think this is, it's a question that's, I think it's revealing about how people think, and I like asking it for that reason. I also think, of course, it's like a way too specific hypothetical, and, you know, the actual answer might be like, you know, by the time that is mature, I already can kind of do that with my like, apple glasses or whatever, just by like twitching my eyes, and that's like comfortable enough that, you know, nobody needs to drill in the skull. So I'm not that confident that, you know, I'll end up wanting to do that when it becomes available, but I would be definitely open to it in a, you know, in a narrow hypothetical where it's like, the world is like it is today, but you could have that. I have three kids and they have my hands full pretty often, so it's definitely, there's

• 1:06:50Nathan Labenz?

  a lot of moments where I'm like, you know, taking them on a walk through the neighborhood or whatever, and I'm like, I wish I could just jot this thought down before it totally escapes me. So honestly, for that alone, I would be, and I might, you know, could also be a wearable, perhaps in the future, could be a little less radical, but. Even with the implant, I would be open to it.

• 1:07:07Speaker 00

  I'm like, I already had an advantage. I already wear glasses. So like, I could just throw it on the glasses, right?

• 1:07:14Nathan Labenz?

  If you listen to Robert Scoble, that's coming, the first version that's coming this year, I don't know. There's, there's a couple of these like stealthy device, you know, AI things that I really don't know much about, but could make a huge splash in the potentially like not too distant future. Uh, okay, last one, just zooming out, big picture, um, as wide as you kind of can. What are your biggest hopes for and fears for society? As this AI wave kind of washes over everything?

• 1:07:48Speaker 00

  My biggest hope is that we get to a place where we're frankly, working less. And I put an asterisk on that a little bit, because I think what's more exciting to me is that we get to a world where work looks more like the type of work a lot of people do, kind of on weekends or on vacations. That's casual, you're thinking about things when you feel inspired to think about things. And you're able to solve like the bigger problems at hand instead of, kind of, at least for me, like the work week is just so busy and full of so many things all the time.

• 1:08:26Speaker 00

  Um, and, uh, yeah, I think, I think it sounds like pretty great to feel like you're on vacation all the time. Uh, but still able to like, can I contribute and drive what you care about forward? Um, biggest fear, I think about this a lot, which is, um, a major like motivating factor for how I spend my time. Is I think that we want to have a future with diverse businesses, diverse perspectives. And if, uh, the way that things play out, only 1% of people are actually adopting this technology. I think that's a worse future than us having broader adoption, um, and continuing to have like competition.

• 1:09:09Speaker 00

  Uh, and so like what I'm more afraid of is Yeah, centralized adoption as opposed to democratized access.

• 1:09:18Nathan Labenz?

  Cool. Well, you are doing your part to try to avoid that future. And I really appreciate you taking the time to join us today. Rachel Woods, thank you for being part of the cognitive revolution.

• 1:09:29Speaker 00

  Thanks for having me. This was fun.