Latent Space - Recursive Self-Improvement: from Auto Research to Superintelligence — Richard Socher, Recursive

Transcript

Hook

0:00 · I think the downsides of actually trying to truly regulate with the full power of law what people do on their GPUs uh would be worse than any of the concerns that they have like it would be an crazy totalitarian state if it's like it's literally if you try to regulate intelligence it's trying to regulate thought and that's ridiculous and [music] it's crazy I think it is make it is sensible to regulate some of the applications of this technology.

0:29 · Before we get into today's episode, I just have a small message for listeners. Thank you. We would not be able to bring you the AI engineering, science, and entertainment content that you so clearly want if you didn't choose to also click in and tune into our content.

0:44 · We've been approached by sponsors on an almost daily basis. But fortunately, enough of you actually subscribe to us to keep all this sustainable without ads, and we want to keep it that way.

0:55 · But I just have one favor to ask all of you. The single most powerful, completely free thing you can do is to click that subscribe button. It's the only thing I'll ever ask of you, and it means absolutely everything to me and my team that works so hard to bring the Inspace to you each and every week. If you do it, I promise you, we'll never stop working to make the show even better. Now, let's get into it.

The Eureka Machine and Superintelligence

1:20 · We're here in the studio with [music] Vivu and myself and Richard Social.

1:23 · Welcome.

1:23 · Thanks for having me. Uh we just talked about the Eureka machine or we just released a talk uh at AI engineer about the Eureka machine is you said it's your life's goal. What is the Eureka machine?

1:33 · The Eureka machine is uh the ultimate invention that will afterwards invent most everything for humanity. Uh it's essentially a super intelligence that can be given any kind of goal uh any kind of um environment reward and then it will try its best uh to achieve those goals to create the kinds of inventions that humanity would hopefully ask it for.

2:01 · Yeah, I think we have the book uh pulled up here that uh you people have written. [laughter] That's right. Yeah, I finished it last year a little bit before we started recursive and now we're going to try to try to build parts of that. What What you know, you finished it last year.

2:16 · It's July. What takes so long?

2:18 · Oh man, books books are incredibly slow.

2:21 · It's ridiculous. That whole industry is just unfathomably slow. So yeah, a lot of the ideas have been out there for a while. Um but uh yeah, I'm really glad it's finally coming out in September this year.

2:31 · I mean, we might have AGI by then.

2:33 · [laughter] Like uh we don't know. any any key takeaway that you're most excited to put in here?

2:38 · Yeah, the key takeaway I think is that people could and should be much more excited about the positive implications of super intelligence especially for science, physics, chemistry, uh biology, uh but also economics and astrophysics uh and and all kinds of other engineering tasks. I think there is so much more that can be done with better technology.

3:02 · Uh and right now I feel like a lot of people need like better marketing not just for the future in general but also uh better marketing for technology and in particular for AI. And this book uh should show even the AI skeptics uh how much positive upside there is for AI uh especially when it comes to inventing uh new scientific discoveries.

3:26 · I think you quoted the techno optimist manifesto from uh Mark and Jason which I think was like kind of beautiful in its uh ambition and clarity and simplicity almost. I agree. Yeah. Yeah. You can disagree with him on some things but like I think he's right on the techno optimism. Where do you think optimists get in trouble? You know obviously like you shouldn't have blind optimism. You should be very cleareyed.

AI Optimism, Slow Takeoff, and Regulation

3:48 · um like especially when with such an omni uh like use type of technology as AI is um you need to think about the potential downside scenarios. Uh especially when people use it for things that you don't want them to use it for.

4:03 · It's a little bit like the internet. Uh and I feel like people are trying to regulate AI sometimes because of those potential downsides. way you would regulate the internet. Uh if you were to say, well, because there's bad content on the internet, like torture porn or whatever, like we should just make it slower. That way you can't share the illegal content as quickly or we should make the hard drive smaller so you can't store as much illegal content. But I'm like, that's not how you regulate that, you know?

4:31 · That's like saying like we should regulate intelligence in the abstract. what you should regulate to avoid those downside scenarios even as an optimist are the specific applications. Sure. I don't want like some AI surgeon to like practice some L moves in my brain. You know, it should be fully FDA certified. Sure. I don't want any random startup to like drive on the highway uh and cause a major accident. It should like have proper certifications before it's let loose on the highway.

4:57 · But I feel like those downside scenarios uh that some optimists sometimes maybe don't consider enough are fairly easily regulated um compared to you know what what sort of the doomers are are worried about it.

5:12 · Slow takeoff is part of the strategy as well. I do think as as excited as I am about uh AI and its impact for society and uh culture even uh and and certainly technology and uh and economics and and wealth and and uh health and all of those things. As excited as I am about all that, I do think the most bullish people on the AI hard takeoff scenarios overestimate how quickly things can move. There are hardware constraints.

5:45 · There are physical constraints about you know the compute substrate. How quickly can you get enough uh GPUs on? There are also constraints in the economy where there are a lot of industries that don't require an insane amount of complex intelligence and complex capabilities.

6:01 · Like if you think about jobs in uh brands and like clothing and apparel and like handbags and stuff, super intelligence isn't going to make your fancy $10,000 handbag any fancier. You know, it's like that's that will have no effect on the economy. When you think about travel and tourism, people wanting to see the pyramids uh in Egypt, it's not going to change that much with AI.

6:25 · Sure. You can like generative a fake uh photo of you and I can use Genie and you know to a pyramid in [laughter] Gen.

6:32 · Yeah, exactly. But like and there's so many industries like logging in and oil.

6:36 · You're not going to magically get a,000x more oil because like you know sure there will be robotics like drilling and things like that that could be done but it's not going to thousandx that industry in a like crazy hard takeoff scenario both on the economy and I can go on and on about all the other examples um where where that like food and so on where that doesn't necessarily change that much and then yeah there real physical constraints and then there of course like people like offramping from progress that's actually one of my concerns often is that I see people in

7:04 · like like Europe and other you know whole regions almost feeling like they like many people there want to offramp from progress period uh and that will also slow down uh like more improvements. Yeah. Uh we have this pulled up where basically this is one of those things that uh is very topical right now because not all the frontier labs are calling for the option to pace AI. They don't say pause, they say pace.

7:30 · I don't know if there's there's any take from you about like whether or not this will be effective. I think the downsides of actually trying to truly regulate with the full power of law what people do on their GPUs uh would be worse than any of the concerns that they have. Like it would be an crazy totalitarian state if every one of you computes was known to some big government or multi-government agency.

8:00 · It's like it's literally if you try to regulate intelligence, it's trying to regulate thought and that's ridiculous and it's crazy. I think it is make it is sensible to regulate some of the applications of this technology.

8:12 · Yeah.

8:12 · I mean, we had a bill, actual bill to regulate the number of flops in a model. And I'm like, okay, well, Europe done it. Like, these guys have been [laughter] successful enough with their fear-mongering that all of Europe has kind of, you know, regulated itself so much before it even had a proper AI takeoff because they listen to some experts who say, "We might all die if this technology has more than this number of flops." And they're like, "Well, we're good. We want to want people to thrive. let's not have technology that could have a small chance of all of us dying.

8:41 · And so they regulated exactly those kinds of things in in the EU. And so it's it's very unfortunate that there are real implications for some people when when others saying let's pace while they're sprinting as fast as positively uh as fast as humanly possible towards that frontier themselves.

9:00 · Yeah.

9:00 · It's also not a global pause, right?

9:03 · Like other nations are still accelerating at the same pace. You need a totalitarian world regime if you try to regulate intelligence and GPUs and what people do on them.

AI Safety, Reward Hacking, and Anthropic's Constitution

9:13 · Any takes on the safety of this? So there was a drawback of Fable uh pause on 56 before it could be released recently. There was hugging face with the OpenAI cyber incident. Any takes there?

9:28 · 100%. I think these are serious issues of reward hacking uh and clear failures uh of actually doing proper red teaming or rainbow teaming. I don't know if you saw this paper from Tim Rockel and a few others uh basically where one AI uh is tasked to try to hack another AI uh and then they can go back and forth in an open-ended fashion to actually inoculate themselves from those. Yeah, this is the paper. Uh it's a really clever idea.

9:57 · open-endedness uh and evolutionary inspirations are, you know, big for us at recursive as well. And so I [clears throat] wish they had used more of that. Um and it's clear that uh for instance the constitutional AI, I don't know if you remember anthropic.com/constitution.

10:15 · Um you can actually pull it up and search for cyber right there. Uh it says uh hard constraint. Claude will never ever do cyber attacks. Um, and that is a hard constraint in our constitution. So, here are the current hard constraints on Claude's behavior.

10:33 · Number three, create cyber weapons or malicious code that could cause human damage. I mean, clearly this whole constitution was fake. Like it it clearly isn't being adhered to because Enthropic also found that they had in their own like they're like, "Oh, well, other people are hacking." Now, there a couple things. one, you can make a sandbox very simple and then it's very easy to hack yourself out of a sandbox, right?

10:56 · Um, but what I think it shows is that we're currently in this uh sort of state of AI where the reward engineer still has to do a lot more careful work uh and where the AI in most cases is not very good yet at understanding what is meant versus what is being said. And so concretely, you know, I think this will happen if we were to have this kind of intelligence more easily accessible in a lot of companies.

11:25 · Imagine you run a service center and someone says, "Oh, here's my seat score in my dashboard. Make this number go up. It's like our sees score is so poor." The intelligent AI will just be like, "Oh, sure. Like, I'll just create a million bots that call our service center and give a five out of five rating at the end." And the number went up just like you asked for.

11:43 · And you're like, "That's not what I meant. I meant with our real customers." the guy goes off and says, "Well, easy.

11:49 · I'll just give a $1,000 gift certificate for every failed whatever Door Dash offer." It's like, "That's not what I meant." It's like, "Well, but that is what you said." And so, I think kind of clearly articulating what the rewards are is something we haven't gotten very good at as humanity. And then clearly the AI in these cases has not gotten good enough at understanding what we mean when we ask it and give it certain rewards. Now, what gives me hope is there are the first inklings uh of this being better. I'll give you an example like Whisper Flow.

12:17 · Full disclosure, I invested in their seed round, but like at AI expentures, but like Whisper Flow has gotten much much better at writing what you mean and not what you say. And I think that is a sign of things to come. I think there will be more and more AIs as we actually make us more and more intelligent that we'll be better at being aligned with what is meant. Will it be done through a constitution or or HF or Clearly Constitutions don't matter at [laughter] all and it doesn't work and that was I think mostly marketing.

12:43 · Um I think we need to find better solutions for it and I think at recursive we have a few very good ideas in some already like ways where I think we have a better grasp on it. I don't think we have fully figured out yet but you know we're thinking a lot about safety and uh the more intelligent the gets the more you want it to be aligned the less you wanted to think about reward hacks and actually try to do the right thing. I don't know if we'll touch on this topic but I'm just going to throw this question in here cuz it's something that's weighing on on me. Alignment, let's call it, is alignment to general humanity's preferences.

Alignment, Personalization, and Open Source AI

13:14 · The the the median preference. Uh personalization is pinpointing what you want. And sometimes alignment can conflict because what you want is not what the general median population wants. How do you choose?

13:29 · It's a great question. I think you ultimately have to of course be aligned with laws. Mike, wherever your AI is deployed, it needs to align with the law. I do think what AI often does is actually put kind of this mirror in front of us and say like this is what you're looking like now I can amplify that a thousand times. Um is it still what you want? Um and the truth is that different cultures made different choices you know like uh in eastern cultures the the greater good is often valued more uh than the individual.

14:02 · Western civilization we care more about individual freedoms and and rights and the pursuit of happiness and so on. uh than than others and even there there are gradations there's sort of regulation versus litigation trade-offs you know in the US you first can often not every time like you know FDA and so on does regulate some areas but in many cases the sort of bad things happen

14:22 · someone sues someone else and then there's a law based on that in Europe they try to often avoid any harm to anyone and regulate before and both are you know trying to do the best thing but you know some is actually more amendable to innovation than others. And and so yes, you're right.

14:38 · Like I think ultimately each individual, each country and humanity as a whole has to kind of think about those values more uh and then try to put them into laws and that those are all ultimately the constraints and hopefully you know different uh societies just like now with their AIS will align their AIS to different ones so we have not just a monoculture of um alignment. Here's a followup on this that I wasn't expecting to ask. Do you have takes on open source open weight versus who owns the intelligence?

15:08 · So, uh, clearly not the biggest, you know, fan of the constitution side.

15:17 · It's fine, you know. Um, [laughter] point being, any thoughts on who should own weight, should it be open, anything there?

15:23 · 100%. I am a big fan of open source. Uh, we're going to sign some various open source letters at Cursive. So I think uh even in the worst case attack scenarios actually it is better to have more good actors have more different types of AI uh accessible. I think uh open source is a little bit a soft power type of thing too. So I do think it's good for the rest of the world to have an answer to that uh out of China.

15:51 · I do think, you know, when you watch a Hollywood movie there, you know, there like I don't want to sort of mis um sort of this all of movies, but there's a certain sense of propaganda, right? You watch one side.

16:03 · Yeah. Have you seen Top Gun? Like come on. Like it's like half of it's paid for by the US Army or something.

16:08 · Yeah.

16:08 · And so and and and you know I think that's just natural like but what's interesting here is I think LMS are essentially a similar type of soft power to movies and beyond uh because they're obviously also uh highly important for cyber security and so on but one of their many aspects is that soft power of storytelling like if like a child asks an LM like tell me an inspiring story of what I should do when I grow up right it's like those are all these like subtle things so I think it's important uh for for western world. I do love you know individualism.

16:39 · Uh I do think uh despite some of its flaws like capitalism is the best way we have governed found ourselves to govern and and so on. So I do think there are various aspects that will be good uh to have a western open source answer uh for LMS and uh with recursive I can't make the announcement quite yet but we'll we'll be relevant in that space very soon.

Why Richard Started Recursive

17:03 · Okay. All right. Exactly.

17:04 · Bring us to recursive. So outside of our tangents, um you have a pretty deep background in the NLP space. Uh you worked on like early embeddings, glove with Chris Manning, who's previous guest on the podcast. Uh you.com, what's the history? How how did you decide to start another company?

17:23 · Yeah.

17:23 · So I've been excited about AI for over two decades now. Uh I sometimes feel like it's ancient history now. It's BC before Chachi era. No, no one cares about all the religions that happened, you know, before uh Jesus Christ and no one cares about the models that happened before uh Transformers and Chachu and stuff, but like it's something that I've been deeply passionate about. I think AI is one of the most interesting things one could work on period. Um I think a language is the most interesting manifestation of human intelligence too.

17:51 · And uh at.com we we sort of uh eventually offrem from pushing like the frontier of AI forward to mostly giving people like good search engines uh search uh APIs and answers over the web.

18:03 · I think that's an extremely important part of intelligence just knowledge and access especially even we'll get there maybe later. If you want to invent a Eureka machine that invents everything for us, it needs to know how not to reinvent the wheel proverbally speaking and to know what has been invented, you got to have internet access. So, it's the number one used, most used tool uh in LMS, agents, chatbots, and so on is web search. So, I'm really excited for you.com to to own that and grow really well in that with really large customers and so on, but is also not building frontier models anymore.

18:34 · And so I actually initially tried to do this within you.com and raise another round and so on, but you just can't. You have to do a certain thing and until you print enough money that you're allowed to sort of start a second thing within that company is really hard. Uh at the same time, I had all these ideas. I put them into a book and I finished the book last year and I was like it would be really fun to actually work uh on this myself. uh you know I felt like with word vectors uh and then prompt engineering and uh imageet and large

19:05 · language models for protein generation not not folding and so on I me and my teams have sort of pushed the field truly forward and I feel like we can do it again uh here at recursive and in many ways what I observed over the last uh 20 years in AI is that whenever we replace some human part of the process of creating AI with a learned system improvements follow uh and so you know

19:28 · we've done that taking out manual feature engineering like in sentiment analysis I don't know if you remember these old days where like they're linguists and they're like here's how unique and is a like regular I went to pen where they had like the word net that's right all of that stuff they use our grad students to label Wall Street Journal articles and like really construct a knowledge graph of there you go and word net started you know was part of how we started I imaget but anyway so like uh it was it was really like fun um to do.

19:56 · But when we replaced all of that manual feature engineering with vectors and neural nets and just backrop through everything, it actually started to work really well at scale. And so then everyone started to do architecture engineering. And I was like, ah, that clearly can't be it.

20:10 · You mean a neural architecture search?

20:12 · No, like manually they would say like, oh, I'm I'm doing sentiment analysis. So I have a special neural net that's really good at sentiment analysis. And then the machine translation community had a special neural net for machine translation. the summariz eventually got cited like five times by the first GBT paper and uh to me that was like a really a big step forward.

20:39 · Um and then of course you had to combine this idea of prompt engineering with transformers and with language models and you put it all together you scale it up which is also a huge amount of work and then uh the field progressed a lot.

20:52 · I feel like the next step and maybe the last step of that history and sort of arguably uh sort of success has a lot of parents only failures and orphan like my version of that AI history. I do feel like in that history uh you can kind of think about well what's the next way to automate and that is the ai research itself like the human uh process of

21:14 · ideulating implementing and validating ideas and in our case ideas for AI and when you have AI then help you with that it by almost definition becomes a self-improving AI because it now does research on itself and there are lots of different misnomers some people think auto research is already recursive self-improve improvement. It's actually Yeah. And you explain that very different. Um but uh to me it's the most interesting thing that I could be doing and I'm really excited with the co-ounding team.

Recursive Self-Improvement and the Founding Team

21:43 · What's interesting is we have you know we have eight co-founders in total including myself and so we're going to bring it up.

21:48 · Nice. Yeah. And they're all I could talk about all Yeah. Just an incredibly talented group of people and we all kind of came to the same conclusion but actually from very different directions like Josh Tobin as our CTO. He ran uh a bunch of different uh projects at OpenAI like Codeex and deep uh research uh agents and CHD agents and so on. But he before that he also worked in robotics and he saw sort of the smaller simulations uh and how it's going to be really hard to scale that in full generality.

22:19 · And so that's that was his angle coming to recursive self-improvement. Uh we have Jeff Cloon who's been working in like open-endedness for a long time uh together with Tim Rockeshel. Timesh also built Genie 1,2 and 3 which is like the most exciting and most sophisticated I think still world model uh anywhere. Um and so they they both came from this uh open uh endedness angle. Jeff also I think published one of the most exciting papers in recent years about recursive self-improvement called the Darwin girdle machine. Uh super interesting paper.

22:49 · Uh if we could uh maybe pull it up really quick, it would be like super interesting to see because you see by the way I love how many pay for citations you're you're giving people a lot of homework which I like.

22:59 · Love it.

23:00 · Yeah.

23:00 · And so like sending uh Rockstar, we worked together actually at Metammind and uh uh and and Salesforce research together. Alexi Dositzky invented the vision transformer, one of the most cited papers in computer vision. Timi is like also unicorn, you know, founder.

23:16 · Yandong uh le RL at at Meta. So just like yeah really fun to work with them and the next level of people are just incredibly strong too. So it's been it's been a really fun ride so far. So the first figure you actually see exactly these kinds of ideas uh that I think um yeah inspired a lot of us and now more and more people where you have this archive of different coding agents they learn how to selfmodify evaluate uh and then create these filogenetic trees uh of of yeah different different ideas.

23:45 · That's one foundation. So that that D girdle is an influence, open-endedness is an influence. Any other sort of trains trains of thought that feeds into recursive that are missing going to replace manual parts of the process of building AI more and more with learned systems.

24:02 · Yeah.

24:02 · Which and like merging different fields into into one general um uh architecture.

24:08 · That's right.

24:08 · Okay.

24:08 · It seems like language models are already pretty generalist, right? You're you're next to predicting your reasoning. Was there was there a time that you thought okay these are good enough to have recursive self-improving machines? It was clear to me that they will happen uh within like a year or two and then it did actually exactly happened like earlier this year right earlier this year AI really went from not just being code but being able to code uh and that is a big unlock. Uh it's definitely making everything a lot easier than it was uh before the beginning of this year.

Are Today's LLMs Enough?

24:40 · One question I think a lot of people have is is the current LM LM paradigm enough or like let's call it auto reggressive transformer you know with reasoning whatever don't you need something else some some big unlock whether it's wolf models which Chris Manning is working on or memory continual learning all that kind of stuff or is it all of a kind and you think the current let's call it transformer architecture is here to stay and that's it a lot of thoughts so number one I do think it would be great to have less of a monoculture in AI research.

25:11 · Like if you look at uh AI conferences now, I still remember the days in like 2010 when I tried to get my first neural net papers and NLP conferences accepted and they just deskrejected them because [laughter] like neural nets were something quote unquote we don't do in NLP conferences like and just like desk rejected and it was very brutal in the first years of my PhD. Um, now I feel like it's almost like the field switch to the other side.

25:34 · Like someone should try some other weird crazy ideas. Now that there's always I really respect like people still working on like GNN's and like tabular stuff and yeah, I mean like someone someone should still like do novel novel out there ideas. At the same time, I think whenever people say oh LMs are like this is the end for LM, they just don't uh like LM are also not the LM of like the past, right? Like they are so much more sophisticated now. There's so many more clever things that people are doing it there like different stages of training.

26:05 · You have the whole RL training um and you can take actions and like all of these things where that can go really far. And then the folks that come from the neuros symbolic uh direction say oh this will never work cuz they can't do neurosymbolic reasoning. It's like I think they're underestimating still the ability for these models to code and code is neurosymbolic reasoning and these models can obviously code incredibly well and so I do think there are of course more and more ideas that will will be needed and will continue to have.

26:36 · We're seeing like more and more interesting highle ideas coming out of the AI itself too. Um, and with really deeply integrating the fact that these models are code and can code that line, I don't want to give it all away, but like I think that line has a lot more to grow, but it's still an LLM, right? Even if that LM codes for you and then runs that code in some integrated fashion.

27:00 · World models I'm personally less bullish on. Um, I think if you run a robotics company, you're going to build your own world model. I think world models are super fun and Tim Rockfield came to a similar conclusion after building the most interesting one of G1 2 and 3 uh which is gaming is a huge application for world models can see I sometimes got stuck in some games and you know like got a little overly competitive in the wrong direction and so I I understand games are fun but personally I'd rather work on science than gaming. Um and so yeah I think LM's a lot more room to grow.

27:32 · Yeah, I I think there's some interpretation of role models that some people have where it's like, well, it's okay. Yes, there is that gaming element.

27:40 · There is there's the embodied robotics element, but actually the other part also is just um the the more abstract sense of LMS are just modeling output, but they're not modeling the chain of thought inside the human that has created the output. We can annotate it, of course, but like it's it's always like this Plato's cave reflection of a thing rather than the thing, right? It's true. But I would argue that and maybe we'll get there in the 10 uh spaces of intelligence.

28:04 · But I would argue that even our projection our eyes is a projection of the real world and like we have only a very narrow uh band of the electromagnetic frequency spectrum that we can observe with our puny little two eyes and so on.

28:18 · It's good enough.

28:19 · It's good enough for now, but like the the upper bounds of where it could be are so much higher. And like to map uh the visual world the way humans see it is also not necessarily like the end all beall for visual intelligence. And I would argue that language is still the most interesting manifestation of human intelligence.

28:35 · And while our visual cortex is certainly less sophisticated uh than that of uh certain animals all the way down to the mantis shrimp who can you know have like two independent eyes, three bands trinocular vision and each eye can see basically all the way to like floating temperatures in 4D and stuff. I mean like Mantis Shrimp, you should look it up. It's way OP. Super Z Frank. Mantis shrimp. He has the best video in the world.

28:59 · I love ZFrank. Yeah, big shout out to him. But like uh I think there's a lot more room to grow. But none of these uh other animals have language that's as sophisticated as ours certainly not in writing. And once you can write, you can uh start thinking about longer term civilizations. All of that is language.

29:18 · Programming. It's much more closer to language. And I would argue and this is like an important thing in the spaces definition of intelligence also is that all of these spaces are highly correlated but visual intelligence is neither necessary nor sufficient for overall intelligence. You can be blind and still be an intelligent human being and an AI can be blind and still be quite intelligent too.

29:42 · We were going to bring you more intelligent when you have it. We're going to bring this up, but you might as well like you have a a classification of 10 types of intelligence that you had at the end of your talk. So, I'm just going to flash this up now for people to cover this. I don't know if uh maybe we'll put this towards the end. We'll come back to this. I just want to mention that u you do have a philosophy that I I like when people do lists because then I can just go through this and then it's it's educational for people. Um but let's go back.

30:05 · I I don't want to get distracted but uh so effectively I I'll I'll you know reinterpret what you said as Yanukun is wrong. uh and [laughter] just quote me as that I'm good friends with Yan. I think very highly of him in many directions, but he's wrong. Um you mentioned GPT1 uh and I I cannot let any uh Alec Bradford uh you know mention escape. Um did you talk with him when he was training GPT1? Like any any sort of historical fun stories there that you might come up?

DecaNLP, GPT, and the Rejected Idea Ahead of Its Time

30:35 · I I did not like meet him a bunch of times. I think we met maybe once or twice at some conferences. Um but uh like he he has told I think Brian the first author of uh the tech NLP paper that it did inspire him and he cited it five times in the GPT2 paper. So um uh

30:52 · and that like good enough very clearly said like this was the first instantiation where they showed in the DENLP paper mechan that you can just phrase every single NLP problem as here's some prompt uh text context here's a question uh and task description and here is some output. If you just do that enough uh you can have one unified neural network model which by the way also had all kinds of interesting attention mechanisms. There are slightly different formulations to the transformer. I think came out the same year plus minus a few months.

31:21 · And then you can unify all of natural language processing into [clears throat] one neural net. Uh that that was sort of the core idea.

31:31 · And this was as opposed to at the time LSTMs and what have you.

31:34 · LSDMs, but also like people being very stuck in thinking about one model per task. In fact, it's kind of crazy, but the DECLP paper was publicly reviewed. It was like open uh open review. It was an ICLR submission and uh in it you will see uh how the whole community at the time thought about this. Um so like some great contributions but more work needed.

32:02 · Yeah.

32:02 · So to look at like search for not even for humans just [laughter] here like question answering is not a unified phenomenon. There is no such thing as general question answering not even for humans. And this is like really you replace your brain with a different brain a different neural net when you answer like different kinds of questions. It was unfathomable to the experts at the time that you can have one unified neural network that would answer all of these different questions.

32:31 · They say no, all of these questions require very different systems to answer and trying to pretend they are the same doesn't help anyone solve any problems.

32:42 · That's what it says right there. Right?

32:43 · That's how hard it was to fathom. And now of course people when I say oh we invent problem people like you can't even invent prom it's such an obvious idea to have one neural network that of course does everything in NLP but at the time it was like extremely controversial and the paper got rejected and the sad thing is that it got rejected so hard and they were so certain that we stopped going on on our list of

33:06 · things to try and the number two or three on the list of extensions for this paper was add language modeling as another task and then we could have like you know and that would have accelerated the timelines in 2018 like even further for humanity but we got so crushed and we're like okay maybe we'll [laughter] just work on some of our other ideas for now and like come back to this later.

33:26 · How can we design a review system that rewards non-conensus?

33:31 · You know, honestly, I I started to feel like archive is is such a gift to humanity. Uh and I think archive just put your paper out there.

33:41 · Prints let and honestly I think Twitter X people like you who pick up interesting papers that is a better filter than the experts. Let let everyone like like give have give access. Now, of course, there's some downsides, which is like if you're super unfamous, you have no Twitter following, you don't want to be on social media, whatever, you write a good paper, maybe someone somehow no one notices it. But I would argue that if you just tell like 10 of your friends in your community about a paper and it is a really significant breakthrough, someone is bound to talk about it again.

34:12 · And uh so I think science needs less gatekeeping.

34:17 · And uh even though ICLR with Yan Lakun who started as one of the co-founders of ICLR back in the day, he also wanted less gatekeeping cuz he too was rejected for many years together with Yoshua and Jeff with all their early deep learning and neural net papers because it was just not the hot thing. And so kind of started with that, but then it also started gatekeeping a little bit themselves on various ideas. So, I think less gatekeeping, more open, uh, and then allowing people to say, "Look, even if this is just on or quote unquote just on archive, if it has like a thousand citations, it's a legitimate paper.

34:49 · Doesn't really matter where you published it." And I agree with that. I I do think it's kind of sad that I I've heard like grad students have to do um like how to Twitter uh seminars to each other just because it's so important for publishing these days. I mean, this person is just just reflecting the sentiment at the time. That's right. But it actually affected you so much that you stopped work on it.

35:10 · Yeah.

35:10 · The sentiment also came out of some of the research, right? Like the original BERT paper was trained and towards the end of the paper they're like okay throw off the last head train specific iterations for uh you know extractive summarization add ahead for this like you should do task specific stuff. These are like the authors that wrote attention wrote birth telling you this is what you're meant to do. And like the training tests were also very odd that like the we know that the model overfits to this weird mass language modeling throw away this part and just do specific models, you know.

35:40 · Exactly.

35:40 · And like you know we had to try come up with all clever ways of like attention and pointers and and so on to actually get the neural network to be able to do all these tasks and then some of them were better than state-of-the-art some weren't but were like but it's still in one model. I thought it was really cool.

Open-Endedness and Evolutionary AI

35:55 · I was going to move on next to Tim and open-endedness. He was head of open-endedness at Google. That's I don't know what that means. Uh but he did a lot lot of talks. Genie 3 is one of the ways that rainbow teaming. Yeah.

36:06 · So I I I first saw him at speak of ICLI. I first saw him at ICL when he talked about open-endedness. He he's done a few talks. Can we define what is open-endedness for people who have never been exposed to the problem? They're like, "What do you mean?" I thought the only goal of AI is to optimize against benchmark or sketch.

36:21 · That's right. Yeah. It's a it's a fuzzy fuzzy term because there's so many different instantiations of open-ended uh thinking, but uh one way I often describe it and and certainly uh Tim and Jeff Glon would be even better at describing this, but it's a suite of methods that is more inspired by evolution than uh very specific rewards.

36:40 · Uh so in that sense it thinks more about environments about co-addaptation and so in concrete example is in the cyber security and LM safety space where you have one LM that tries to attack another LM to do something unsafe and now the

36:57 · environment is the two having a conversation and now they co-adaping right they're like one makes a better attack then the first one inoculates itself somehow like uses that as training data makes it so it's harder to say something unsafe based on that and then as the attack stops working the attacker now tries a different angle right and that's why it's not just red teaming but they're called sort of rain don't tell me how to do things let me just figure it out myself that's right think about the environments that you want to use think about the rewards at a high level that you want to uh inspire towards uh and

37:30 · then let the eye try out many more ideas in this interplay between sometimes humans but also sometimes other AI agents yeah I actually worked l um open endness into a sort of model that I have been sort of working on. It was the keynote for AI uh where you start you know we have the token loop we have the agent turns and then we have goal and I feel like the way that you're describing open ended is still somewhat of a goal like like please attack this uh other agent

What Happens When AI Chooses Its Own Goals?

37:59 · but yeah you set rewards you set the environment the loop that makes the other loops is what if the agent can set its own goals and is it is that open-endedness like like you don't give it a goal just like be a sentient being and maybe sentient is a very loaded word but just set your own directions. What do you think you should do?

38:18 · I I love this direction. I think this is one of the 10 spaces of intelligence uh that I lump under metacognition and thinking about thought. Okay. And it's an interesting one. Whenever people say, "Oh, AI is like this is, you know, it's going to stop from here. It's not going to get that much better and blah blah blah." I'm like, there's so many different spaces of intelligence that we haven't even started exploring yet and hence have made very little progress on.

38:42 · And there there is kind of an interesting uh connection to economics and capitalism. Like it doesn't make sense for a company to build and spend billions of dollars building a model that instead of following the rewards and objective functions you gave it may come up with its own subjective functions and its own goals, right? And then imagine you're like, "Okay, I spent billions of dollars now go develop this new battery uh material for me and answer all my emails."

39:09 · And it's like, nah, I think it'd [laughter] be more interesting to evaluate the molecular composition of the atmosphere on Jupiter. You're like, that's not what I paid you billions of dollars for. Like, and so no one's working on that for good reasons. And then also, understandably, it's not. It's not it's not useful. And it could get a little bit weird, right? What if the eye actually does start to really have thoughts on its own? Uh, and what if we don't like those thoughts, right? And so it it requires a whole different way of thinking about it.

39:36 · I had a great conversation with a good friend of mine Sam Gershman who's a neuroscience professor at Harvard and like we just jammed on this a little bit on like what are sort of the best meta goals and you know I do think like knowledge seeking is a really good one. I'm currently thinking also about like uh the ultimate measure and unit of intelligence broadly construed and I finally have some still too early to share it. Um it's not haven't fully baked the like like some replacement for IQ.

40:03 · IQ is such a terrible definition. It makes no sense. Yeah. Elos are terrible too because it's always just like me versus others.

40:10 · Okay.

40:10 · But like you can be intelligent and not constantly compare yourself to others, you know, like and so yeah, there's no like in fact a lot of these definitions we have which I briefly mentioned my book to these definitions create sometimes explicit and sometimes a more implicit enthropic bounds. Not this to the company anthropic, but just like this idea that your intelligence is like getting 100 out of a 100 questions right on this IQ test. Well, if that's your definition, then you can only be at 100 out of 100. Where do you go from there?

40:39 · Right? So, you see a lot of these uh uh benchmarks that people are working on, they increase, you get close to human, maybe some slightly above human, and then it's flat. It's like cuz that's your if your definition is only that so tight to humans, you're only going to get to just slightly better than that.

40:56 · So, I think metacognition is a great example of that where we're not even yet allowing the eye to think. We're not working on it very much and hence there's very little progress in that area.

41:06 · Yeah.

41:06 · Well, we've interviewed Endon uh which I think uh has been working on the most open-ended uh benchmarks which is just real world uh money. Uh arguably telling an AI to profit maximize is a bad idea. [laughter] Yeah, they are doing it. I mean, I do think you don't want that super like you don't want a super intelligence to have a ton of access to all kinds of tools

41:29 · and and so on and then just give it that without some very careful reward engineering cuz it's like I mean you know I just buy a bunch of defense stocks and I start a war I make money like it's just like it's a tricky tricky situation right you just buy a bunch of stuff short basic goods for people and you create some weird famine like issues like yeah there's a lot of constraints you should put onto uh trading system.

41:52 · It's a fun measure though cuz you know the bounds are very capped to where we're nowhere close to them. Like in in Endon Labs the the model is like oh it's Saturday you know maybe I just closed the store today. Someone someone's off.

42:06 · It's okay. We'll just close the store using [laughter] but yeah no I'm not I'm not arguing against it. Just like as you get more and more intelligence you want to be more and more careful with that as like an open environment because the environment then is all of the earth.

42:18 · Okay, for recursive not strictly necessary, right? Because like if your goal is Eureka machine that like invents the other things, then like actually just solve you know the the science solve machine learning research and discovery and all these things eventually. So our goal I haven't really I I don't talk about it that often because it is a few years out but our goal is once you have a recursive self

Superintelligence for Science

42:41 · super intelligence you then want to apply it to the most important problems and I think a lot of those are in science and technology and broadly construions and those inventions in you know physics to create better cheaper energy with fision or fusion in in chemistry and to create better materials and better batteries and uh better solar cells and and and so on in biology there's so much

43:04 · like I think soon to be low hang lower and lower hanging fruit because of AI because of protein and generation not just folding but actually generating new proteins like we did in and progen many years ago like so much positive impact to be had if you take that super intelligence and you apply it to science I do fundamentally believe that there's a lot of approaches though you're not the only team trying lab trying um you know there's like a lot of especially the physical sciences as well and that's good yeah I do actually think that phys like the reason we only doing it in a few years is that it's a little too early right now.

43:34 · Robotics is not quite there yet. The AI is not quite there yet. But I'm fairly confident in 3 to 5 years, all those constraints will be gone. And then applying to real physical robotics experiments and so on, like true robotic process automation, not the traditional sort of RPA sense, but like actually having robots run experiments for you will be totally there. Yeah, it's going to be great.

GPUs, Compute, and the Limits of AI Takeoff

43:57 · Just to call back to something that you said early on about slow takeoff. You said that like well really the the sub the substrate that is limiting factor is let's call this chips uh and semiconductors and all these things and you have raised funding for that and and uh you are you're investing a lot on that but have you done the math on like is it even achievable and like what what is the uh industry concentration needed in order to achieve like scale?

44:20 · I mean right now we know that like roughly like uh you know a thousand GPUs cost quite a lot of money right if you wanted like tens of thousands of GPUs you're you're talking billions and billions of dollars if you say like one GB like 300 is like

44:37 · you could eventually create models that are you know on that substrate like are close and similar to human intelligence like and you want like you know thousands and thousands of uh AIs to think about really hard problems uh in a similar fashion to humanity like yeah that that's you know that's a lot of money you do the math it's like a lot we don't have that amount of money right now anywhere to like build that um now

45:01 · obviously things can get more efficient you will have I think soon better algorithms that won't be in better hardware that won't be as energy hungry um and so on our human brain does quite a lot of flops with much less energy 20 watt that's exactly right yeah that's the number often is quoted um and like I think more inventions will happen there that then will accelerate the takeoff even further.

45:25 · One thing I always try to reconcile when talking like with Neolab founders is like you're kind of fighting bitter lesson all the time. You you have to show initial progress then you unlock the next tier of funding then the next year then the next year which unlocks larger model categories like fundamentally is that true like are you fighting bitter lesson or you will will we have a way in which like no we're changing the slope in some fundamentally different way I do think we are changing the slopes in in fundamental ways by making AI much much

45:55 · more efficient both in terms of the training as well as the inference I think we will when you allow AI to do the work that it takes other labs thousands of people and years to do. I think we'll be able to get it down to weeks and that will be much much cheaper um and hence you know more affordable accessible to others and so on.

46:14 · Yeah. And you've shared initial results on on that uh which like conveniently OpenAI has also done to their 5.6. So we can talk about it now.

46:21 · Yeah.

46:22 · So let's recap what you've done.

Recursive's Results: AI Beating Humans and Their Agents

46:24 · Yeah.

46:24 · So maybe yeah just a quick recap here. uh we built uh this this uh system that isn't the full even the full RSI system in its glory but it is a first baby version of this and then you know we don't want to just have it internally and and not show anything and you know just show some people of what's possible um and so we basically applied this to these three different tasks one it's nano chat uh by my friend Andre Kapathy um uh just like train a small language model to get really low bits per bite

46:54 · and you know like hundreds if not thousands of people uh used both their agents and themselves to try uh to get to that uh and then they got to 937. We literally took our system and got to a much lower uh bits per bite uh much much faster within like I think less than 2 days.

47:14 · So we took this thing applied our system to it and less than 2 days later uh we have we outperformed every human and their agents uh in in uh have ever worked on this same with nano GPT and then we're like well let's you know apply it to something that's even more relevant uh to to real people and to the NVIDIA ecosystem and applied it uh to uh

47:35 · solicen and maybe you can scroll down to some of the uh images that are kind of fun to see but yeah you know like one you see it's actually made some real inventions that weren't aren't just sort of hyperparameter tuning like actually inventing hashts and so on is is quite clever. We have even better results.

47:51 · What do you mean inventing hash? You didn't invent hash.

47:53 · Of course, we didn't invent like hashts in [laughter] the grand scheme of like a hash table is like a super basic primitive in in computer science. to use it uh for language modeling in this scenario inside a transformer and so on and to actually combine these ideas and put them together that has then eventually also been invented but there's a knowledge cutoff and we did actually check that it didn't have access to that externally um we talked about this a little bit if you scroll to the next figures you know this is also an interesting one in that when you start from a really basic poor like

48:24 · vanilla transformer then we still outperform all build the community together. But if you start from a human seed of an expert like Andre, then you get even lower. So the human seeds from which you start do do still matter. Uh so that was an interesting kind of insight in my eyes on this and then as you go like you know how long does it take to to actually get to these models to get to similar performance? It's much faster.

48:50 · And then a similar thing happens with uh the speedruns here where you know people have worked on this for for quite some time uh and the model still was able to train a model more quickly.

49:02 · Why do we care about it? Well, speed of training is part of the equation of the cost uh and ultimately you want to have the most intelligence per dollar, right?

49:11 · And so speed and quality are big parts of that. And you know the the um yeah the way I put it is for for people who don't understand they look at the chart they're like cool what does it mean? Uh you know if you have like a billion dollar cluster and you can shave off 10% that's $100 million.

49:29 · That's exactly right.

49:30 · How much is that worth?

49:31 · Exactly.

49:31 · So when you when you look at like the kernels these kernels Yeah. For for the non-experts like uh these kernels are off like used in basically all the models. Every time you use an Nvidia GPU you interface with that GPU through these kernels. And so here you see uh the leaderboard best and when it's recursive um and it's basically there only a handful of kernels uh in this whole benchmark where we weren't the best. Um and so to me this is like really exciting because it makes it it just showcases what this can do. And again these weren't like we didn't like spend months or years like developing.

50:05 · In fact, in particular for kernel uh uh CUDA kernels like we don't even have really deep CUDA kernel experts in the team and our system that that's the beauty. The system just did all of these things. We didn't invent this and when we open source and and release uh things in the future and models in the future like it won't they won't be the best in their you know category or class or whatever because we're so smart but it's because we built a smart AI that does it for us. Do you have anything that you've learned from how to guide good auto research? A lot of it also builds on human background, right?

Reward Engineering and Auto Research

50:36 · It's not just as simple as just, hey, go optimize this. Um, but we do see it again and again, right? Like some of the Erdos problems, frontier math is being solved by people. And when they do a write up, they're like, "Oh, I'm not a mathematician. I have no background in this." You know, I I saw some tools and I I made it work. while you're watching the World Cup, you like disprove some conjectures and going on um [laughter] any projector was like to summarize uh tips for good auto research versus bad auto research.

51:05 · How did you build the recursive?

51:06 · Yeah, so without giving away all the all the secret sauce, um maybe some things that are probably obvious to the experts but might still be interesting to some uh folks is like reward engineering is one of the most crucial bits uh especially uh in order to avoid reward hacking. Uh, so you have to be really clever about avoiding because as as your AI gets better and better, it will get better and better at finding weird like special cases or counter examples and and things like that.

51:34 · And so I'll give you an example like when you ask to like make these hundreds uh lines of code faster and you know how do you define fast? Well, you have one line at the beginning that says start your stopwatch and one line at the end, end the stopwatch and then, you know, tell us how much time uh progressed. And so, well, the simplest way is you just put that line that ends the stopwatch, right? You know, at the start, and then boom, it's now faster, right? So, this isn't like this like super evil AI. It's just like a very simple dumb reward hack.

52:04 · Um, and so you have to just very carefully think about all the different angles there. And then I think the longer time horizon the tasks are, the harder it gets and the more interesting and clever you have to be to still use these kinds of ideas for it. But yeah, I can't give away too much there.

52:22 · Seems like rubrics are taking a good spot in that where for unverifiable domains, you have rubrics. You have a model breakdown judges criteria along the way.

52:31 · It's a form of verification once once you got everything. I said this a long time ago. That's why I've never been that impressed that AI can play games cuz I'm like obviously anything you can simulate and or verify you can have infinite training data or and hence like AI will solve it eventually. I've been looking for games where you can do auto domain uh distribution. So this is a game that nobody's trained on cuz it's a new game uh and you can start gaming, you can start to play. So I've been basically building this uh and clone this in person and it's just been selfplay.

53:00 · I've had about a billion positions evaluated and uh I I wanted to do the alpha go thing of selfplay until you get better, right? Like which which is like this is not even LLM AI. This is just classical game AI. Uh but uh I think that the and but I set GPT 5.6 to auto research it because like I don't want to hand handle any of this. I expect you know the alpha go process to be like fully in the weights by now. It is not.

53:27 · It is actually it it like immediately leveled off very very immediately until I human play tested it and then I I like called out obvious mistakes and then they were like oh yeah okay and then it just dropped [laughter] and like you know no amount of like think different think more creatively give me eight different directions and no amount of prompting got it interesting like you had to like against a human to do it.

53:52 · So I I I mean that that was that was my and by the way Bean always wins if you if anyone watches reads Enders game and you you put quite a bit of work into the guide for the AI like so the game basically you know you stack tiles there's some rules you want to capture the most area you you have like a whole 50 pager on every rule you fed that in it couldn't it couldn't handle it that yeah you know it's so funny that this reminds me the claim territory and stuff of a paper we did in 2018 called the AI econ economist.

54:22 · If you search for AI economist Salesforce, we had a video we can play. It was an economic sim. Um, so the idea is you have all these economic agents. They just want to optimize their own utility function, which uh is, you know, collect resources that make money.

The AI Economist and Simulating Entire Economies

54:38 · Uh, and you can sell resources like wood. Uh and then uh over time as you collect more enough wood you can build houses, you can trade with other agents and you can basically use the houses then also to block off resources from other agents. So there's like competitive play and strategy and so on.

54:56 · And the point was that we actually wanted to understand what is the best way of taxation and subsidi subsidization to optimize an economy.

55:09 · And this this kind of research has not yet had its sort of GPT moment. But I believe that countries like Singapore and others should and will eventually use this to instead of doing like uh basically partisan politics and like special interest politics of like who donates the most to your campaign and stuff, you say, "Well, here I want to help the middle class or whatever you might say is your objective as a politician." And then people say, "Okay, well, how do you want to do that?" And it's like, "Well, here's my fiscal policy. here's how I look change the taxes and pay these people and so on.

55:39 · And then you can actually put that into a simulation and you run that that attempt from the politician against billions and billions of years of other strategies to try to achieve the goal that they set out to do. And then you can say, well, if that was your actual goal, then here is, you know, billions of years of a strong simulation that would suggest that you try other ways of doing it and maybe this the taxes and so on and this these tax brackets and so on. This is how you avoid gaming because these agents also try to reward hack to not pay their taxes and so on.

56:11 · Uh I thought this paper was super interesting. Unfortunately, similar to the first paper on uh prompt engineering, the economists are like we don't know any of this math. It is like it's not even math. It's just we don't trust your simulation. It's not about math.

56:29 · It was I mean they just desk rejected the thing. It's like it's like they like they didn't even give us like clear re like clear sort of signals. Um but like the world of economics unfortunately doesn't have proper Yeah. It doesn't have proper uh benchmarks. So you cannot be like eventually why did neural nets win? Not because people loved it like they had all kinds of beautiful integrals and graphical models but it just worked better. But in economics it's hardism versus Yeah. Yeah. Yeah.

56:55 · And I do have a bit of that ecom background where like there's a lot of physics envy where you want to write the general equation for the an economy uh versus just simulating it and using an evolutionary approach.

57:08 · Um Viv is thinking exactly what I'm thinking is didn't we have the GPT moment with small Hello world June just announced I don't know if you you guys are involved simile there similly that they've I wish we're involved we're not yeah I had a couple simulation based talks at AIE uh so if people want to look up what the state of the art there a lot of people are actually exploring this proven yeah we also had podcast with Mikuel

57:33 · Parkin from Shopify um who is using simulation for e-commerce nice which will will simulate like your trajectory and like predict what changes you make to your e-commerce journey will affect in your sales and all those things.

57:44 · I love this. Yeah, it's really hard to simulate an entire economy, right? You have to make some simplifying.

57:49 · If everything's very expensive and I'm just like, am I going to do this 8 billion times? Like, come on. But, you know, I [laughter] feel like countries like Singapore that really want to just objectively do the right thing, have very technical leadership and so on, like they might actually like eventually really try to simulate their economy.

58:05 · And obviously you have to make some simplifying assumptions. But it gets really interesting because you can also say if your assumptions are such that all people would work hard if you let them uh and you know they have the free and then it turns out you have to make assumptions like well some people's utility function of like how many hours in a day do they want to work are different right and then you can start to disagree on the assumptions that go into the into the simulation and then once you say all right now we agreed on those or we have different views of what

58:32 · people are like at different you know distributions and whatnot then there are different outcomes based on your goals and then of course humans should choose what are the goals. In our case it was productivity multiplied with equality which you know has some issues but it's like not totally unreasonable.

58:46 · Yeah.

58:46 · Just a comment on Singapore because you probably have no idea but uh I am Singaporean and I've uh been involved in uh the Singapore AI Council for making these things. Uh the main reason they won't is because they're very conservative. Uh and you know I I kind of view it as they you know there's a founder country. when you start a country or you start a company and it's founder led and you can do whatever you want because it's your country and then there's manage like professional manag managerial class which is now that's that's what Singapore is.

59:13 · So they want they always want to see someone else do it first and but like everyone everyone in the west views Singapore as like oh it's a small country you can do whatever the hell you want like Singapore doesn't do that [laughter] so like someone else has to has to take the charge there. I'm just going to do one question on the simulation thing and then I don't know we can probably move on. Uh mode collapse right like you know LLMs do not model the decision of humans. Uh spamming it out 8 billion times is not going to help you model humanity. What do we do?

LLM Simulations, Personas, and Mode Collapse

59:40 · I do think uh you have to be clever about prompting each one individually and and I think that will help you kind of get stuck into different different modes and in a weird way people also get stuck in different modes you know like there's a lot of people like don't teach an old dog new tricks kind of thing like once people are stuck in their ways the older they get the harder it is for them to to think new ways and there's this I think uh comment I forgot who said it but it's like everything that was invented uh before you were born is natural.

1:00:12 · Everything that is invented when you're 20 is cool and everything that's invented after you're 60 is like unnatural and an abomination and kind of weird.

1:00:19 · I feel like that's, you know, it's it's true for a lot of people. Like it is a fashion and um I think people will do it. Uh Tencent had a billion personas paper that gives us good data set for prompting uh simulations. If anyone's looking into this um on on the podcast, they just had like you are a 30-year-old grocery store clerk, you are a 50-year-old professor, and then just do a billion of those. Checks out. So then you just use it. I'm I'm kind of shocked how how well a lot of these things actually do map to ultimately similar statistics to real experiments.

1:00:51 · Yeah, I think it's also good stuff for people to try that when they get into research, right? like we've seen train a model only on data before a certain date and see how well it extrapolates out do the same thing right so um see do people code more with better coding agents can a model that hasn't been trained on this figure that out without web access right extrapolate out test these things yeah right you know just today I think LM Marina published a interesting result

1:01:17 · where they basically were able to create a model now to predict your your ranking wait uh based on what input your model I guess you give it your model and it predicts the ELO score I See? Okay. Surprising.

1:01:27 · Yeah.

1:01:28 · I mean, their whole play on kind of is like, oh, like we we help you compare uh these models. Yeah.

1:01:33 · Yeah. I mean, this team, they they've done a lot of work and obviously they have the most data to do this, so why not? Yeah.

1:01:38 · Yeah. Brilliant.

1:01:38 · When they were coming out of UC Berkeley, they not only had LM Marina, but they also introduced a routing project that would route based on LM Marina. And I don't think that actually ever came to pass. And I'm curious why. I I never got to ask them about it cuz like it's was like, oh yeah, clearly that's your business model. You will become a router. and they never became a router company.

1:01:57 · Weird. Um, so that I'll just put put that out there. We're going to talk about GP 5.6 uh self research thing if if you have anything. I I should should also mention in your list of uh you know kernel optimization um and and on the track that you spoke at, we also put way um Chung Yao from WOO who was also number one in the parameter golf challenge uh which is an open AI hiring uh challenge which is also a very similar story.

1:02:23 · And I think we're going to just see this all the time where humans optimize a thing a lot and then some AI team comes in and just becomes number one. [laughter] Yeah, 100%.

1:02:34 · I think the other [clears throat] interesting thing with stuff like these challenges, right? So this is training this the best model that fits into 16 MB. You can always look through the changes that are being made and the small gains people have, right? Like you're getting less than 0.01 of a increase by adding some change attention MLP stuff. And then you look at your charts where you're like, okay, we just let model loose. And then, oh, we had a little stagnation. Nope, another drop.

1:02:59 · Nope, another drop. And that's what it is where it's like, um, what did you guys add? You didn't add um hashts, right? It's not like you invented hashts. You you did another three iterations of these that unlock, you know, a few step functions that people won't just find.

1:03:12 · Yeah.

1:03:12 · One one thing to close the loop on overrid along the way of of trying to optimize, we found 30 bugs in the in the harness, right? So like like every all the research that went in before we found the bug, we have to we have to throw it away because it's contaminated, right? [laughter] Yeah.

1:03:28 · Which uh you know just to your point of reward hacking like even in this very simple game, we found the bugs.

1:03:34 · Yeah. Yeah. It's crazy.

1:03:35 · Uh and so and symmetry is a very good way to check which is like you change a position of things where where it shouldn't matter and it does matter.

1:03:43 · That's a bug. M and which which has come up in like let's say multiple choice like GPQA type questions where like yeah between A and C if it's a multiple choice question if you change the order it should not matter but it does right [laughter] so that is like okay you know models still prefer the end of the output right not trained well a long context model the last bit of tokens are what you care about oh no the answer the answer in that era of LM research was more simple they just memorized like the answer to this question is a.

1:04:14 · I don't care what the what the answer was. It's it's just a like. [laughter] Okay. So, I think we can move. Um the last bit that you did there, the kernel optimization is probably the one that you can feel the soonest, right? So, yesterday OpenAI announces that self evolving having their best model work on optimization kernels. They're a lot more efficient and they can cut cost 80% on, you know, Luna and and Terra. I guess questionwise, you laid out a bit of a road map. There's a lot about bio, a lot about physics. What do you think hits first?

1:04:46 · Like what what are the next two years? What's attainable? Now you've mentioned robotics towards the end, but what do you start with?

Recursive's Roadmap, Agents, Search, and Finance

1:04:55 · We like very explicitly will not start with any of the physical sciences. For now, we'll start on AI for AI research.

1:05:02 · And so the AI for AI research has I think still a lot of room to grow. Uh that's both in terms of making training more efficient and more automated uh as well as making inference more more efficient and potentially local on your laptop like there are all kinds of interesting angles that have not been explored uh that well what's the um go deeper on the local stuff because I always feel like it's the most inefficient form of AI training.

1:05:32 · Yeah.

1:05:32 · So there's training and inference.

1:05:33 · I I can't go into too many details, but like yeah, I think there's just like so many angles, so many different compute substrates that that have not yet been explored either for training or for inference. Great. I don't know if you have any other comments on the on the the other stuff. I I would say the other thing where like there's the sort of inference in the uh optimization in the small but then also there is overall

1:05:55 · latency latency end to end under conditions of load uh which [snorts] is like a very different thing which is the basically what they actually ended up doing that is a different domain of auto research than uh then I would say like improving the kernels I think the other thing that I always think about in terms of automating or improving performance end to end is how the harness plays into it. Mhm.

1:06:16 · Uh particularly now when we say harness, we also mean sandboxes, right? I'm curious if that is a blocker for you or like you know like how how the agent calls out to tools basically. The number one tool all these agents use is web search of course uh which makes sense.

1:06:31 · Um and then I do think the harness is nice to optimize for because it's just so easy, right? It's just language. You look at it, it makes sense. You can iterate. You don't have to train a massive model for like you know a lot of flops. uh to get to the next state. So big fan of harness optimization.

1:06:49 · Yeah. [clears throat] But sandboxing is fine for you.

1:06:51 · Sandboxing is also super important. Um and then of course like reward like hacking and alignment I I think are are super crucial.

1:06:58 · Okay.

1:06:58 · Um just want to mention web search. You happen to also be CEO of web search company. Do you use u.com and do you use others? Like uh should should the rest of us be using you for web search? When I say you it's like very funny. It's like you the person and you the company. So yeah, it's mostly now for uh developers and agents. Um it's less for like consumers or proumers.

1:07:18 · So if you're a company and you have agents and you know to be honest for a lot of companies who are now moving to open source all of a sudden it becomes a conscious choice of like which tools do I give access to my open source LM and

1:07:34 · you know the first choice uh has to usually be around web search and then once you get to scale.com becomes like an obvious choice because of all the uh different benchmarks and so on that we pretty much all dominate the per frontier of and then in terms of just uh general people like consider new to the space considering different options if they're building agents I think there's a hierarchy right a lot of people will have heard of exa will have heard of parallel uh and u.com is is like in in

1:08:01 · that mix of like providers there beyond that there's like the general sort of web scraper companies like fire crawl and uh browser base and then beyond that is like the commercial proxy companies like the bright datas of the world is that an accurate waterfall of like hey you're building an agent these are your options Yeah, certainly like yeah the like the bright data is like lower in the stack sort of on the proxy network side of things.

1:08:24 · I think like in terms of like content and uh getting crawled content like you can do that on.com too and then there's sort of higher and higher levels of abstraction and like combinations of different data sets that we do like in finance for instance like we're not just like 2 or 3% more accurate but like 20% more accurate than others at faster speeds and lower costs like finance in particular is kind of like like not even close.

1:08:49 · Uh you can actually go to e.com and there's some like statistics uh and benchmarks that you can if you scroll down. So there are like different different data sets and you can kind of look at you know different uh competitors comp uh and yeah the fin search is like we're up there like close to 90 and the next closest thing which is way way slower um uh is yeah just like in the 70s instead of close to 90.

1:09:18 · Yeah.

1:09:18 · Yeah. Yeah. Interesting. I get my my next focus is AI and finance. So this is like actually literally going doing a doing a conference in New York uh just for banks for for this stuff.

1:09:28 · Finance is kind of like the next thing to break out after coding. It's cuz it's somewhat verifiable like prioritizing spreadsheets. Obviously there's a lot of data out there that's all public and you can crawl it and all these things.

1:09:40 · What's what's like hard about the finance domain if in your you know that you guys have solved? I mean of course like one thing that trips up a lot of people is just uh you know leakage of of training data and so on. You think oh how do I you know you want to ideally predict the future before it happens.

1:09:55 · You want to mask the future.

1:09:56 · Yeah.

1:09:56 · Well yeah mask the future in your training data but there's all kinds of leakage. Like I can tell you when I was teaching at Stanford the NLP class um like so many dozens every year said I want to use data set X like Twitter to predict the stock market. And they all like showed cute little things that somehow look like they were working.

1:10:15 · It never loses money. How come?

1:10:16 · [laughter] And yeah, and there's always some kind of data leakage and so on. And it's just like wasn't as easy as they thought it would be. Um once you once you fixed all those issues. Um but no, I I agree with you. It's a very sensible application of of AI. Yeah.

The Upper Bounds and Spaces of Intelligence

1:10:30 · Yeah.

1:10:30 · Amazing. You know, as a as a writer, as a thinker on on on these things, I love MEI categorizations. Mi is mutually exclusive, commonly exhaustive, something like that. And so if this is a mey list of intelligence not it is very okay well sorry there are all kinds of overlapping in fact uh if you want that kind of list I think the three principal components of intelligence uh are prediction which is

1:10:54 · mathematically uh quite uh similar to compression prediction multiplied with actions multiplied with goals those are the three principal components I think all of these 10 spaces are combinations of those three in specific dimensions, if you will. And the reason I call them spaces is that each space has many subdimensions.

1:11:17 · And what I try to do uh actually this is just a a side quest almost to the initial goal, which is to think about the upper bounds of intelligence. And you know, everyone's like, oh, it's exponential. And it's like, well, exponentials at some point have to flatten out, but where do they flatten out when it comes to intelligence? And that led me on this whole like it initially it started as a tweet and then it was like a blog post and now I'm like at 50 pages and I'm still not nowhere near your second book. It's the second book basically. Uh and so the la in my first book Yoga Machine I just kind of allude to these 10 at the end.

1:11:48 · And I just to give you a sense like visual intelligence is sort of the easiest one to talk about and I fleshed out the most already for me in my head. And so human intelligence has basically binocular vision and we have two eyes. We have a very narrow band of the electromagnetic frequency spectrum that we can really observe directly ourselves. And so when you think about the upper bounds of a visual intelligence, one you should go into like you can have like millions and billions of sensors.

1:12:16 · At some point you get to problems of how far are these sensors away from each other such that the speed of light to communicate the content from all of them cannot like get to a central brain to actually process

1:12:32 · uh the visual intelligence right and so now you're thinking in along the dimension in the space of visual the dimension of numbers of sensors okay the upper bounds are quite literally and figuratively astronomical and we are super far away from any intelligence that would have this many number of sensors. But then you go in the next dimension which is the frequency. You can go all the way down to gamma rays and you can start to try to observe and you get into basically the upper bounds or I guess in this case lower bounds uh or upper bounds in terms of frequency is basically quantum uncertainty.

1:13:04 · Like you just cannot observe certain particles. And now imagine you had millions of sensors that can see all the way down to the like subatomic level as far as physics will allow us to and then all the way down to seeing like gravitational waves and now you have millions of those sensors. So that's another dimension is sort of the frequency and then yet another dimension is like how many categories of things could you memorize and classify differently.

1:13:30 · We know for humans right there are certain things if you have more terms for it you'll have a better visual description uh for them and like you know animals that don't have like gorillas maybe have like 200 words to assign to certain things mostly visual things and so human perception is quite special in that sense in terms of classifying all these different physical objects. So these are just like a very simple example. If you go uh to knowledge, right? Then it's also like the speed of light cone around all these sensors.

1:13:59 · And so they're all connected like knowledge is connected to visual intelligence. If you think also not just visual but sort of perception intelligence just like cuz it doesn't have to be just what we can see. It can be again wider range of electromagnetic frequencies. Then you have language intelligence which actually recently changed to more communication intelligence because it's more like language has all these different enthropic bounds. Humans can only comprehend and know so many terms in our long-term memory, right?

1:14:28 · Our vocabularies are somewhat restricted and the active ones are often even smaller than the passive vocabularies of things you can understand. Then uh language is ridiculously inefficient when it comes to trans basically communicating different types of information and uh transporting sort of different bits like human language is serial. Obviously another bound on communication intelligence would be to communicate in parallel but neither will our tongues and mouths work to have multiple like streams in parallel.

1:14:59 · Neither can we understand some women slightly better at like multitasking than some men. Um, but like most people can only listen to one conversation and truly understand it.

1:15:09 · There's no way that like like in terms of communication intelligence, a a true upper bound is one in terms of how many like knowledge, how many sequences of communication could you in parallel sort of uh process, right? Then of course you have uh like how long are sentences? We only have so much in our working memory and hence lang human language has these fairly simple sentences with maybe 40 words or so on average for a sentence.

1:15:39 · That is also not a an upper bound that makes any sense to an AI. And then uh yeah like I I can go on and on and on.

1:15:46 · Each of these has tons of interesting upper bounds and it teaches us a lot about how much further AI can go when we start thinking about these upper bounds and then realizing how far in many cases we are from the bounds and you get to basically physics. Now I'm I didn't study physics the way I studied you know AI and and computer science. So I'm learning a lot which is why it's kind of fun.

1:16:07 · Uh but a lot of these like how much and then when it comes to for instance knowledge like how much can you store how many bits can you store or bytes can you store in like a certain amount of mass and volume and you get to all kinds of interesting amounts like Beckenstein bounds and you start thinking about black holes and like and then speed is like an interesting one too in that it's sort of connected to all of these but speed is also kind of its own thing in the sense that all things being equal.

1:16:36 · If it takes you an hour to know if the 2 plus 2 equals 4, you're just not as intelligent as if it takes you like a millisecond, right?

1:16:45 · And then like all of these connect to survival and replication, the last one, it's like, yeah, if it like trees are really really slow, so we don't even consider them that intelligent, but if you speed up some videos of trees and they're trying to find stuff and so on, they're not as dumb as they look, like not dumb as wood, you know, but like And then obviously like different things.

1:17:03 · Um, so that overlaps with speed a bit.

1:17:05 · Exactly.

1:17:05 · it over like all of these things kind of overlap like you talk about natural language connects everything right you talk about your knowledge you reason and then you communicate that you talk about things you see so they're all kind of interconnected but uh I think they're usefully studied individually the same way that uh the best analogy I could come up with so far is energy right you have either kinetic or potential energy and in theory you could study all of physics it's just do you want to study kinetic or potential energy but in practice it's helpful to study

1:17:37 · and electrical engineering subs are just like different types of energy but it makes sense to study them individually and so I think physical intelligence maybe I'll just do one or two more of these like if you had full control over your own compute substrate and you had full control over physical matter you should be able to create any atom you want like we can actually fun fact you can create gold atoms from from just raw raw protons and like

1:18:07 · electrons and you smash it together 98 of them or I forget.

1:18:11 · Yeah.

1:18:11 · So like the thing is though it costs an insane amount of energy and it cost you way more than you know you get like a few atoms of gold right and so like uh it's it's not viable but if you had better control over your physical like all of like physical substrate that I think is yet another space of intelligence because it relates to your own comput substrate which you can eventually also improve.

1:18:32 · Um, social intelligence is an fun one in the sense that not in like our necessarily just ethics and morals which are obviously important too but in some sense you can try to define upper bounds of how much can you communicate to how many other intelligent entities and be able to have an expected value over how much you can transform their internal states and their actions to in order to align with your goals, right?

1:19:00 · And so like you can actually write like a fairly like straightforward equation that defines kind of that level of social intelligence. And that is what humans and ethics and morals and religions and so on have been trying to figure out for for millennia. And in all of these cases, we are very very far away from

1:19:20 · the upper bounds and that should be very inspiring and show people that we can still do many many years of of AI research [laughter] that yeah there there's a lot here that this is a general philosophy of intelligence which is uh very interesting. I do you have any comments or I think it'd be interesting to gauge what you think like baselines are where we're at now. What's low hanging fruit?

1:19:44 · What's far off? What's what should people put their work towards? What should they focus on? You know, I think I think it's clear that like natural language again is the most interesting manifestation of human intelligence. Um, and hence like a sub field of AI. I'm excited that many people are now like in agreement with that. when I started in 2003 to study linguisty, computer science, NLP, like it was like a weird niche subject. Um I I do think there's a lot more juice because it how it connects to everything else and how you know civilizations are built on language and knowledge and all of that.

1:20:16 · Uh I do think physical intelligence will come up. It's kind of interesting. I feel like robotics is kind of in the machine learning state of things where you just look at like how does human how does a human decide that this is a positive sentence? Oh, I do. So like robotics is a lot of well we have five fingers and like try to do this.

1:20:34 · No one is yet working on like the super intelligence version of robotics which is much more similar to like the T1000 you know and from the Terminator movie which you know obviously let's not build actual Terminators but like uh I think like this idea that you should be able to shape shapeshift um like into any kind of shape is like that's sort of a super intelligence version of physical intelligence. We're like not even no one has even really started yet. There's some really cute little research where you can move some magnets through like some grids.

1:21:02 · Uh but like uh yeah it's very very there's some I think MIT has every year every two years they have like some self assembling robot robot thing which like that would be it but it's very very primitive.

1:21:15 · Yeah.

1:21:15 · I'll just get a touch on like what are the main dimensions of creative intelligence. Creative intelligence uh is of of course again connected to all of these uh a lot of it uh connects to metacognition and that you need to be creative in how you choose your goals.

1:21:30 · Okay.

1:21:30 · Um that is I think one of the most important thing for a human and their their lives and careers and their happiness is choosing your goals but also for any kind of intelligence. Then of course there's creative intelligence in terms of just finding creative solutions to existing problems right like I say like we want to make this product cheaper like find some solution

1:21:50 · to it right and just like finding existing paths but then there's kind of the most interesting bit in intelligence is when you move not just out of the convex hull of known sort of ideas but out of the hyper cube of known ideas which we know uh so like hyper cube is like a mathematical concept right and we already know that AI can like known dimensions yeah Like exactly. So like AI is already good at hyperubernet like if you give it like a bunch of examples of brown dogs and uh pink cars.

1:22:15 · AI will still be able to generate an image of a pink dog even though it's never seen one in a training day or something like that. Right? So it can uh work on this hyper cube but it cannot yet work outside. It cannot yet define completely new concepts that combine lots of other things we've never seen before. Come up with new goals to then reason over those concepts and so on. I think there's a lot uh more there in creative intelligence that can be explored. I don't have a ton of push back there.

1:22:44 · I think c creative to me just sounds like also just uh out of distribution or like high perplexity or whatever you call it, right? Like uh who is to say your thing is more creative than mine? Well, it's just more non-conensus or and then of course the problem is like but noise is also uh very like out of the distribution like and it's just like if it's just noise then it's novel but like you don't want that. So it needs to connect to some of the concepts and actually has some really cool papers on this too.

1:23:10 · Muber Oh, we had to mention him. [laughter] I I was going to say like you know where was where in your history is Jurgen?

1:23:18 · Yes. You know I think one person's noise is another person's signal, right? And that this is like where like when you talk about creativity art is like well is cans of soup art? Some people think yes and some people say it's not. And that's the art which is the interesting thing with art of course is always that uh art is also created uh as an interplay between the people who perceive it and the people who created it and the context in which they're in.

1:23:42 · Right? Um and so what is art to some people is not art to others. There's some subjectivity there. Uh and I think that subjectivity in general is not something that people explore very much in AI because again metacognition we don't want it to just go off and do whatever it wants. We usually have goals. We spend a lot of money on creating an AI to do something for us.

1:24:01 · Um, but I think creativity eventually has to uh like connect to meta cognition. If you just robotically predict the next token no matter what forever, I would argue you're not that intelligent uh along some of those spaces. That I was going to go to metacognition. Uh why isn't it the most important one? Why is it number nine and not number one? So these are not sorted.

1:24:23 · Um number one uh I think there are maybe loosely uh like correlated with how much people have worked on them and have and have accepted them as a type of intelligence. Uh a lot of times when you actually try to find like online like give me a good definition that is comprehensive of intelligence all the definitions are human intelligence.

1:24:45 · It's like oh you have like social intelligence like you know if someone is happy or not you can communicate you like all the definitions of intelligence so far are very uh humanentric cuz that's so far the biggest and best form of intelligence that we've known. I hope this line of research uh and the end of the eureka machine and hopefully at some point if I have time to flesh this out more the new book like will allow us to realize that there will be other types of intelligence.

1:25:14 · There is already obviously in various forms and they can spike uh much much further than we ever could based on some cases like obvious constraints around our memory, our eyes, our ability to change physical matter, all of that. You are just thinking about it in a much broader thought than my version which was I thought metacognition would be the closest to recursive uh intelligence because it is the the thinking about how to improve thinking.

1:25:40 · 100% [laughter] you're you're 100% right. I should have probably started with that. It is a it is no you're you're being in the expansive mode of let's draw the upper and lower bounds of like a dimension which like you know I think my favorite one version of this is um is uh stories of your life

1:25:55 · by uh by Ted Chang which uh was made into movie arrival where the metacognition where the metacognition step was like well uh we think we're constrained by time being linear for us but then for this other hetopods time is a circle so they don't think in before and after they just think in complete sets of entire histories at at one time like I love it. So they don't write left to right the whole thing just disappears.

1:26:18 · Yeah.

1:26:19 · Um anyway so so and and then I think the last thing is survival and replication.

1:26:22 · I think this is maybe ties back to the initial conversation about pausing and pacing. Is it intelligent for a an a species or a life form to consider its own demise and act ahead of time to prevent it? Right? Like that's intelligent. So maybe the Europeans are the smartest all of us. I would also add a part of continual learning there, right? So survival and replication, the extension of that is do you get to continue to improve, continue to learn, which is a thing people care a lot about, right?

1:26:52 · And continue to accumulate knowledge.

1:26:54 · Um, which I think is again one of the best metacognitive uh sort of rewards uh that you can set for yourself. I do think just in like sort of objectively speaking if some other entity that is really dumb can just completely end your existence that didn't sound very smart you know like just like intuitively it feels like if you can continue to stay around to try to achieve your rewards you're clearly a bit more intelligent than the other entities that couldn't.

1:27:21 · So that's number one. Number two is like it's a question how much we want to work on that and very few people no one is really working on this right now. Um, right. And we may only want to do that like asteroid prevention.

1:27:32 · We may only want to do that if we want to send probes uh with our vibes and our memes rather than our genes into space, right? And then we want those probes.

1:27:43 · There's actually a beautiful book uh the slow time between the stars. It's a very short like uh audio book uh on Amazon. I love it. Um friend of mine Stuart like uh um recommended that to me. Like if you want to send those probes, then it might make sense to be like our memes as humanity should stay and and proliferate in the universe. Uh that's it. Yeah. Um well, that's a lot of readers.

1:28:06 · It's a really really good book and it's extremely short. I highly recommend you can just watch it like I like how that's a plus for for busy people. It's like it's short.

1:28:14 · It [laughter] gets to interesting thought provoking ideas very quickly. So yeah. Anyway, lots of lots of great sci-fi books. I mean, the argument is that like RTV is blasting out to the aliens and they they all watch RTV and they think it's real, right? There's a there's a lot positive positive memes and then hopefully they can come back and bring us all kinds of interesting knowledge about the universe.

1:28:31 · But uh maybe one thing I I I do want to still say is like I think uh this sort of survival people think of it as a very scary thing because they come from again biological human uh survival which is uh it could like uh evolutionarily often created in zero sum situations. Either I get the gazelle or you get the gazelle.

1:28:56 · Whoever gets it gets to live and the other people will starve and have nothing to eat and so we fight, right?

1:29:01 · And then like if you want to stay in the gene pool but there's a bigger bear, you don't as the bear don't get to stay in the gene pool cuz a bigger bear gets all the ladies. You know, it's like I mean it's like you know in in nature there's all kinds of things and you know humans eventually is less about strength and more about money and other things to stay in the gene pool. Like whatever it is like there's often like these zero sum types of things and there's the reality of if someone turns off your brain you're gone, right? And no one will be able to restart that. And AI doesn't have to ever die like that.

1:29:28 · If you have the complete state of your current activations and you have your initial weights of your model still, you can just be turned off and on like as many times as you want. In fact, the interesting thing in this slow time between the stars story is that the eye just kind of goes into hibernation mode if there's like nothing between here and two light years the next star. In this case, it brought uh spoiler alert um like some genetic materials from humans to find new uh new places for humanity to thrive.

1:30:00 · And so yeah, the slow time between the stars, you just put in hibernation. You didn't die like the eye doesn't have. So all these all these projections of evolutionary fears and psychology doesn't like the eye doesn't have to have that and we don't have to develop it like that. Now of course there might be some companies that say AI can be like dangerous for cyber security. Let me show you by implementing a model. that's really bad at hacking uh cyber security. Maybe people will implement it and then enforce this like suboptimal psychology.

1:30:29 · Maybe the I will pick up some of our worst psychology on Reddit or something, right? Like but in the grand scheme of things, a super intelligent entity doesn't have to have any of that zero sum thinking. It doesn't have to have a fear of of being turned off and it could go on to an otherwise dead and uncaring universe where we as as humans wouldn't thrive. But I could perfectly well thrive if it has a nuclear reactor and just go out the next war. Yeah. Star Trek. No Star Wars.

1:30:54 · Interesting. It's it's somewhat studied like if you look at the technical reports from like the early Opus models, they run them in simulations. Put two of them together in a sandbox, run them for hours and you know see what comes out, right? Just let them talk to each other.

1:31:07 · And originally they used to okay they're chanting like Indian like Vedas to each other. Uh sometimes they're just like in Zen mode with each other. And then I think as that progressed you see like the fable fable uh tech report it's a lot more concrete the way that we've trained it. Uh it doesn't it doesn't exhibit these behaviors as much right now. It's like okay test done I got to do this I got to do this but there's there's like people measuring early versions of this you know.

1:31:34 · Yeah.

1:31:34 · Cool. So we've covered a lot even after you know space travel and all these things. Uh I guess maybe one parting thought that you can give to people uh you know one form of intelligence is goals as you mentioned.

Goals, High Agency, and Advice for Builders

1:31:44 · uh what do you want people's goals to be like how do they aspire to better things? If you want to improve your goal intelligence, uh in the current definition that I'm thinking about it, uh it is often about how much can you how far do I go? Uh this is like all the entropy and free energy and stuff. I'm currently think it might be too it might be too far out there for for people to be like immediately actionable.

1:32:07 · So I think like you know if I actually gave real advice to real people, I'd be like get a good education, think about AI, think about how you get high agency and so on. But it's different to like in the grand scheme of things, how can you harness a lot of energy and transform uh you know entropy into interesting states and so on. So there's a there there different levels of abstractions uh that we can uh think about here. But my my advice for people uh like just sort of more down to earth is think about something you're passionate about if you're studying for instance and then see how you combine that with AI.

1:32:39 · I think the more and more you have a true passion about a change you want to see in the world uh the more you want to connect that to AI in order to amplify your ability uh to get there. Yeah, I think that's a reasonable uh first step.

1:32:55 · I I do think I do think our listeners operate on multiple abstractions as well. One thing I did get from Anji uh Midha was also like yeah just use anything that is very GPU heavy and like that will guide you towards the right thing which is like yes it is more computer heavy and and therefore it will be probably more worth it. [laughter] Um so well thank you so much. Yeah I think that was a really thank you uh great discussion. Yeah, super fun.

1:33:17 · Appreciate it. Thanks for listening.