The Computer Expert That Just Solved AI’s TOUGHEST Challenge - Rose Yu - #511

Please join my mailing list here 👉 https://briankeating.com/list to win a meteorite 💥 Could an AI physicist soon out-innovate Einstein? In this exciting interdisciplinary exploration, UC San Diego’s Rose Yu dismantles the romantic myth of genius-driven science and instead offers a thrilling look at how AI could become not just a computational tool, but a genuine partner in discovery. Rose draws from her pioneering work in traffic forecasting, pandemic modeling, and high-energy physics to show how machine learning is evolving and what it can offer to the scientific community. She walks us through the fascinating ways AI is already helping us rethink what’s possible in science: from generating novel hypotheses to rediscovering fundamental symmetries, and even beating traditional simulations in both speed and precision. With her characteristic clarity, Rose explains how she’s built data-driven surrogates for complex phenomena like the spread of epidemics and the turbulence of fluids. We also tackle some of the more existential questions: Can an AI ever experience “happy thoughts”? How close are we to AGI? And should we be worried about AI replacing us? If science is humanity’s greatest invention, this episode is a fascinating look at its future. — Key Takeaways: 00:00 Intro 01:04 Can an AI physicist out-innovate Einstein? 02:07 Why are GPUs so good for AI? 12:24 Traffic modeling and AI 16:08 Is AI just imitating physics? 24:53 Epidemiology modeling and AI 30:49 Should we be worried about AGI? 34:32 The benefits and dangers of AI 39:53 AI scientists and scientific productivity 42:19 The impact of AI on academia and education 52:34 Outro — Additional resources: 💻 Rose Yu’s website: https://roseyu.com/ — ➡️ Follow me on your fav platforms: ✖️ Twitter: https://twitter.com/DrBrianKeating 🔔 YouTube: https://www.youtube.com/DrBrianKeating?sub_confirmation=1 📝 Join my mailing list: https://briankeating.com/list ✍️ Check out my blog: https://briankeating.com/cosmic-musings/ 🎙️ Follow my podcast: https://briankeating.com/podcast — Into the Impossible with Brian Keating is a podcast dedicated to all those who want to explore the universe within and beyond the known. Make sure to follow/subscribe so you never miss an episode! Learn more about your ad choices. Visit megaphone.fm/adchoices

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

• 0:01Speaker 00

  Life is all about making decisions. Sometimes you make questionable ones, like buying a new TV that's too big for the wall because the electronics store employee was a little too good at their job. Then sometimes you make great ones, like protecting your home and auto with Progressive. You get customized coverage options so you can be sure you have the right coverage. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers. Not available in all states.

• 0:31Speaker 01

  This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required. Compatibility and availability varies 18 plus.

• 0:56Speaker 02

  This professor says AI will never feel shivers down its spine. AI systems can now automatically discover fundamental symmetries and new physical laws without being taught the underlying theories. Essentially, they're finding the hidden mathematical patterns that govern our universe through pure data analysis. If machines could independently discover new laws like Lorentz invariance from particle physics data without knowing Einstein's theories, it suggests AI might find entirely new physical principles we've never, ever conceived of. Professor Rose Yu's team trained deep learning models on data from the Large Hadron Collider that automatically recognize symmetry patterns and high-energy particle interactions. The same symmetries that took Einstein and other geniuses decades to understand through pure theoretical insight.

• 1:40Speaker 02

  Professor Rose Yu is a computational physicist at UC San Diego, whose AI models have been deployed by Google Maps for traffic predictions and ranked number one among 40 national teams for pandemic forecasting during COVID-19. Now, let's meet this brilliant natural genius who's taking artificial intelligence to the next level. Let's go.

• 2:00Brian Keating?

  Professor Rose Yu, so nice to have you here at UCSD's Arthur C. Clarke Center for Freeman Adoptions into the Impossible Podcast. Great to meet you.

• 2:07Speaker 04

  Thank you. It's a pleasure to be here.

• 2:09Brian Keating?

  You've done so much wonderful stuff and it's wonderful and gratifying to know that you're a colleague here at UCSD, but you're also getting the recognition that you deserve. I want to start off with a little bit of a provocative question, which is fascinating to me, which is the following question: Can an AI physicist ever do what Albert Einstein did? So famously in 1907, Albert Einstein said he had a dream, a thought experiment, that if he was in free fall, like if he was like this and he fell down, that he would feel no gravitational force field. So you could actually get rid of gravity and that's part of what's called the Einstein equivalence principle. I want to ask you, he called that notion, that thought, the happiest thought that he ever had. He said it made shivers

• 2:51Brian Keating?

  go down his spine. Can an artificial intelligence ever have a happy thought? Can it ever feel shivers down its spine or its CPU or GPU? And can it really create new laws of physics?

• 3:02Speaker 04

  That's a great question. So I think the, in general, AI is a bunch of machines that can think. So if you ask whether these machines can have emotions, I'll say most likely no, just by how they're constructed. I don't know whether it's possible even to build machines that have emotions. However, the second question is like whether these machines can create. And I think my answer is definitely yes. Because these models are built with a lot of data and they're distilled with the world's knowledge. And then once they are imbued with this knowledge, they can start to create new content. And you're already

• 3:45Speaker 04

  seeing these examples now with large language models, they create new mathematics theorems, they can also generate new hypotheses. Recently, they have shown these models can create new molecules. So as we are seeing more and more examples, it just vindicate that it's definitely possible for these models to create new theories, like what Einstein has come up with.

• 4:09Brian Keating?

  Obviously, everyone out there is familiar with LLMs like ChatGPT, Claude. I use them all. I pay for them all. There's probably a bunch of paying for that I don't even know about. I'm just, my credit card. Gets deducted. I use them for everything. They read my kids bedtime story. Originally, I was kind of embarrassed and I was ashamed. Ah, I'm using AI to read a bedtime story or come up with a bedtime story and then it reads to them. But I mean, a book is kind of like that. It's just a natural intelligence and I'm just reading or listening to an audiobook. So I've gotten over my guilt on that. But there seems to be something different about physics and that's the only reason I'll bring that up is because that's what I do. I'm a physicist. And that it's an empirical science. Mathematics, I could understand and I might have predicted that you could make new mathematical theorems and test things because

• 4:52Brian Keating?

  proof is possible in mathematics, but it's not possible in physics. I can't prove the earth is round, you know, for example, it's not possible to prove that statement. You can exclude and falsify other statements. When I think about L L Ms, I think about, you know, these original devices that were made on GPU systems. So we could talk about the computer architecture later on. But as I understand it, you know, the GPUs were invented to, you know, make Grand Theft Auto six and Minecraft and run really fast and optimize for, you know, my, my kids to, you know, kill their friends at, you know, one millisecond before their, their, their friends kill them on the game. They weren't designed for, certainly for AI. They just turned out that they were good for it. Why are GPUs so good for it? Uh, good for AI? And are they good for non L L M type AI systems?

• 5:37Speaker 04

  So the GPUs are particularly good because they allocate a lot of those resources to arithmetic computing, right? Because in, in general, when we talk about computer chips, they're mainly designed for two types of tasks. One that is responsible for arithmetic computing, like doing matrix multiplication. And the other one is handling all this logistical instruction processing and then handling the scheduling of different jobs. So for AI, I think when I talk about AI, it's more narrowly referred to uh, the set of algorithms called deep learning, right? And these deep learning algorithms are essentially a lot of matrix multiplication and matrix computation.

• 6:16Speaker 04

  So by leveraging GPUs, that was specifically designed to put a lot more weight on processing matrix multiplication operations. It's much more efficient, usually like 10 times, if not 100 times more efficient. Of course, like these type of algorithms are not only limited to applications in L L M's. Uh, in my class in deep learning, we talk a lot about other applications in deep learning. For example, generated videos with diffusion models, generate trajectories for autonomous driving cars. Or, you know, generated music even. And sometimes we in our, you know, own research work, we use it to generate new symmetries of the universe and new molecules in chemistry. So yeah, they're all based on the algorithm of deep learning.

• 6:59Brian Keating?

  At some level, the G P U structure is naturally useful for doing linear algebra, matrix multiplication and and and doing all sorts of inversions and so forth. At extremely high speed and that's because, you know, computer screens originally are, you know, are just two by two matrices in some level with different third dimension representations, brightness, color, saturation. But physics is different, right? Physics is continuous. So my example that I keep going back to is Einstein in nineteen oh seven knew that the planet Mercury had this weird behavior that it was orbiting the sun in an ellipse, which is normal. But the place at closest approach to the ellipse, the line connecting the what's called the perihelion to the sun

• 7:39Brian Keating?

  was moving a little bit each year, which doesn't happen for any of the other planets. It only happens in a strong gravitational field. And he knew this and Newton even knew this or people shortly after Newton knew this. And the only way to correctly account for that is to include the fact that gravity affects time as well as curved space. So if you just put a bowling ball on a tarp or, you know, some trampoline, it'll bend it, right? But you won't get the the procession of the perihelion of the planet or the orbit of the marble around the bowling ball. Unless you include the curvature of time. Einstein realized that and made spacetime this four dimensional thing. When I tried to do this with my student Evan Mont uh over the summer or last summer and your student helped out, I forget your student's name who helped. It's helped a very, very great graciously of you.

• 8:23Brian Keating?

  Um, but we couldn't get it to work. We had to put in artificial, now maybe you could get it to work, you know, with your more advanced knowledge and understanding than I have. But we had to put in like putting charge on Mercury and and put in all these kludges because we fundamentally had to discretize spacetime. To Quanta pick boxles. We couldn't do that without, we couldn't solve that without doing so. Have you seen or encountered an example where an AI will create, you know, kind of that new leap that you see? Well, no, no, you could discretize it in time all you want or in space all you want, but until you include time curvature, nothing happens. Are there truly new things that you couldn't understand? Not beating humans in chess or folding a protein faster than we can understand. Are there new like creations like four dimensional, five dimensional?

• 9:07Brian Keating?

  You know, is it going to come up with string theory? How can it with GPU limitation? Do something that's just not just faster, but truly novel? Can it, can it really do something?

• 9:17Speaker 04

  I think we need to like kind of a, you know, disentangle between like the underlying, you know, algorithm or underlying hardware infrastructure with the type of models that are basically used for creating new things. So there are different kind of concepts because we have GPUs, right? That was originally designed for video gaming. And these are sort of hardware that are particularly good with, you know, arithmetic heavy operations in computation. And then you use GPUs, you can say, okay, I will use that to design a set of algorithms, the models, right? Models in, in particularly generative AI that can generate new content. And that's what we call creation, right?

• 9:59Speaker 04

  So in some of our work, you know, we have shown you can build generative models that can generate new type of hypotheses in science, especially in physical science. So one of the key example we showed in, in our work is I just by looking at, you know, a high energy particle physics data from large hydrogen collider. The model can automatically recognize there is Lorentz symmetry from data, without knowing the knowledge of, you know, general relativity from Einstein. So, so that's something that the model can do, right? And it doesn't necessarily require discretization of space and time. And then I, I would say in general, the scientific discovery goes through this kind of iterative loop

• 10:43Speaker 04

  where you have some observation and then that's your data set, right? And once you have the observation, you will. Create a new hypotheses and then a good scientist will try to, you know, shrink down this hypothesis to a few of these possibilities. And then like the example you you mentioned and trying to find another dimension of time and to better fit the data.

• 11:04Brian Keating?

  Mm hmm.

• 11:04Speaker 04

  And then, once you have a better fitted hypotheses, you can verify them either through theory or through experimentation.

• 11:11Brian Keating?

  Mm hmm.

• 11:11Speaker 04

  And then using this kind of new observation of experimentation, you can refine your hypothesis. And then, you know, go back to this iterative loop. And I think it's throughout this iterative process, like AI can play a role in every single step.

• 11:24Brian Keating?

  Mm.

• 11:25Speaker 04

  Because AI is fundamentally designed to process huge amount of data, generate a large number of hypotheses. It can help scientists sift through like all these possibilities really quickly, given this massive power to process, you know, internet knowledge.

• 11:40Brian Keating?

  Mm hmm.

• 11:40Speaker 04

  Right? And then you can also create design experiments to to help verify them. I would say the current limitation is that a scientist can go to the lab and then do the experiment manually, right? If they sometimes they have help with the robots.

• 11:54Brian Keating?

  Mm hmm.

• 11:54Speaker 04

  Um, but AI algorithms are largely still limited to like computers. And if they wanted to go out and verify the hypothesis they have, they still have to rely on humans to do this experimentation and collect the data.

• 12:08Brian Keating?

  Mm.

• 12:09Speaker 04

  But a lot of these processes in scientific discovery are already being accelerated by AI.

• 12:14Brian Keating?

  Perhaps, yeah, the understanding of the fundamental symmetries and underlie physical law, which Einstein and Emmy Noether and many other people, you know, contributed to have to do with conservation laws that underlie physical principles, including independence of position on time and so forth. That's really exciting. I wasn't really with that, so I'd love to to learn more about the Lorentz invariance violation from CERN. But on a more practical level, people talk about, well, what is AI good for, you know, making new chatbots or whatever? And everyone's played around with that or making some, you know, uh, logos or, you know, music. My my daughter's made a song about herself and her her friends and what they do and stuff, and that was kind of cute. And it's brilliant and they they sound, you know, they're they're listenable, they're not horrible. And now with Google VO three, that came out very recently.

• 12:59Brian Keating?

  You know, people are having fun and getting scared about what it will mean for. But I feel good because the uh this format where you and I are talking and we're not AIs, you know, we can we can uh vouch for each other at least. But if a normal like talking head video or you don't or you don't see the person, these are just computer generated podcasts. Those are there's many of those like Notebook LM came out with this audio format thing. Uh but one of the more most practical implications, applications of your research had to do with traffic modeling. And this you came up when you were a postdoc at Caltech.

• 13:28Speaker 04

  Yes.

• 13:28Brian Keating?

  A little known uh technical college uh north of San Diego. And and that's because, you know, everyone wants to get out of LA and come down to San Diego. Talk about the traffic uh modeling and the application of AI that you invented as a postdoc. That were later, as I understand, taken over by Google and now used in part of Google Maps or will be. Yeah, so talk about the traffic modeling.

• 13:47Speaker 04

  Uh sure, so the idea was also, you know, uh there's a lot of data collected by, you know, road network sensors, right? These sensors are recording the traffic volume and the speed at very high frequency. So before, you know, this data were just like kind of uh looked at like with simple models and people would use very simplified assumptions to to try to capture the pattern in the traffic. And now with the deep learning rising, I think around twenty twelve, right? And then we saw like, okay, what about we look at deep learning models for traffic forecasting? So for traffic forecasting in general is not a typical, you know, image data. Because the sensors are distributed as a Euclidean, uh non-Euclidean space, right?

• 14:30Speaker 04

  On the graph and then you cannot use any technique that were designed for videos or images directly apply them to traffic forecasting. So we have to come up with our own, you know, deep learning model. In particular, we were inspired by this uh paper from a physicist and and he described how traffic flows related to fluid flows. You know, kind of lattice and he described the Navistos equation, based on Navistos equation to predict the traffic volume. But that was also based on this kind of assumptions that you have. But this connection he pointed out in the paper kind of inspired us. And then we actually designed a deep network called diffusion convolutional.

• 15:09Speaker 04

  Neural networks, also specifically we put this concept of fluid diffusion and to model traffic flows on sensor network. And that model was like very successful because before deep learning, people were only able to forecast accurately for about ten to fifteen minutes. Then after deep learning was taking over and we were able to train these more complex models with very little assumptions, right, with a large amount of real time sensing data. So then we were able to accurately forecast the road network traffic for up to one hour. And that's why it was a big deal and it was deployed by Google Maps.

• 15:46Brian Keating?

  Uh huh. So does it take into account statistics like on average, you know, ten people per, you know, hour will be using their cellphone to text and they'll crash in. Or is it actually not using statistical averages or past behavior in an ensemble, you know, of past histories? And just statistically forecasting traffic. It's actually knows what the traffic will be more or less. I mean, there's some variables like a meteor could hit L A or something. But what levels of actual simulation versus prediction are occurring with these types of convolutional neural?

• 16:18Speaker 04

  Right. So this type of, you know, diffusion convolutional networks takes in historical traffic patterns. Like for example, the the velocity, the speed, the volume of the traffic for different locations. And we also know have data about, you know, latitude, longitude information, the traffic accident information from police reports. Right. All this data are fed into the model. And then the model implicitly learn the statistics from the traffic pattern trying to project how the traffic pattern is going to change in the future. So because it's generating the future traffic pattern, so you can also think of it as a simulator, right, for future traffic. But this not like a simulator based on differential equations that people have done before. You rather it's a simulator trained with data.

• 17:00Speaker 04

  So sometimes I call this type of model data driven simulator.

• 17:04Brian Keating?

  Data driven, some answers like. So when we see things in, you know, the physics space that, you know, the now some models probably you were involved with can can accurately simulate what, you know, smoke looks like or water in a turbulent, you know, medium. Are they actually simulating the individual particle trajectories? Or are they just like representative, like, oh, a smoke will rise vertically and it'll have this kind of density? Or is it actually solving physics? Or is it really just imitating what physics looks like? And you could do it with a good artist.

• 17:35Speaker 04

  Yeah, that's a very good question. So actually, when you try to simulate a fluid, right? There are two types of representation for fluids. One is called the Eulerian representation, where you basically just like sit at a particular location and watch how the flow is going to change. So that you don't need particle based representation. The other one is based on the Lagrangian formulation, where you actually think of think of fluid as a collection of particles. And you trace the movement of individual particles and integrate them, right? That give you the overall movement of the fluid flow. So actually, for both representations of fluid, there has been AI models that were designed to forecast or to like simulate the movement. And I have seen success for both cases.

• 18:16Speaker 04

  In terms of the whether they are actually solving the equation, I think most of the time they are not solving equation in the traditional sense, because in when you solve an equation with, you know, numerical methods, like finite element or finite difference. And that's what kind of consider the traditional solving the equation. But but these models, they are based on deep networks, right? So the fundamental algorithm is back propagation. They don't solve the equation in terms of finite difference or finite element, but they give you a solution, which evolves over space and time. So using that solution, you can predict the future evolution of the fluid. And then it's there much faster than traditional numerical methods.

• 18:55Brian Keating?

  What are the biggest limitations for what you do? What what gives both the largest systematic errors? And the largest bottleneck maybe to what you do is a GPU count? And you just need bigger computers, faster computers? Or is it some other rate limiting process that that restricts how much progress you can make in a given application at a given moment?

• 19:15Speaker 04

  Right now, we have a project called the multimodal.

• 19:18Speaker 00

  Life is all about making decisions. Sometimes you make questionable ones. Like buying a new TV that's too big for the wall because the electronics store employee was a little too good at their job. Then, sometimes you make great ones. Like protecting your home and auto with Progressive. You get customized coverage options, so you can be sure you have the right coverage. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers, not available in all states.

• 19:47Speaker 04

  Foundation model for automatic hypothesis generation.

• 19:51Brian Keating?

  What's the acronym for that?

• 19:53Speaker 04

  But we have an acronym called GNI, Generative Hypothesis Model. But I forgot what the rest of the letters represent. But it's a GNI. And then what we hope it can do is can generate hypotheses like a scientist. And then we wanted to synthesize multimodal information from both the textbooks, research articles, and large-scale numerical simulations in climate. One of the bottlenecks that we're facing is that a lot of the existing foundation models like ChatGPT or Cloud, they just fundamentally cannot handle large-scale, high-dimensional simulations. So we have to come up with our own model. But then, again, we were also limited by the data, right? So generating this type of data to pre-train a foundation model is extremely expensive.

• 20:36Speaker 04

  And then there, you know, we weren't able to train really big models like 70B models. We were only able to like fine-tune a 7 billion model. With 7 billion models, still, we need a lot of data to train. And then this data have to be curated in a way that makes sense, right? Because we want to mesh the textual information from research articles and textbooks with the simulation data.

• 20:59Brian Keating?

  Is that a thing called supervised learning?

• 21:00Speaker 04

  So like pre-training, like, you know, self-supervised pre-training. But for self-supervised pre-training, you need to like carefully curate the dataset. Otherwise, you're just adding noise to the model, right? And then the other challenge is how do you even define the evaluation metric? Because in, in general, when you train like an LLM, right? People would use, you know, actually users as a labeler and what they call the reinforcement learning with human feedback. And based on human feedback, they iteratively improve the model. But then for Our model because we have to rely on our domain scientists, in this case, a climate scientist.

• 21:33Brian Keating?

  Yeah.

• 21:33Speaker 04

  We have one climate scientist in a team and his student.

• 21:36Brian Keating?

  Right.

• 21:36Speaker 04

  It's definitely not very scalable if we just rely on our user to give our feedback.

• 21:40Brian Keating?

  Right.

• 21:41Speaker 04

  So we need to come up with a much more scalable way to collect the supervision and to evaluate the model, right? To tell us whether the model is generating the right hypothesis or the wrong hypothesis.

• 21:50Brian Keating?

  Mm hmm. So the types of data that you're talking about here, are they time ordered, you know, time series data? Are they weather, you know, images from satellites? Obviously, you talked about traffic data. Let's go back to the traffic because that's easier for me to understand. It's sort of one dimensional in some ways, but it's not Euclidean in other ways. So what would be the kind of rate limit or the bottleneck there? Is it that, you know, some of the data, as you say, is noisy? By the way, I found out the root of the word in Latin for noisy or noise comes from nausea, like to be sick. Which is kind of interesting because I'm sure you get sick when you see bad data, uh, noisy data. Talk about, what does the process look like? Is it some person in Nigeria looking at, you know, LA traffic on Interstate five one day? And then someone in, in, uh, you know, Bosnia looks at whatever.

• 22:34Brian Keating?

  What does the supervision look like? What are the, what's the input to the model? You said it's police and this and that, but there has to be some human element that curates, if it's human reinforced, non self-supervised, I guess, self-supervised.

• 22:46Speaker 04

  Oh, so this, actually, this problem, this project was, I think, uh, it was 2018. So it's like seven years ago. And at the time, I think it was still before the LLM era, right?

• 22:56Brian Keating?

  Yeah, right.

• 22:57Speaker 04

  And, and then the data that we used was this like, uh, loop sensor that was already installed in like every highway in California. And there is a particular, uh, system that was built many years ago, trying to just collect all the.

• 23:11Brian Keating?

  It was time series data, mm hmm.

• 23:13Speaker 04

  The time series data.

• 23:14Brian Keating?

  Velocity.

• 23:14Speaker 04

  Yeah, it's a velocity volume and that it gets sent over the internet, right? To, uh, to some central traffic bureau. And then, you know, the system collects the data as you can download them, anybody can download them.

• 23:27Brian Keating?

  And then how do you, how do you convert that condition into a trained data set for your purposes?

• 23:31Speaker 04

  Oh, yeah. So, so for ours, you know, once our task, we define the task as a forecasting, right? Which your input is, you know, historical. Features of the traffic, it could be like, you know, historical uh volume, speed and potential like uh accident patterns. And your your your target, your prediction goal is the future.

• 23:50Brian Keating?

  Yeah.

• 23:51Speaker 04

  Uh traffic. And in our case, we were looking for like fifteen minutes as an interval, but predicting rollout for one hour.

• 23:57Brian Keating?

  Mm hmm.

• 23:58Speaker 04

  Right? So so that's once we define the prediction task and we can essentially extract subsequences from the time series to curate the data set.

• 24:05Brian Keating?

  Are you still involved in that project or is that that's you did as opposed to like, but are you still involved at all?

• 24:09Speaker 04

  Oh, not anymore. Yeah, because it was actually I spent like a year and a half at Google and trying to look at scale up with this project and it was done.

• 24:17Brian Keating?

  So when we look at, you know, the apps now, I can put in the app and say, you know, how close can I cut it to, you know, go and meet my wife somewhere? And and it'll say like, oh, well, if you leave now, it'll be this much time to get there. But then it actually allows me to put in, well, no, I have to be there at five thirty. So when should I leave? It's a different question. Is that? Is the type of research that you applied, is that now being used in these apps and and actually?

• 24:39Speaker 04

  Um, because I I wasn't aware about the features they developed after our forecasting model, right? Because our model was a forecasting model. It tell you, you know, how the traffic is going to evolve. Of course, like the situation you mentioned, use cases of like, you know, ETAs or, you know, planning, and they're all depend on forecasters.

• 24:57Brian Keating?

  Yes.

• 24:58Speaker 04

  So so I'm sure that Google and other companies have built on top of this technology, but I I was not involved.

• 25:04Brian Keating?

  What are the alternatives to GPUs? Are there other types of hardware technologies that could be useful for any type of AI, ML? You know, whatever, deep learning, what what are the alternatives to GPU based L, you know, GPU plus L L M or, you know, chat in video? Kind of as I as I think about it. Are there alternatives to GPU uh a hard on the hardware side?

• 25:25Speaker 04

  As far as I know, I think like TPU is a one option, right? Tensor processing unit. And then there's also these uh essentially onboard computers like on FPGAs or on on, you know, Justin. And these are, these are specifically designed for edge devices. But I know there's also people thinking about quantum computers, essentially trying to look at beyond the classic computers and think about using. Uh, you know, complete a different paradigm to, to compute and, and that thing that also have implications on a machine learning.

• 25:56Brian Keating?

  How, how would you use a quantum, if I gave you a quantum computer, Willow, you know, whatever IBM or Google has, I mean, how would you use it right now? Would it be of use to you or not yet? Or the algorithms?

• 26:07Speaker 04

  Because we're the users of computers, right? I imagine if, you know, the quantum computer is becoming a reality, then yes, we're going to have a massive acceleration in how we can calculate things that we care about now.

• 26:20Brian Keating?

  Okay. The next question I have um is involving what we just went through and I can't say the name of this particular pandemic because YouTube's AI is so sophisticated that if I say the words, you know, C, whatever ends with a D, it will put a warning label on the video, you know, that the United Health, World Health Organization, anyway, it kind of like does something and influences and they're very sophisticated. They look at every frame of this video. They listen to every, you know, byte of data and the audio and they can tell what we're talking about. But talk about, you know, for a general epidemiological studies that you were involved with for a recent event that took over the world, you know, five years ago, uh, from now, uh, earlier than now in 2020. Talk about what, uh, your

• 27:05Brian Keating?

  work involved and, and how AI was useful in determining both forecast and predictions and even implications and getting stuff prepared, you know, perhaps for the next occurrence because we're just a matter of time away from the next event happening. So what did you do for the previous, you know, concern that the planet had this worldwide issue? But also what can AI help us in your work in particular, help us to avoid or prevent or minimize the effect of the next one?

• 27:30Speaker 04

  Yeah. So like basically in epidemiology modeling, they have very similar issues as traffic forecasting, right? Basically you have this like.

• 27:37Speaker 05

  Progressive's Accident Response can contact emergency services if you're in a major accident. The only thing worse than someone you love being in an accident is not being able to help them. So Progressive is here to help when you can't. So get peace of mind knowing your loved ones are protected with Progressive's Accident Response. Learn more at progressive.com. Progressive Casualty Insurance Company and Affiliates. Limited availability in California. Driver is responsible for emergency and towing charges not covered by the policy. Coverage subject to policy terms and conditions.

• 28:08Speaker 04

  Very complex numerical simulator. In this case, it's an agent-based, a multi-agent simulator that is also stochastic. So we need to solve this complex numerical problem. And typically it takes about a week to generate a what-if scenario. Like what if we change the school lockdown policy? Which what if the virus characteristics changes? And you want to understand the potential impact of these changes. So to get that result in a in a week is just too slow, right? During the wartime of epidemic modeling and control. So we designed physics-guided deep learning methods that are hybrid in of using deep learning inspired by these principles and physics to forecast the progression of pandemic up to

• 28:54Speaker 04

  four weeks. So that we were actually ranked number one among 40 national teams in this competition. And then, you know, another thing we contributed is to build a very fast surrogate model. So this surrogate model is also based on deep learning. So it's taking data from both a simulator that we have and also the report data from CDC and slash Johns Hopkins Physics Lab. So then our simulator or our emulator was able to mimic the behavioral simulator, not only in terms of the average prediction, but also the confidence intervals of this prediction. So this confidence interval are particularly important for risk assessment and the decision-making. And then our model was able to reduce the turnaround time originally for one week to one day.

• 29:39Brian Keating?

  How much commonality? I mean, if you ask a normal person, you know, does the traffic in L.A. have anything to do with, you know, forecasting a pandemic? I don't think the average person would see a connection. But you do a lot of what's known as cross-disciplinary and interdisciplinary research with a variety of people ranging from epidemiologists to weather, climate scientists to physicists and at both the theoretical and the application observation layer. Can you talk about, you know, what motivates you? What drives you? How did you get, you know, to a point where you're as conversionist talking about traffic as virus? And how can our listeners who are brilliant, you know, how can they sort of benefit or what lessons can they learn? From your career that they could apply in their in their lives to see connections between different fields that to me seem

• 30:23Brian Keating?

  completely foreign and unrelated to each other, but to you, obviously, they had some relationships.

• 30:27Speaker 04

  Yeah, I think sometimes I think the biggest drive for me is just to make an impact to this world. And you know, when I started working on traffic problem, just because I was stuck in L A all the time in the traffic. And and I was actually commuting between downtown L A and Pasadena.

• 30:44Brian Keating?

  Yeah.

• 30:44Speaker 04

  Um, so so I was really frustrated by by by the situation. I thought I I should do something to improve the status quo. Um, and it just happened that I I I, you know, I reached out to relevant people, got the data. And and then, you know, as a computer scientist, we are trained to make abstraction of about the problem and trying to find solutions to this abstracted problem. And then, once you have the data, you have the model, you have the right abstraction. And you can easily come up, not easily, but you can come up with a solution. And similarly, for, you know, the pandemic, right? And, you know, I was actually stuck at home during the pandemic. And and then I was moving to supposed to move to San Diego, but I couldn't for a year.

• 31:27Speaker 04

  And because of the pandemic, so so I also wanted to solve this problem to make an impact to not myself, not only, you know, myself, but people around me and the community. So I just, I know, I have collaborators who worked on the space and and I reached out to them. And then, you know, I worked with them closely. So I think the commonality I have seen from this past collaboration is one, find a problem that you feel you are, you have the eager, you know, how you're urged to to solve them. Like you feel like you you wanted to solve them and it's so painful to you and I'm sure it's painful to other people as well. Uh, and second, uh, reach to the domain expert and try to understand, uh, what is the pain point?

• 32:08Speaker 04

  And then especially to understand how your expertise in a particular field can help address their pain point. And then after you understand the pain point, right? And then you can formulate the problem and then formulate and take the right abstraction. And then you will think about how to solve them.

• 32:27Brian Keating?

  Do you think we've passed the Turing test?

• 32:29Speaker 04

  I think in a lot of the benchmarks, they have shown, yes.

• 32:32Brian Keating?

  Uh huh. How far away do you think A G I or A S I, artificial superintelligence is? Where AIs are just, don't need us anymore to do anything. As I understand it, they'll just kind of iteratively improve forever without much human input.

• 32:46Speaker 04

  I think that's a very hard question. I, I, as a computer scientist and especially somebody who works a lot on forecasting, and I know most of the forecasts are very uncertain, so I don't want to make a certain forecast about the one we're going to have.

• 32:59Brian Keating?

  You could model it.

• 33:00Speaker 04

  Yeah, actually, I have written a paper about the future of, of trend of AI scientists. And then like we're trying, we actually collected a lot of the research papers in our archive and trying to understand how papers are being formed and how is ideas is evolving, how quickly are ideas evolving?

• 33:18Brian Keating?

  Oh, wow.

• 33:19Speaker 04

  And, and so that's like a forecast of future trajectories of science, but then also has implications on, on AI. But I, I do see like right now, it is very astonishing like how quickly AI is progressing like every single day. Like even for me, working in a field for more than a decade, I have been quite impressed by, by the progress and the speed. But in terms of whether AI is going to replace humans, and I, I don't think so because eventually, uh, the thing, the goal is to have a partnering situation, right? Like AI will work side by side with a human and you know, they provide the type of efficient

• 34:00Speaker 04

  and diagnosis, like knowledge comprehension and just to accelerate the productivity.

• 34:05Brian Keating?

  Mm hmm. Yeah. And be an augment to our creativity, which is somewhat unique to humans, I think it is. So as you know, this podcast is named after Sir Arthur C. Clarke's famous statement that the only way of knowing the limits of the possible is to go beyond them into the impossible. So that was, uh, Sir Arthur C. Clarke. So we're gonna zoom in, my producer Carlo, zoom in or as close as we can. So, um, you are actually the recipient of the very first Into the Impossible Medal of, of Excellence, which has on the front, uh, Sir Arthur C. Clarke's picture. On the back, it has a monolith and the reason for that, I don't know if you knew that, but the, the word podcast. Comes from Arthur C. Clarke, because it comes from the movie 2001 A Space Odyssey.

• 34:47Brian Keating?

  When Dave is talking to Hal, the artificial intelligence that was thought to, you know, be super intelligent back in 1968, when the movie came out or around then, when the book was written. He asks Hal, "Open the pod bay doors."

• 34:58Speaker 04

  Oh.

• 34:58Brian Keating?

  And the pod then became the iPod. Steve Jobs liked that name. One of his engineers came up with it. And now, so we wouldn't have the word podcast without AI and without Arthur C. Clarke.

• 35:07Speaker 04

  Oh, that's so beautiful. Sorry.

• 35:08Brian Keating?

  So, yeah, so on the back, it says the catchphrase, "Only way of determining the limits of the possible and into the impossible." And then on the rim, you know, all coins have three sides. People think coins have two sides, but they have three sides. The coin could land like that. Actually, this one's pretty good. Alvin, shout out to my undergraduate assistant, Alvin. And it has your name and the date of the podcast.

• 35:27Speaker 04

  Thank you so much. Thank you so much. It's an honor.

• 35:29Brian Keating?

  It's not a Nobel Prize, but you know, it's as close as I can get, given all my writing about it. But when we think about AI, the kind of the trope in the popular culture is that it will be dangerous, malicious, super intelligence. I talked with Nick Bostrom and Coin, that term, you know, super intelligence. The notion that it's going to be running away and then turning this all into paper clips or maximize, you know, paper clips. You're shaking your head, so I guess you might be a skeptic of that. Malicious, malevolent AI. Should we have guardrails on it? Or should it be open? We don't know what ChatGPT really does under the hood. We know a little bit more about Llama and the open source ones. Which ones will win? And which ones are more dangerous? And what should we be worried about?

• 36:09Speaker 04

  I think these worries or concerns are definitely valid. And I think there's a lot of research on AI safety.

• 36:17Speaker 00

  Life is all about making decisions. Sometimes you make questionable ones, like buying a new TV that's too big for the wall because the electronics store employee was a little too good at their job. Then sometimes you make great ones, like protecting your home and auto with Progressive. You get customized coverage options so you can be sure you have the right coverage. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers. Not available in all states.

• 36:46Speaker 04

  And it is a big deal, right? Any technology has both sides. Like even the the old in old times, like the nuclear technology, it has both sides. So so AI and obviously as a technology, as a very powerful emerging technology that has also has a positive and negative side. So yeah, so we definitely should have guardrails. And and then one of the, you know, kind of focus in our group is also trying to build these trustworthy AI scientists, right? Because a lot of times when we produce tools for our scientists and these scientists don't trust them. Just because the models are predicting things that don't satisfy laws of conservation. And these are relatively easy to check, right?

• 37:26Speaker 04

  So using our first principle knowledge as a guardrail for AI scientists is something relatively easy. But then there are other situations that are much, much harder to to put guardrails on these technologies. Uh with that said, I would say still it should not stop us from progressing the technology. What I'm worried about is people are too um terrified by the negative uh impact, the potential negative impact of AI. And then it's actually like kind of slowing down the, you know, the necessary technology for us to progress and eventually overcome this negative impact, right?

• 38:05Speaker 04

  So so I I think I'm a big advocate for let's try to disentangle or separate the societal impact with the technological advances. And let the people who are passionate about the technological advances work on the technology, right? And then let separate teams who have great expertise on this to design the guardrails. And obviously, there should be communication between teams, but then let's not stop the the the the development of AI.

• 38:31Brian Keating?

  Try to have on all the different perspectives from Max Tegmark to Jan La Kun, who's on in December. I'll put links to those uh episodes up here in the video. They couldn't be more diametrically opposed, the AI safety, you know, pure safety, it's safety at all costs. And once we release this, it's like, you know, creating our own destruction. You seem uniquely positioned in that you've worked on. The things that connect humanity across a global scale, there aren't that many of them. Pandemic, something propagated virally through the atmosphere. Nuclear weapons, obviously, they affect things in the very thin shell, the atmosphere that surrounds this fragile planet. And then you work on that research as well, infusion as well. And then AIs, of course, now everybody has an opinion about it. Which of these, if you could only work on one, fusion, climate's the other one.

• 39:15Brian Keating?

  Climate is also in our atmosphere. So you think about all the three things in the atmosphere, nuclear weapons, pandemics, diseases, and climate change is in the climate. And then you think about the other type of cloud, the internet and propagating systems. How would you rank those, in concerns of existential risk and worried about the future of humanity perhaps? Where do you rank the different levels? Whether it be climate change that you're involved with, whether it be looking for nuclear fusion as a reliable energy source to alleviate the climate issue and power AIs perhaps? And then epidemiological issues like we discussed earlier? And then the threat of AI? Like, where do you rank? Like, which ones are most important to you to kind of create?

• 39:59Brian Keating?

  Not guardrails, but just to be aware of safety implications and then the benefit? Like, you're a very optimistic person. So what are you most optimistic about? What are you most scared about?

• 40:08Speaker 04

  Yeah, because I think these projects that I am involved, they're actually going to point out to the same theme. Because I care a lot about AI for sustainable development. So when we think about our society progressing as a whole, we need to be thinking about the impact of different things happening on Earth to our children, our children's children. And it's important that we think for them and then trying to find solutions now rather than later. So the climate project I was involved is trying to look at the long-term impact of these different policies and hopefully provide suggestions for policymakers. And then more actionable kind of project is nuclear fusion. What if we can create a new type of energy sources that is cleaner?

• 40:53Speaker 04

  And hopefully that's more sustainable, right? And that's a more actionable solution to sustainable development. Epidemiology is on, you know, undoubtedly a a critical part of this because if we cannot be healthy, right? In general, that we cannot sustain.

• 41:10Brian Keating?

  Shut down research.

• 41:11Speaker 04

  For for for a long term. So in terms of poking garbage on on on different things, I I think, yes, like definitely we we should put guardrails on things that are more tangible and becoming like faster becoming reality, right? But again, I I think it has to be a a dialogue between policymakers and and technological developers. And and and we should have the right people doing their job, like, you know, the expert doing their according to their expertise to do their job, right? We should not force technologists to come up with policies. We should not force policymakers to focus on technological development.

• 41:48Brian Keating?

  Right. People should be experts in their domain and then, you know, more or less. It's important to be cross disciplinary. I mean, you're doing that uniquely well.

• 41:55Speaker 04

  But even for me, like, you know, I I I am a computer scientist and I I am involved with a lot of interdisciplinary project. But I I I always feel like I'm never a climate scientist. I'm never going to be a physicist. And because I have really good collaborators in my team and that's how we were able to make the progress in progress.

• 42:12Brian Keating?

  But you're obviously so learning in all these fields, you know, not not too many uh computer scientists can talk about Lorentz in variance violation, for example. As we wrap up, I got two major topics that are super important to me and my audience as well. We have a lot of listeners out there in academia and science, obviously. We've had twenty one Nobel Prize winners on the podcast, but I want to ask you about the AI scientist concept. First of all, if you can define it, what led you to this this concept and what they what they can do for for us? But but start off with with the genesis of it.

• 42:40Speaker 04

  Yeah, so like it actually just like naturally grow out of the project that we have been working on so far, right? Because we felt like, okay, what if we can discover symmetry from data? And later we thought, what if we can discover equations from data? And we have shown examples that AI algorithms were able to successfully do that. So then that's like then we were discussing with a lot of scientists that in different fields from biology to chemistry to physics. And they all seem to be passionate about improving their own productivity, right? So so then. That's, you know, naturally comes the idea of AI scientists. Again, you know, the goal is not to replace scientists, but rather have more like a scientific assistant.

• 43:19Brian Keating?

  Uh huh.

• 43:19Speaker 04

  Right? So these, you know, AI scientists will work alongside with our scientists. And then they will accelerate each step of the scientific discovery process. As I actually mentioned, uh, in the beginning of this podcast, you start with, you know, data collection or observation, generate hypotheses, and then you analyze and then test the hypothesis experimentation. And then you go back to generate the more hypotheses, right? This feedback loop can be accelerated by AI in every single step. Like example is, okay, before we have to do literature search, right? We will have our student or ourselves, you know, read a book or going on the internet. But these days with AI, you can read three thousand books with, within a lunch break. Like, I don't think any human is capable to do that.

• 44:04Speaker 04

  So this kind of massive improvement in productivity, right? It is just like a very strong indication that AI scientists is going to happen. And it's going to be a big, you know, productive boost for all the scientists.

• 44:17Brian Keating?

  Uh huh. How can we implement it now? I mean, can I, uh, I heard Google has a project called AI Scientist, something like that. But I can't get access to it, even though I'm a scientist. I, I got access to making videos. I applied, but I haven't gotten, I got access faster to VO three than I got to that. And so VO three, I just use for making, you know, uh, interest of the podcast. But how can our listeners and viewers out there, how can they actually get access to it? And put it on their phones or put it on their computers and actually start, you know, doing research with it rather than, you know, hopefully in the future, being able to do it. How can we do it now? Or is that possible?

• 44:46Speaker 04

  So yeah, I think these concepts are still pretty much in the lab type of concept. Like, uh, things that we are working on in our group is that we open source a lot of our code and paper, right? You can all find them on my website. Then you can try it out. One of the web app that we developed is basically showing all the molecules that we, uh, generated with our algorithm. And then if you are a medicinal chemist and you can go on this web app and you can test whether this, uh, you know, molecules are up to your standard. And if you really like these molecules, you can even take them to, uh, synthesize. That's the kind of the feedback loop that we hope to have. Uh to build, right? And now we're also talking with people in, um for example, in in materials. And um there are some professors here at UCSD, they have robotic labs.

• 45:30Brian Keating?

  Um.

• 45:31Speaker 04

  And these robotic labs can quickly, you know, test the different combinations of materials and test their properties. And you can imagine there's a, you know, AI brain that drives all this experimentation. So that you can really connecting, you know, the AI in computer with the AI in the real world.

• 45:47Brian Keating?

  Um, yeah, I'd love to do a tour and figure out what the uh what those are like. Yeah, I was gonna ask you about the next frontier that you're excited about. Is it robotics? Is it, you know, embedded AI systems? We're supposedly gonna have Optimus from from Tesla coming out not too long from now. Or XAI, I guess it's called. What do you see as the as the, you know, benefit to, you know, sign? Let's just stick to sign. I mean, I know it'll be good at unloading my groceries from my car or something. But what can it do? What can a robot like Optimus, can it do anything for me as a scientist?

• 46:16Speaker 04

  Yeah, so I feel like the biggest frontier right now is what people call the foundation model, right? So before, like, people would have to build separate models for individual tasks. Like even in my own lab, you know, we build models for call or discovery, for forecasting, for um hypothesis generation. Um but it's just becoming more obvious now, like a lot of these models share the similar backbone, architecture. And they are also like like transferring knowledge from each other. So the concept of foundation model is to have a single model that, you know, is pre-trained with common knowledge and then fine-tune for different tasks. So essentially, I'm trying to automate my lab, myself out of this loop.

• 46:57Brian Keating?

  That was my last question, yeah, so. In sixteen hundred or so, Galileo's, my other favorite avatar, my favorite scientist in in history. He uh he actually came up with the initial laws of what we now call Lorentz invariance relativity. But he was also a professor and he made his living not only uh being a teacher, but he also had to have his students live in his house. So can you imagine? You're gonna you're gonna have twelve graduate students.

• 47:21Speaker 04

  I don't even have a big old house.

• 47:23Brian Keating?

  Yeah, you don't have a house yet? Okay.

• 47:25Speaker 04

  I don't even have that big of a house.

• 47:27Brian Keating?

  Okay, okay, well, I'll have to get you a bigger house. Uh UCSD should do everything they can possibly to make you happy because you're so valuable to our campus. But the future of academia and in particular of our our profession as professors, you're an educator as well. It comes through so clearly in your passion, enthusiasm, and the word education comes from Latin to bring out of, which is that you bring things out of your students and and you do so so well. And I just love this article in Quantum Magazine that was about about your research and and just your story. It's just so fascinating. We'll have a link to that in the article and the video in the show notes and the podcast notes as well. I want to ask just a selfish question, you know, is our profession, is it under threat? Because why learn, you know,

• 48:07Brian Keating?

  Galileo came up with the with the idea that you should test gravity in the lab by rolling things on an inclined plane. So it was much slower. They didn't even have clocks back then. They had pendulum, you know, they had their pulse. That was it. That was basically all they had.

• 48:22Speaker 00

  Life is all about making decisions. Sometimes you make questionable ones, like buying a new TV that's too big for the wall because the electronics store employee was a little too good at their job. Then sometimes you make great ones, like protecting your home and auto with Progressive. You get customized coverage options so you can be sure you have the right coverage. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers, not available in all states.

• 48:51Brian Keating?

  So to measure something falling in free fall was really hard to do that accurately. And that's why he took 1600 years, you know, from Aristotle to him to prove that heavier things fall at the same rate as light things with the leaning tower of Pisa, whether he did that, we don't know. But it's certainly his story is great. And then the question I have is, is, you know, as a professor, why learn, you know, physics 1A from Brian Keating? Why learn, you know, about the ball rolling down an inclined plane, which from from me, when you could have an AI avatar, you could have a robot, you could have all of his words are digitized. We did the first ever audio book, which was took 21 hours to read. All this exists, or Einstein, learn relativity from Einstein, not not from Brian

• 49:32Brian Keating?

  Keating. What are the risks from me to me to you perhaps? Although I think you're more impervious than than anybody I've ever met. You're like the MacGyver of. Of of science and and and technology. But but tell me, what do you see as the future of education? Are we gonna have jobs? Are our students gonna have professorships? Or maybe, you know, sooner than that? Is a professor still gonna look the same in twenty years?

• 49:53Speaker 04

  Yeah, I think that's a very intriguing question. Like, you know, obviously academia is facing a, you know, an identity crisis now, and especially in computer science, right? Because I think the most exciting application for L M is for coding. And for writing code.

• 50:08Brian Keating?

  Yeah, I love it. Even I'm horrible. I never learned to code. And I feel like, oh, it was good for me to wait because I could use that energy to build stuff in the lab. And now I have cursor do it and, you know, does it much better than I could do.

• 50:19Speaker 04

  Exactly. So so, yeah, like, I think back to your question, like, we do have to rethink like our role as educator, as a professor in university, right? So knowledge, you know, knowledge sharing are kind of just like educating the next generation will look very different from now. Because and I always feel like, well, are we going to be out of jobs? Um, maybe. But then, you know, as I think about it, a lot of times when I interact with my student, it's actually not about the knowledge I share with them or like a specific technical concept I teach them. Right? It's really just about, you know, this personal interaction and how they are were able to like learn from this very nuanced the details of how I teach and how I think

• 51:04Speaker 04

  and how I try to solve a problem. And I feel these kind of nuanced details or interpersonal interactions are are always going to be there. And that's actually very important for for education, right? Um, and I think I don't think these type of, uh, you know, nuanced interactions is all is is going to be replaced by by A I. Right? As because A I is just going to be a much more uh, efficient tool out there for us to use. But it's not going to replace these type of very important roles.

• 51:36Brian Keating?

  I feel like A I is almost training us in some ways. And and the education that the reason I brought up the Latin of that, which, you know, education, educare, it means to pour out of, not to pour into, which we don't want to think, oh, I'm just spraying information and my students are just, oh, I love your information. No, but they don't care. But really, the job is to bring stuff out. And I feel like that's what a prompt is. You know, so we're teaching people, you know, we should be teaching people how to prompt. You know, the joke ten years ago was learn how to code if you want a job. Now I'm proof you don't need to really know how to code. You know, it helps them know a little bit. But teaching, you know, my kids how to prompt. I mentioned this, this Sora, Sona, or sorry, I forget what it is. You can make, well, you can make videos, but you can make these beautiful songs. And she was prompting and she was like, oh, I really want, you know, Dua Lipa's, you know, kind of

• 52:21Brian Keating?

  style. And then the AI comes back and says, you used a forbidden word. You can't use anyone's actual style. So she learned, oh, okay. So I have to say like, describe what Dua Lipa is like, you know, what her sound is like, her style is like. So she learned how to prompt better from the AI, even though the AI, you know, kind of rejected this prompt. So I, I find it very, very fascinating. And so I want to just conclude maybe with if personal story about how you came to be, again, the name of the podcast, Into the Impossible. I like to conclude a lot of the interviews that I do with this question, you know, if, if the only way to know the limits of the possible is to go into the impossible. What kind of lessons or teachings would you give to a younger Rose Yu, a 20 year old Rose Yu or 15, whatever, a kid, as a kid? Maybe a little bit of your personal story, your backstory.

• 53:06Brian Keating?

  If you had 10 minutes to talk to her as a little girl or whatever, what would you say to her? What would you do to give her the courage and the, and the intensity, the passion, the, the, the caring and the intellect that you have that allows you to go into the impossible? What would you say to her?

• 53:19Speaker 04

  Yeah, I think, you know, if I were going to talk to my younger self and I would say like, you know, just take a little bit more risk. I think because as an immigrant, you know, I came here after my undergrad and to get, got into a PhD program in the US. So that was a big deal to my family. But then, you know, because I was, I'm an immigrant and I always have this kind of anxiety that, that I, if I don't follow things, you know, exactly, if I don't, don't follow the, the dogma and that I won't be able to stay in this country. And but it does seem now it's even getting harder to stay in this country. But then if I, if I had an opportunity, right, then I would say, well, why not, why don't you take a little bit more risk and then just trying to do things that.

• 54:03Speaker 04

  You feel you can make an impact, right? So, so like, you know, but, but, but when I, when I started out as a PhD student, I was just simply following whatever my professor gave to me at the research project. So I spent like two years working on projects I weren't really passionate about. Um, and then like later on, right, as I started to become more confident in, in things and I know how to pick projects myself, even those projects were a little bit far away from, you know, my advisor's research direction. And, and later on, I expanded to domains that are not even in, in computer science, like outside of computer science, right? And, and you always get questions from, you know, sixty or eighty years old professors

• 54:45Speaker 04

  in, in this domain, like whatever you're doing is, i- i- is garbage. And, and I, I don't, I, I don't think, you know, uh, people like, uh, in young, in the, in the audience, and, you know, if you wanted to do something you're passionate about, just go for it. And obviously, you're not going to please everybody, right? And, but you feel, feel you're solving a real problem. And, and I'm, I'm sure there will be people who are appreciating and will give you support and, and encouragement that you need to, to make, to make progress.

• 55:14Brian Keating?

  That's wonderful. Well, if they try to come for you, they're gonna have to go through me, Rose. I won't let you go. You're too valuable to, to us, UC San Diego. Uh, thank you so much for everything you do.

• 55:24Speaker 04

  Thank you very much. It's a lot of fun.

• 55:26Brian Keating?

  It's so much fun to have you here. I will put a link to your, the article about you, which I think is just the first of many. Uh, it's so impressive and I'm so proud to have colleagues like you at UCSD. It's what makes being here so wonderful.

• 55:36Speaker 04

  Thank you, the honor. Thank you very much.

• 55:38Brian Keating?

  Thank you.

• 55:39Speaker 02

  I'm so proud to have colleagues like Rose Yu at UC San Diego. She's exactly the kind of fearless, cross-disciplinary maverick thinker that we need. As AI reshapes everything from traffic patterns to the fundamental laws of physics itself. Her optimism about human-AI collaboration, combined with her track record of solving real-world problems, gives me hope that we're heading towards an augmented intelligence, rather than a replacement of our human intelligence. And if you enjoyed this deep dive into AI's role in scientific discovery, you'll love my episode with Jan LeCun, where we explore whether or not artificial intelligence can truly understand the world, or whether or not it's just a very sophisticated pattern matcher. Two brilliant minds, two very different perspectives on the future of machine intelligence.

• 56:21Speaker 02

  Don't forget to like, comment, and subscribe, so you don't miss our next breakthrough episode.

• 56:28Speaker 00

  Life is all about making decisions. Sometimes you make questionable ones. Like buying a new TV that's too big for the wall because the electronics store employee was a little too good at their job. Then, sometimes you make great ones. Like protecting your home and auto with Progressive. You get customized coverage options, so you can be sure you have the right coverage. Get your quote today at progressive.com. Progressive Casualty Insurance Company affiliates and other insurers. Not available in all states.

• 56:57Speaker 06

  They're running a Cover 2 defense. But you're running to get game day deals on Uber Eats. Keep everyone at tonight's watch party fed with deals from your go-to restaurants. Or, stock up with deals on your favorite snacks and groceries. Whatever the occasion, get game day deals on Uber Eats all season long. It's not football without food. Uber Eats. Order now. Terms apply. Product availability varies by region. See app for details.