Will Everyone Have a Personal AI? With Mustafa Suleyman, Founder of DeepMind and Inflection

0:05Today on No Priors, we're speaking with Mustafa Suleiman, co-founder of DeepMind, the pioneering AI lab acquired by Google in 2014 for$650 million, and now co-founder and CEO of Inflection, along with Reid Hoffman and Corinne Simonian. Inflection just launched their first public product, Pi, last week. Mustafa, welcome to NoPriors. Thanks so much for joining us. Thanks for having me. I'm super excited to be here. Yeah, we're very excited to have you today. I think one thing that'd be great to maybe start with is just a little bit of your personal story, because I think you have a really unique background.

0:35You're very well known, obviously, for DeepMind and your pioneering work in the AI world. But I think before all that, you worked on a Muslim youth helpline. You started a partnership and consultancy that was focused on conflict resolution to navigate social problems. I'd just love to hear a little bit more about the early days of things that you did before DeepMind. And then maybe we can talk a little about DeepMind and sort of more recent stuff as well. Yeah, sure. I mean, the truth is I was very much a kind of change the world kid growing up, Like a big believer in grand visions, doing good, having a huge impact in the world.

1:09And that was always kind of what drove me. So when I grew up in London and went to Oxford, but at the end of the second year of my philosophy degree, I was kind of getting a bit frustrated with this sort of theoretical, you know, nature of it all. It was full of hypothetical moral quandaries. And so a friend that I met at Oxford was starting a telephone counseling service, a kind of helpline. And it really appealed to me. It was a non-judgmental, non-directional, secular support service for young British Muslims. And this was like about six months after the 9-11 attacks. And so there was quite a lot of like rising Islamophobia.

1:55and the government was talking a lot about anti-terrorism. And, you know, in general, I think that like sort of migrant communities were feeling the pressure. And this was a support service that was staffed entirely by us, by young people. I was 19 at the time. And, yeah, I spent almost three years working pretty much full time on that. And it was an incredible experience because it was basically my first startup. and, you know, fundraising was the name of the game, except the numbers were much, much smaller than they are these days. And, you know, the service was staffed by almost 100 volunteer young people, which was just amazing because we felt like we can actually do something.

2:41You know, it was quite liberating and energizing to actually give this a shot. And, you know, I was very much inspired by the kind of human rights principles. It was deliberately not religious, even though it used some of the kind of culturally sensitive language that helped people feel heard and understood. So, yeah, it's had a very formative impact on my outlook. Yeah, no, it's super interesting. And I think we can talk more about that in the context of AI in a little bit. One other thing that you did is you also started a consultancy where you worked as a negotiator and facilitator. And I believe you worked with clients like the United Nations, the Dutch government and others.

3:16Can you tell us a little bit more about that work as well? Yeah, I mean, I was always trying to figure out how to scale my impact. And, you know, I quite quickly realized that delivering a sort of one to one service via a nonprofit was not going to scale a great deal, even though it had an amazing impact, you know, on a kind of human to human level. And so I was super interested in these like meta structures, like how does, you know, the UN actually influence, you know, behavior at the country level? And, you know, how could we run more efficient decision making processes where there's tension and disagreement?

3:57So we worked all over the world, actually, in Israel, Palestine and in, you know, in Cyprus between the Greeks and the Turks. my colleagues worked in South Africa, Colombia, Guatemala, and I think it really taught me that learning to speak other people's social languages is actually an acquired skill. You really can do it with a little bit of attention to detail and some patience and care. It's kind of a superpower being able to deeply hear other people and make them feel heard such that they're better able to empathize with people that they disagree with. And that's been an important theme throughout my kind of career, something I've always been interested in.

4:41So I think I co-founded that and worked on it for, I think, three years and soon realized the limitations of large-scale human processes. I mean, in 2009, I facilitated one part of the climate negotiations in Copenhagen. And, you know, it was a kind of a remarkable experience, like, you know, 192 countries, literally a thousand NGOs and activists, many different academics, everyone proposing a different solution, a different definition of the problem. And, you know, in one way, it was sort of inspiring to see so many different cultures and ideas coming together to try to form consensus around an issue that was clearly of existential importance.

5:29On the other hand, it was just like deeply depressing that we weren't able to achieve consensus. It took another decade to even get mild consensus on this or half a decade, 2015. And I think that was sort of an eye-opener for me. I was like, the world's governance systems are not going to keep up with both the exponential challenges that we face from globalization and carbon emitting, but also like technology. And that was the next thing that I saw on the landscape. So how did this lead into your interest in AI? And, you know, I believe that you met Demis when you were quite young, and I think he and your other co-founder worked together later in a lab.

6:10But I'm a little bit curious, like, how your background and interest in these sorts of global issues then transformed into an interest in AI and the founding of DeepMind. Yeah, well, around about that time, actually, like, I guess it was like 2008 or so, I was starting to keep an eye on Facebook's rise. And I was like, this is incredible. I mean, this is like a two or three year old platform at that point. And it had hit like 100 million monthly actives. And that was just a mind blowing number to me. And it was obvious that this wasn't just the kind of neutral platform for giving people access to information or connecting people with other people.

6:47The frame, because I had come from a conflict resolution background, our entire approach was like, what is the frame of a conversation? Like, how do you organize space? How do you prepare individuals to have a constructive disagreement? How do you set up the environment basically to facilitate dialogue? And so that was the lens through which I looked at Facebook. I was like, well, this is a frame. There's a choice architecture here. There are significant design choices which are going to incentivize certain behaviors. Obviously, at that point, there wasn't really ranking. but even just having a thumbs up or like the choice of, you know, which button you place in what order and how you arrange information on the page and what all of that drives behaviors in one way or another.

7:34And, you know, that was a big realization to me because I was like, well, this is actually reframing the default approach to human connection at a scale that is like completely unimaginable. unimaginable. I mean, perhaps only akin to, you know, the default expectations in a religion, for example. Everyone grows up with an idea that there is a, you know, a patriarchy, a male god, that, you know, that there's a particular role for women. Like that's, you know, until a few decades ago, that was just an implied sort of undertone to an entire social structure for thousands of years. And that's kind of what I mean by frame.

8:13There's sort of these implicit design choices which cause hundreds of millions of people to change their behavior. Yeah. And I think that's super interesting because I remember working on a bunch of Facebook apps at the time when the platform launched and people were purposefully thinking about that stuff, but on the micro level, right? How do we get more users? How do we get people to convert? How do we drive certain behaviors? And so everybody I think was very explicitly thinking about this as a behavioral change platform, but not at the level of society. You know, we were thinking about it in the context of just like, how do you get more people to use this thing?

8:44And so I think it's really interesting that people then later realize the big ramifications of this in terms of how that actually cascades in terms of social behaviors and other things. How did that lead to starting DeepMind? Well, it was clear to me from that moment on, like I left Copenhagen in 2009 thinking this is not the path to significant positive social change. it still needs to continue and I support those processes obviously but I'm just saying it is just not something that I feel I could continue to work on full-time and so my heart was set on technology at that point so I reached out to Demis who was the brother of my best friend from when I was a kid we got together we had a coffee we went and actually we played poker at one of the casinos in London because we both love games, both super competitive, both good at poker.

9:36And on that night, I think we both got knocked out pretty early in the tournament. So we sat around drinking Diet Coke, talking about ways to change the world. And we basically were, you know, having exactly this conversation. Like, you know, is it going to be, I mean, obviously at that point I was mostly inspired by platforms and software and social apps and connectivity and so on. Whereas, you know, Demis was way more in the kind of robotics land and sci-fi land. I mean, he was fully thinking that, you know, the way to manage the economy, the way to make economic decisions was to simulate the entire economy, right?

10:18And he thought that he was very much obviously had just come off the back of his games like Evil Genius and Black and White and so on, which were kind of simulation-based games. So I think that was his default frame at that point. Yeah. And then we spent many months talking and spent a lot of time with Shane Legg as well. And Shane was really the core driver of the ideas and the language around artificial general intelligence. I mean, he had worked on that for his PhD with Marcus Hutter on definitions of intelligence. I found that super inspiring. I think that was actually the turning point for me, that it was pretty clear that we at least had a thesis around how we could distill the sort of essence of human intelligence into an algorithmic construct.

11:04And it was his work in, I think for his PhD thesis, he put together like 80 definitions of intelligence and aggregated those into a a single formulation, which was how do we, you know, the intelligence is the ability to perform well across a wide range of problems. And he basically, you know, gave us a measurement, an engineering kind of measurement that allowed us to constantly measure progress towards, you know, whether we were actually producing an algorithm, which was inherently general, i.e. it could do many things well at the same time. Is that the working definition you use for intelligence today?

11:46Um, actually, no, I've changed. Um, I, I think that there's a more nuanced version of that. I think that's a good definition of intelligence. But I think in a weird way, it's over rotated the entire field on one aspect of, of intelligence, which is generality, you know, and I think, um, open AI, and then subsequently anthropic and others have taken up this default sort of mantra that like it all that matters is can a single agent do everything you know can it be multimodal can it do translation and speech generation recognition etc etc i think there's another definition which is valuable which is the ability to direct attention or processing power to the salient features of a of uh an environment given some context right so um actually what you want is to be able to take your raw processing horsepower and direct it in the right way at the right time because it may be that a certain tone or style is more appropriate given a context it may be that a certain expert model is more suitable or it may be that you actually need to go and use a tool right and obviously we're starting to see this emerge.

13:10And in fact, I think the key, and we can get into this obviously in a moment, but I think the key element that is going to really unlock this field is actually going to be the router in the middle of a series of different systems, which are specialized, some of which don't even look like AI at all. They might just be traditional pieces of software, databases, tools, and other sorts of things, but it's the router or the kind of central brain, which is going to need to be the key decision maker. And that doesn't necessarily need to be the largest language model that we have. It's really interesting because I feel like a lot of what you described is actually how the human brain seems to work in terms of you have something a little bit closer to a mixture of experts or MOE model where you have the visual cortex responsible for visual processing and then you have a other piece of the brain specifically responsible for empathy and you have mirror neurons.

14:02And, you know, it feels like the brain is actually this ensemble model in some sense with some routing, depending on the subsystem you're trying to access. And so, you know, the generality approach seems like a really, it almost goes at odds with some of those pieces of it, unless you're just talking about some part of the hippocampus or something, right? Well, I think that's long been the inspiration, right? I think for everybody, neural networks are the obvious example, but in many other elements, reinforcement learning, you know, and so on, are all brain inspired. And I think that, you know, there's been a lot of talk about, you know, sparsity as well, which is sort of what you're describing.

14:37And, you know, so far we've had to do, you know, very dense all-to-all connections because we sort of haven't quite learned the algorithms for sparse activations. But I think that's going to be a very promising area. And, you know, in many ways, what I'm describing doesn't actually require sparse activations because, you know, you actually could just train a decision-making engine at the middle to know when to use which size model, So maybe in some contexts, you would want the highest quality, super expensive 20-second latency model. And in most other contexts, a super fast three-second mini model might work fine.

15:14I think that's going to be the key unlock, actually. And quite sort of remarkably, that's an engineering problem, perhaps more than it is an AI problem, which is just a pretty surreal moment, just if you actually observe that, given where we are in the field and stuff. when you um started deep mind i think it was reasonably unpopular to do what you were doing right and so i think you ended up getting funded by um founders fund and peter thiel and nila musk but i remember at the time there was like three or four parties that funded a lot of ai things and nobody else was really doing it in terms of the types of approaches you were taking in terms of saying we're going to build these big ai systems that can do all sorts of things right yeah i mean it was wacky like i i can't say that enough like it it was especially for the first two years so because we founded in 2010 and for the most of the sort of spring and summer of 2010 actually most of the rest that year i was going to gatsby computational neuroscience unit at ucl sneaking in with demis and shane to just sit in on the lunches that uh peter diane ran and i remember shane like sort of saying to me, like, you know, the language here is machine learning.

16:28Yeah, you can say AI. Don't say AI. Yeah. I was like, okay, okay, I'll keep my mouth shut. Don't worry. Like, we certainly don't say AGI, you know, and that was a kind of, that was pretty weird. I mean, you know, there weren't, you know, there weren't very many funders for us. Like, you know, Peter Thiel, you know, to his credit, did actually have significant vision here although he sold pretty early i think and now doesn't seem to be in the game so but uh yeah he certainly he certainly saw it first um and you know i think that all changed pretty quickly first with uh you know alex net of course in 2012 and then with dqn uh the atari paper in 2013 um you know and then a kind of succession of breakthroughs after alpha go and people got more more more sort of aware of it but it still surprises me the extent to which the rest of the world is like suddenly waking up and obviously we've seen that like crazy in the last six months so yeah and then i guess last question on sort of your time with uh google and deep mind and because i think there's a lot of really exciting things to talk about in the context of inflection and sort of the broader field and world what are some of the things you were most excited to have the team create a deep mind over the years or some of the breakthroughs that you're most proud of?

17:43Yeah. Well, I mean, in some ways, we definitely sort of pioneered the deep reinforcement learning effort. And I think, you know, in principle, it's a very promising direction. I mean, you clearly want some mechanism by which you can learn from raw perceptual data, and that directly feeds into a reinforcement learning algorithm that can update and essentially iterate on that in real time with respect to some reward function, whether that's online or offline, like directly interacting with the real world in real time, or it's, you know, in a kind of batch simulation mode. You know, and that turned out to be very valuable for a specific type of problem, where a game-like environment had a very structured scalar reward, and we could play that game many millions of times.

18:40That's part of the reason why we started the Alpha Fold project, because it was actually my group that was looking around for other applications of DQN-like, AlphaGo-like tools. And in a hackathon that we did one week, someone stumbled across this problem. We'd actually looked at it back in 2013 when it was called Foldit, which was a very small-scale kind of version of this. And just for context, sorry to interrupt, you know, AlphaFold was focused on folding proteins, which at the time was a really hard problem, right? People were trying to do this molecular modeling and they couldn't really make any real headway and lots of the traditional approaches.

19:21And then your group at DeepMind really started pioneering how to think about protein folding in a different way. So sorry to interrupt, I just want to give context for people listening. So I think the hackathon was probably 2016. And then as soon as we saw the hackathon that, you know, start to work, then we actually, you know, scaled up the effort and hired, you know, a bunch of outside consultants to help us with the domain knowledge. And then I think the following year, we entered the Casp competition. So, you know, these things take a long time. And, you know, sort of longer than I think people realize, you know, there's a lot of that was a very big effort by DeepMind and eventually became a company wide strike team.

20:00So in hindsight, these things do take a huge amount of effort. Yeah, the fascinating thing here is that, you know, the work started with AlphaGo, which was how to play Go better, right, or how to beat people at Go. And then the same underlying approach could then be morphed and applied to protein folding, which I think is an amazing sort of leap or connection to make. And, you know, I used to work as a biologist and I remember you'd spend literal years trying to crystallize proteins in different solutions. You do all these different salt concentrations in each well so that the protein would crystallize so you could hit it with x-rays.

20:32And then you'd interpret those x-rays to look at the structure, right? And so you had to do this really hard sort of chemistry and physics to get any information about a protein at all. And then you folks with the machine ran through every protein sequence, literally in every database for every organism. And you're able to then predict folding, which is pretty amazing. It's very striking. Yeah. I mean, I think if I were to sort of summarize the core thesis of DeepMind, it was that it would be possible to – the motivation for generality was that you would be able to learn, you know, a rewarding behavior in one environment and transfer in a more compressed or efficient representation the insights that had made you successful in one environment to the next environment, right?

21:23That transfer learning has always been the key goal. And that was one of the very exciting proof points that it is increasingly looking likely that that's possible. So I definitely think that's pretty cool because when you think about the sorts of problems that we're facing in the world today, we don't have obvious answers lying around. There's no genius insight that's just waiting to be applied. We actually have to discover new knowledge. And I think that's the attraction of artificial intelligence. That's why we want to work on these models, because we're sort of at the limit of what the smartest humans in the world are capable of inventing.

22:06And we have, you know, very pressing, urgent global challenges, you know, from food supply to water to decarbonization to clean energy, transportation, you know, with a rising population that we really want to solve. So amidst all of the stresses and the fears about everything that's being worked on at the moment, it is important to keep in mind that there is an important North Star that everybody is working towards. And we just got to keep focused on those goals rather than sort of be too sidetracked by some of the fears. Let's talk about inflection. What was the motivation for starting another company?

22:47um well i guess back in sort of 2018 2019 it wasn't clear that neural networks were going to have a significant impact in language if you just think about it intuitively um for the for the previous sort of five years cnn's had been effective at learning structure locally right So pixels in an image, in the input. So pixels in an image that were correlated in space tended to produce subfeatures, which were a good representation of what you were trying to predict. Maybe there were lines and edges and they grew into eyes and faces and scenes and so on. And that kind of hierarchy just intuitively seemed to make sense and seemed to apply to audio and other modalities.

23:35Right. Whereas if you kind of think about it, a lot of the structure of predicting the next word or letter or token in a sentence seems to exist in a very, very, very spread out, you know, far removed from the immediate next step of the prediction. Right. And so it didn't look like that was working. And then, to be honest, like when GPT-3 came out, that was like a big revelation. I had seen the GPT-2 work and hadn't quite clicked for me that this was significant. It was really only when I saw the GPT-3 paper that my eyes were wide open to this possibility. It's pretty amazing that you could attend to a very, very seemingly sparse representation and use that to predict something which, on the face of it, It seemed like there were billions of possibilities of what might come next in a sentence, right?

24:33Maybe tens of millions or something, but a lot. And for me, it was early 2020 that I went to work at Google, and I got involved in the large language model efforts. I got involved in the MENA team that was called at the time. I know that you guys had Noam on the show recently. Noam's super awesome. And it was me and Noam, Daniel, Kwok Lee, and a few others. And it was just unbelievable what was being built there. And when I joined pretty small models and very quickly, we scaled it up. It became the Lambda group. And we started seeing how it could potentially be used in various kinds of search.

25:19Started looking at retrieval, grounding for improving factuality, started getting a feel for all the hallucinations and so on. And that was just really a mind-blowing few years to me. And while I was there sort of in the last year in 2021, I tried pretty hard to get things launched at Google. We were all kind of banging on the table being like, come on, this is the future. And, you know, obviously David Luan from Adept was also in and around that group. So the three of us in our own ways were pushing pretty hard for launch. And it wasn't meant to be. Uh, just, you know, timing is everything. And, you know, Google just wasn't, wasn't the right timing for Google for various reasons.

26:01Uh, and, you know, I, I was just like, look, this, this has to be out there in the world. This is, this is clearly the new wave of technology. And so, yeah, in January, I left, got together with Karen, my co-founder, um, who I worked with at DeepMind for seven years. We bought his company back in 2014 at DeepMind. He led the deep learning scaling team at DeepMind for years and worked on all the big breakthroughs at DeepMind. And then, of course, Reid Hoffman, who's been one of my closest friends for like 10 years. And we've always talked about starting something together. And I was like, this is the obvious thing.

26:35Now is the time for sure. And so the rest is history. It's been a wild ride since then. It makes me feel a little bit better than somebody who's been such a pioneer in the field and working on this all the time. is still constantly surprised, as I am also constantly surprised. I remember when you were first starting to get this going, another thing I was surprised by is the focus you, I mean, I came around to it in writing the investment memo, but you had this focus on the idea of companionship rather than information as the right initial approach. You've talked about, worked on, thought about empathy for humans and other populations for a long time.

27:18It seems counterintuitive. Like, why companionship? Yeah, it's a great question. So I think to step one step back from that first, I think my core insight about what was missing for Lambda was interaction feedback. And in a funny way, that was exactly what was motivating Karen to having beaten all the academic benchmarks and achieved SOTA many times. He had come to the same conclusion. I had seen the same thing from Lambda. What we were missing was user feedback. And actually, when you think about it, all of our interfaces today are fundamentally about interaction. You're giving your browser feedback all the time.

28:03You're giving that web service feedback, same with an app or anything that you interact with. It's actually a dialogue. And so the way I'd position Lambda at Google is that conversation is the future interface. and Google is already a conversation. It's just an appallingly painful one, right? You say something to Google. It gives you an answer in 10 blue links. You say something about those 10 blue links by clicking on it. It generates that page. You look at that page. You say something to Google by how long you spend on that page, what you click on it, how much you scroll up and down, et cetera, et cetera.

28:40And then you come back to the search login and you update your query and you say something again to Google about what you saw. That's a dialogue. And Google learns like that. And the problem is it's, you know, using 1980s yellow pages to have that conversation. And actually now we can do that conversation in fluent natural language. And I think the problem with what Google has sort of, I guess, in a way, accidentally done to the Internet is that it has basically shaped content production in a way that optimizes for ads. and everything is now SEO'd to within an inch of its life. You know, you go on a webpage and all the text has been broken out into sub bullets and sub headers and, you know, separated by ads.

29:31And, you know, you spend like five to seven or 10 seconds just like scrolling through the page to find the snippet of the answer that you actually wanted. Like most of the time, you're just looking for a quick snippet. And if you are reading, it's just in this awkward format. And that's because if you spend 11 seconds on the page instead of five seconds, that looks like high quality content to Google. And it's quote unquote engaging. So the content creator is incentivized to keep you on that page. And that's bad for us because what we want is a succinct. Well, we as humans, all humans clearly want a high quality, succinct, fluent, natural language answer to the questions that we want.