Demis Hassabis: Agents, AGI & The Next Big Scientific Breakthrough

continual learning, long-term reasoning, uh some aspects of memory, these are still unsolved.

I think all of these are going to be required for AGI.

Depending on what your AGI timeline is, you know,

mine's like 2030 or something like this,

then if you start off on a deep tech journey net today,

you have to just consider AGI appearing in the middle of that journey.

It's not bad necessarily, but you have to take that into account.

You have to have an active system uh

that can actively solve problems for you to get to AGI.

So, agents are that path, and I think we're just getting going.

>> Demis Hassabis has had one of the most unusual careers in tech.

He was a chess prodigy as a kid,

then designed his first hit video game, Theme Park, at 17.

He then went back to school, got a PhD in cognitive neuroscience,

published foundational work on how memory and imagination work in the brain,

and then in 2010 co-founded DeepMind with one mission, solve intelligence.

And I think they've done it.

Since then, uh his lab has gone on to do things most people thought

were decades away.

AlphaGo beat a world champion at Go, AlphaFold cracked protein structure prediction,

a 50-year grand challenge in biology,

and they gave it away for free to every scientist on Earth.

That work won him the Nobel Prize in chemistry last year.

Today, Demis leads Google DeepMind,

where he's building Gemini

and pushing toward the same goal he set

when he was a teenager,

artificial general intelligence.

Please welcome Demis Hassabis.

So, you've been thinking about AGI longer than almost anyone.

Uh when you look at the current paradigm, large-scale pre-training, RLHF,

chain of thought, how much of the final architecture for AGI do you think

we already have,

and what's fundamentally missing right now?

>> Well, first of all, thank thanks, Gary, for that great introduction,

and it's great to be here.

Thanks for for welcoming here.

It's amazing space, actually.

I'll have to come back here often.

Very inspiring that you all get to work in in in this space.

So, the question is I think the the components that you just mentioned,

I'm pretty sure will be part of the final architecture for AGI.

So, I think they've come such a long way now,

uh and we've proven out so many things about what they can do.

Uh I can't see a world in

which we'll sort of realize in a couple of years this was a dead

end.

That doesn't make sense to me.

But, there still might be one

or two things missing on top of uh of of of what you've you

know,

what we already know works.

So, um continual learning, long-term reasoning, uh some aspects of memory,

these are still unsolved.

Um and how to get the systems to be more consistent across the board.

Um I think all of these are going to be required for AGI.

Now, it might be

that the existing techniques can just scale up to

that with some innovation

and some incremental innovation.

Um but, it could be

that there's still one

or two big ideas left uh

that need to be cracked.

I don't think it's more than one or two if there are out there.

And I think, you know,

my betting is uh about 50/50 if that's the case.

So, of course, at DeepMind at Google DeepMind we work on both those things.

>> I guess that's what I mean.

Working with a bunch of identical systems,

the wildest thing to me is to what degree it's the same weights ev-

over and over.

So, this idea of continual learning is so interesting because like yeah,

right now we're sort of cobbling it together with duct tape, you know?

>> Yes.

>> These dream cycles at night and things like that.

>> Yeah.

It's pretty cool, the dream cycles,

and we we used to think about this with consolidation with episodic memory.

It's actually that's what I study for my PhD is how the hippocampus works

and integrates,

you know, new knowledge gracefully into the existing knowledge base.

So, the brain does that amazingly well.

It it it does it through you know,

during sleep uh especially things like REM sleep,

replaying back episodes that that are important so that you can learn from it.

In fact, our very first Atari program DQN,

one of the ways it was able to master Atari games was by doing

experience replay.

So, we sort of borrowed

that from from neuroscience

and replayed successful trajectories uh many times,

you know, that's way back in 2013 now in the in the dark ages

of AI.

It was uh a really important thing.

And and I agree with you, we're kind of using duct tape right now.

So, like shove it all in the context window.

Um this but it seems a bit unsatisfying, right?

And actually, even though uh we're working on machines, not biological brains,

and so you potentially you could have, you know,

millions or tens of millions size context window or memory,

and it can be perfect,

there's still a cost to looking it up

and finding the right thing uh

that that's actually relevant for the specific uh decision you've got to make right

now.

And that's non-trivial that cost, even if you can potentially store it all.

I think there's actually a lot of room for innovation in in areas like

memory.

>> Yeah.

I mean, the one thing is like it feels like a million token context

one is actually bigger than I mean,

it's plenty big, honestly.

You can do stuff.

>> It's plenty big for for for most things that it should be used for.

I mean, if you think about the context window is sort of equivalent to

working memory,

you know, humans have we have like a few digits, you know,

it's like a a dozen digits maybe, you know, average of seven.

We got million or, you know, 10 million context windows,

but the problem is is that we're trying to store everything in that,

you know, things that aren't in not important, things that are wrong.

It's pretty brute force currently, and that doesn't seem uh right.

And then the problem is

if you're an agent trying to try

and process live video,

and you're just going to naively record all the tokens,

then actually a million tokens isn't that much.

It's only like 20 minutes.

So, actually you need more if you want something that's going to understand your,

you know, your what's going on in your life over maybe a month

or two.

>> DeepMind has uh historically leaned into reinforcement learning and search.

Uh AlphaGo, AlphaZero, and MuZero.

Uh how much of

that philosophy is actually embedded in how you're building Gemini to get today?

Uh is RL still underrated?

>> Yeah, I think potentially it is.

It sort of goes in in ebbs and way- waves.

You know, we've worked on agents since the beginning of DeepMind.

In fact, we also That's what we said we were working on.

And so, all of the Atari work and AlphaGo, most specifically, they're agent systems.

And what we meant by that is systems that are able to, you know,

accomplish goals on their own.

Uh and make active decisions and and make plans.

And so, of course,

we were doing it in the domain of games to to to make it

tractable.

Uh and then doing increasingly complex games, things like StarCraft after AlphaGo, AlphaStar.

So, um we basically did all the games that are out there.

Um and then of course, the question is,

can you generalize those models to be world models or models of language,

not just models of simple games, uh or even complex games.

And that's what the last few years has been about.

But really, you can think of a lot of the things we're doing today,

all the leading models with thinking modes

and chain-of-thought reasoning as aspects of what was sort of pioneered with AlphaGo coming

back now.

And I actually think there's a lot of work we did back

then that is relevant today,

and we're sort of re-looking at some of those old ideas um at scale

today in a more general way,

including things like Monte Carlo tree search

and other other ways of doing augmenting the RL uh on top of the

the reinforcement learning we're ready to do today.

And I think a lot of those ideas both from AlphaGo

and AlphaZero are really really relevant to to where we are with today's foundation

models.

And I think a lot of

that is what we're going to see of the advances the next few years.

>> One question I would have like obviously today you need bigger

and bigger models to be smarter

and smarter.

But then we're also seeing distillation working.

And then smaller models can be like quite a bit faster.

I think you know,

you guys have incredible flash models

that are like nine like you're finding

that they're 95% as good

as the frontier and at like 1/10 the price.

Is that right?

>> I think that's one of our core strengths is I mean you have

to build the biggest models to to to to have the frontier capabilities.

But I think one of our biggest strengths has been distilling

and packing that power into smaller

and smaller models very quickly.

Obviously we we you know,

we invented the kind of distillation process

and and people like Jeff

and Oriol and and others.

And we're still world experts in that.

And we also have a huge need to do it

because we've got to serve the biggest probably AI surfaces there are.

Obviously there's search with AI overviews and AI mode.

Then there's Gemini app.

And now increasingly every single product at Google has you know,

maps and YouTube and

so on has some aspect of Gemini

or Gemini related technology in it.

And so that's billions of users a dozen more than a dozen billion user

products.

And they have to be served extremely fast,

extremely efficiently and cheaply and with low latency.

So that that gives us a really important incentive to to make these flash

and even smaller models flashlight models extremely efficient.

And hopefully that ends up

then being really useful for many of the workloads

that all of you use for.

>> I'm curious about how much smarter these smaller models can actually be.

Like, are there limits to the distillation process?

Like, could a 50B

or 400B model be

as smart as like a mythos for today?

>> Yeah, I didn't I didn't see any I don't think we've got to

any kind of or at least none of us know yet

if we've got to any kind of information

or limit.

I mean, maybe at some point

that will be the case where there's just an information density

that can't we can't get beyond.

But, I think for now there's the assumption we make is that you know,

a year later after one of our leading, you know,

pro models or frontier models goes out, half a year later, a year later,

you'll have them in the the really tiny, almost edge models.

And you'll also see some of that goodness in our Gemma models,

which hopefully you're all enjoying our Gemma 4 models,

which I think are really amazing power for their sizes.

So, again, that uses a lot of this uh these distillation techniques

and and the idea of how to make things really efficient in these very

small models.

So, I don't really see any limit yet in terms of like some kind

of theoretical limit.

I think we're still pretty far off of that.

>> That's a mean I mean, that is really good.

>> Yes.

>> Uh you know, one of the weirder things

that we're seeing right now is like engineers can do like 500 to 1,000

times the amount of work

that they were doing like 6 months ago,

I guess.

I mean, the people in this room there are people who are doing about

like a 1,000 x the work

that like I Steve Yegge talks about this.

It's like a 1,000 x the work

that a Google engineer from the 2000s was doing.

>> I think it's very exciting.

I mean, I think models have many uses.

One is obviously cost, but the speed can allow, you know,

if you think about coding even or other things,

you can iterate a lot faster.

Also, especially if there's if you're collaborating with the system.

I think there's a there's a a lot of need for having fast systems

that maybe are not quite frontier,

like you said, like 95%, 90%,

but that's plenty good enough

and actually gain back more than the 10% on the the iteration speed.

So, and then the other big thing I think is running these things on

the edge.

Again, for efficiency reasons, but also for privacy and security reasons, too.

Um if you think about different devices

that you might run these systems on

that that,

you know, process very personal information.

Can also think about robotics, as well.

Um you know, robots in your house.

I think you're going to want very efficient, uh very powerful, uh local models,

which maybe are orchestrated you know, with some bigger models,

frontier models that are in the in the cloud,

but you only delegate to that in certain circumstances.

And perhaps you, you know, you process all of the audio-visual feed,

let's say, locally, and that stays local.

I could imagine uh

that would be a very good sort of um end state.

>> Y Combinator Startup School is back.

We're hand-selecting the most promising builders in the world

and flying them out to San Francisco for July 25th

and 26th to discuss the cutting edge of tech.

Apply now for a spot.

Okay, back to the video.

>> Going back to context and memory, models currently stateless, but, you know,

continue like what would the developer experience even be like for someone who's using

a continual learning model?

Like, you know, any idea like how you'd steer it?

>> I think it's really interesting.

I think that's one of the not having continual learning currently is one of

the things holding back agents from doing full uh tasks,

you know?

I think they're really useful for aspects of tasks right now,

and you can patch them together and do some really cool things,

but they don't adapt well with the context that you're in.

And I think that's the missing piece for them being really kind of fire

and forget,

and they'll figure it out themselves.

You know, I think they need to be able to learn um about the

specific context um that you're going to put them in.

So, um I think we have to crack that to get full general intelligence.

>> Where are we on reasoning?

So, models can do really impressive chain of thought now,

but they still fail on things a smart undergrad wouldn't.

What specifically needs to change and what progress do you expect in reasoning?

>> There's a lot of innovation left in in think the thinking paradigms,

I would say.

Again, I think we're fairly we're doing fairly simplistic things, fairly brute force.

One could imagine I think there's a lot of scope for example in monitoring

the chain of thought,

maybe interjecting midway through a thought process.

I often get the impression with our systems

and and our competitor systems

that they're almost overthinking.

They're almost getting into sort of loops of things like one thing I sometimes

like to do is is play chess against Gemini

and you know,

it's the all the leading foundation models are pretty poor at games,

which is quite interesting.

It's very cool to kind of look at the thinking traces cuz obviously these

are can be a well-understood,

you know, I can tell quite quickly

if it's going off on a tangent

and it's very sort of provable what the what the the thinking is doing

whether it's useful or not.

And so, what we see is that, you know,

sometimes it will it will it will consider a move.

It will realize it's a blunder, but it can't find anything better,

so it kind of goes back to that move and does it anyway.

So, it you know,

you just shouldn't be seeing

that happening in a in a very precise reasoning system.

So, there's just sort of huge gaps, I think, still,

but it may only be one

or two tweaks that are required to fix those kind of gaps just to

be clear,

but I think that's pretty pretty obvious they are there

and that's why you get this kind of jagged intelligence.

You know, on the one hand, it can solve gold medal problems in IMO,

which is super hard, but on the other hand, as we've all seen,

it can still make basic elementary math errors

if you pose the question in a certain way,

right?

So, or elementary reasoning errors.

So, there's just something to me about the almost an introspection about its own

thought process that I feel like there's there's something maybe missing there.

>> Agents are really big.

Some would say they're hyped.

I personally think they're just getting started.

It's totally insane.

What does DeepMind's internal research tell you about where agent capabilities actually are right

now versus,

you know, sort of the hype out there?

>> I think we are I agree with you.

I think we're just at the beginning.

You have to have an active system

that can actively solve problems for you to get to AGI.

That was always clear to us.

So, agents are that path and I think we're just getting going.

I think all of us are getting used to how do we best work

and you're leading the way in a lot of this in your own personal

experiments.

I'm sure many of you are doing that.

I think how do you incorporate it into your workflow in a way

that isn't just sort of a nice to have,

but actually starting to do fundamental things. is at the moment we're all experimenting

we're experimenting a lot of things,

but we're only in the maybe the last couple of months starting to find

the really valuable places.

And the technology probably only getting good enough for that to be the case,

right?

Where that it's not a kind of toy nice demonstration,

but actually really adding value to your to your to your time and efficiency.

I'd often wonder I see a lot of people working on like setting off,

you know, dozens of agents for like 40 hours,

but I'm not sure I've seen the output

that yet of that quite justify

that level of input going in,

but I think it will come.

So, I still think we're in the experimentation phase.

We haven't seen a AAA game

that tops the App Store charts

that was sort of vibe coded yet,

right?

I've seen and I've programmed

and I'm sure many we've all done little nice demonstrations

and it's like amazing.

I can do a prototype of theme park in half an hour now

which took me 6 months back

when I was 17.

It's kind of mind-blowing

and I and I wish I I got this feeling

if I spent the whole summer working on it,

you could make something really incredible, but it still needs craft and you know,

human sort of soul into it and taste.

I think that's that's something

that can that's you have to make sure you still bring

that to to whatever it is you're building.

And I think it still shows like it's not quite there yet

because why haven't we seen a kid making a hit game that's

that sells 10 million copies,

right?

That should be possible given the effort that's gone in.

So something's still somehow missing.

Maybe it's to do with the process,

or maybe it's to do with the tools.

I'm not quite sure.

You will probably know better than me cuz I'm sure you're all experimenting on

that.

But I haven't seen the result yet

which I would expect once this is really delivering

that full value.

Which I think will come in the next 6 to 12 months.

>> Some of it is like how much of it will be autonomous versus

I mean,

I don't think we'd see autonomous first.

We would actually probably see people in this room operating at 1000X,

and then >> That's what you should see first, and then many of you,

you know, they'll be like games companies or you know,

other types of companies that have built some kind of best-selling app,

best-selling game using these tools.

That's what you should see first, and then more of that will get automated.

>> I mean, some of it is like there's a human in there,

and then the human doesn't want to say

that the the the agents did it yet.

>> I think part of it might be

though that um this we want to discuss like creativity.

What I often say about

that is like if we look at the things we've done like AlphaGo.

So obviously very famously you'll all know about the move 37 in game two,

and for me I was waiting for a moment like

that to start the science projects like AlphaFold.

We started AlphaFold like the day we got back from Seoul,

which is 10 years ago now.

I'm going to Korea after this to celebrate the 10-year anniversary of AlphaGo.

But it's not enough to come up with move 37.

Like that's pretty cool, very useful, um but can it invent go?

That's what I want a system

that can invent go

if you give it a high-level description,

you know, like a game you can learn the rules of in 5 minutes,

but it takes a many lifetimes to master.

It's beautiful aesthetically, um

but you can play it in a few hours in an afternoon.

So, you know, maybe you could imagine

that would be the high-level description I would give

and then I'd want the the return the thing I get back is go.

Right?

And um clearly today's systems, I think, can't do that.

So, the question is why?

Um and I think there's something still missing there.

>> Well, someone in this room might might make it.

>> Then the answer would be there's nothing missing.

It just was the way we were using the systems.

And that might actually be the answer.

It might be that today's systems are capable of

that with a brilliant enough creative person using it

and providing that impetus

that the soul of the project

and being able to probably being au fait enough with the tools to like

almost be at one with the tools.

I could imagine that would be happening

if you experimented with the tools all day

and all night like probably many of you are doing

that and you combine

that with proper deep creativity,

um something, you know, more incredible could be done.

>> Switching gears to open source, I mean, or open open and open weights.

I mean, the recent release of Gemma,

you're making highly capable open and accessible ones that can actually run locally.

What do you think

that means for you will AI be something

that is in the hands of the users instead of primarily in the cloud?

And does that change who gets to, you know, build with these models?

>> We're huge proponents of in general of open source and open science.

And you mentioned AlphaFold at the beginning, you know,

we put that all out there for free.

And all of our science work, even still today, we publish in, you know,

the big journals.

We wanted to create uh world-leading models for their their sizes.

Right?

And so, that's what we hopefully we've done with Gemma.

And we're, you know, very committed to that path.

And hopefully you all experiment and build and and enjoy and using Gemma.

I think it's been like 40 million downloads now

and uh it's just in you know 2

and 1/2 weeks.

So we're really excited about that.

And I also think it's important for there to be Western stacks on open

source.

You know, obviously a lot of the Chinese models are excellent

and and they're currently well well leading in open source

and we think Gemma is very competitive for its sizes uh in in all

those respects.

And for us, I mean there is a question of resources, talent, and compute.

Like nobody has enough spare compute to just make two, you know,

uh frontier models at maximum size, right?

With different attributes.

So that's pretty difficult.

But also for what for now what we've we've decided is

that our edge models,

the things we want to use for Android and glasses and robotics,

um it's best that they're open models

because they're vulnerable anyway on the once you put them out on the surfaces.

So they might as well be actually fully open, right?

So we've sort of made a decision to kind of unify

that uh at the at the kind of we call it nano size level.

So that actually works for us uh strategically as well.

Um and you know, we hope as many people as possible build on it.

And of course, we'll be building on that, too.

>> Earlier uh before we came on,

I got to show you a demo of uh my version of Samantha from

Her,

which is Yes. uh harrowing for me to try to demo something to you.

>> Yeah, very good.

>> Um and it worked, which is amazing.

Gemini was built multimodal

and I spent a lot of time with a bunch of the models

and I mean the depth of the context

and the tool use with speech directly to model,

like there's nothing like bar none, like the best one actually.

>> Yeah.

Yeah, I think I think that's a sort of still a slightly underappreciated aspect

of of of the Gemini series is we we started it being multimodal from

the start.

That made it a little bit more difficult actually to begin with cuz

then just focusing on text,

for example.

But I we believe we're going to gain from that in the long run.

And I think we're seeing that now for things like world model building,

so stuff like Genie that we build on top of Gemini.

I think it's going to be really important for things like robotics.

So, this is why Gemini robotics, which many of you probably played around with,

I think it's going to be built on multimodal foundation models, the robotics models.

And we think we have a sort of competitive advantage with with Gemini being

so strong at multimodal.

We're using it increasingly in things like Waymo,

um but also if you imagine devices and assistants,

uh that digital assistants that come with you into the real world, you know,

maybe on your phone or glasses or some other device,

um it needs to understand the physical world around you

and intuitive physics uh

and and the and the physical context you're in.

And that's what our systems are extremely good at.

And I think you found that's why you've enjoyed using it in your setup.

We're planning to continue on

that and I think we're far

and away the strongest models on on those types of uh problems.

>> So, the cost of inference is uh dropping fast.

What becomes possible when inference is essentially free,

and how does that change what your team is actually optimizing for?

>> Yeah, I'm not sure inference will ever be essentially free.

I mean, there's sort of Jevons' paradox

and other things about like I think we'll just end up using all of

us will end up using whatever we can get our hands on.

And you could imagine uh millions of agents,

swarms of agents working together on things.

So, that's one way to use the inference.

Or you could imagine uh single agents

or smaller groups of agents thinking for in multiple directions

and then ensembling that.

So, we're experimenting with all these things.

Probably many of you are.

All of that will use up any inference I think that's available.

I mean, one day maybe it can be almost cost zero,

certainly the energy if we solve fusion or, you know, superconductors or, you know,

optimal batteries or some set of those things,

which I think we will do with material science.

And Energy costs will be essentially zero,

but there'll still be the physical creation of the chips and other things.

There'll There'll be some bottleneck, um at least for the next few decades,

I think.

And so, if that's the case, there'll still be rationing on the inference side.

You still have to use it, I think, efficiently.

>> Yeah.

Well, luckily, the smaller models are getting smarter and smarter, which is fantastic.

Uh we got a lot of bio and biotech founders in the audience.

I can see a few.

AlphaFold 3 took us beyond proteins to a broad spectrum of biomolecules.

Uh how close are we to modeling full cellular systems,

or is that still a fundamentally harder problem in a class of its own?

>> Well, I Isomorphic Labs,

which we spun out from from from from DeepMind after we did AlphaFold 2,

um it's it's which is going amazingly well.

It's it's it's trying to build out uh not just AlphaFold.

It's just one piece of the drug discovery process, uh as many you know,

but we're trying to do the the adjacent biochemistry

and chemistry to design the right compounds with the right properties,

and so on.

We'll have some big announcements for, you know,

very soon to talk about on the on that front.

I think that's going really well.

Eventually, you want a whole virtual cell.

So, I've talked about this in many of my science talks about a full

working simulation of a cell

that you can perturb,

and then the, you know,

the the outputs of that would be close enough to experimental that it's useful,

right?

You could skip out a lot of the the search steps,

and generate lots of synthetic data to train other models

that then would predict things about,

you know, real cells.

And um I think we're about 10 years away probably from something like a

virtual cell,

like a full virtual cell.

You know, we're starting out This is we're working on the DeepMind side,

science side, on a, you know, virtual nucleus, cell nucleus first,

cuz it's relatively self-contained.

The trick with all of these things is,

can you pick uh a slice of the complexity, you know,

eventually you want to want to model a human body,

but can you model it down to the right level of detail

and what slice can you take out of it

that will be self-contained enough?

You can kind of model

and approximate the inputs

and outputs into that self-contained system

and then just focus on the self-contained system.

So, a nucleus is quite interesting from that perspective.

Um, then the other issue is just there's not enough data yet.

So, you need data and I talked to various, you know,

top scientists about who work on electron microscopes and other imaging things.

If we could image a live cell without killing the cell,

that would be game-changing obviously cuz

then you could convert it into a vision problem

which we would know how to solve.

Right?

And but at the moment,

there are at least I don't I'm not aware of any techniques

that can give you a kind of,

you know, nanometer resolution but without destroying but it in, you know,

in a live dynamic cell.

So, you can see all the interactions, right?

You can take static images at that resolution obviously.

Really detailed now and that's quite exciting

but it's not enough to turn it just into just into a complex vision

problem.

So, that's one way it could be solved.

So, it could be a hardware driven data driven solution

or it could be

that we build better learn simulators of these dynamical systems.

So, that's that's the more modeling way of solving it.

>> You've been looking at all kinds of science and not just bio.

There's science, drug discovery, climate modeling, mathematics.

If you had a rank

which scientific domain will transform the most dramatically the next 5 years,

what's in your list?

>> all sound exciting and that's why, I mean,

that that for me has been my main passion

and always the reason why I've worked on AI for my whole career for

30 plus years now is to use AI

as the ultimate tool.

I always thought AI would be the ultimate tool for science

and to invite such advanced scientific understanding,

scientific discovery, and things like medicine,

and just our understanding of the universe around us.

So, actually, when you mentioned our original way we used to articulate our mission

statement,

which is still uh the way we think about it,

is there was two steps to it.

One was Step one was solve intelligence, i.e., build AGI,

and then step two was use it to solve everything else.

We had to change that a bit over time cuz people were like,

"Do you really mean solve everything else?"

And we did mean that,

and I think people are sort of understanding what that means today.

But, specifically, I was solve other what I call root node problems in science.

So, areas of science that would unlock whole new branches or avenues of discovery.

And AlphaFold is the prototypical example of what we want to do.

So, over 3 million researchers around the world,

pretty much every biology researcher in the world uh uses AlphaFold now.

And I was told by some of my, you know, former executive friends that,

you know, almost every drug discovered from now on will have used AlphaFold at

some point in its in the drug discovery process.

So, that's something we're very proud of,

and it's the sort of impact that we hope to have with with AI.

But, I do think it's just the beginning.

Uh I I I don't really see any area of science

or engineering that this won't be able to help be helpful with.

And the ones you mentioned, I think we're almost like an AlphaFold one moment.

So, it's we've got very promising results,

but it's not quite solved the the grand challenge yet in that domain.

But, I think we're going to have a lot to talk about in the

next couple of years on all those areas you mentioned,

materials, which I I think is very exciting, all the way to mathematics.

>> In in science, I mean, it feels Promethean.

It's like, here is this capability, and >> I think so.

I mean, of course, along with that,

including what the the the parable of Prometheus,

we have to also be careful with how we use

that and what we use it for,

and also the misuse uh that can happen with those same tools.

>> A lot of people in this room are trying to build companies applying

AI to science.

For them, what's the difference between a startup

that actually advances the frontier in your view versus one that's just wrapping an

API around a foundation model

and calling it AI for science?

>> Well, look, I think there's one of the things I would recommend.

I'm trying to think about and I think you mentioned this to me before.

What would I do today myself

if I was sitting in your place in Y Combinator,

you know, looking at things.

One thing you have to do is obviously intercept where the AI tech is

going.

So, that's one hard part of it.

But, I do think there's huge scope for combining where AI is going with

some other deep technology area.

I just think that

that sweet spot is is whether it's materials

or medicine or other really hard areas of science.

I think that those kinds of interdisciplinary teams,

especially if it involves the world of atoms as well,

there's not going to be a shortcut to that,

at least in the foreseeable future.

Those areas that are pretty safe from just getting swamped by whatever the next

update is to the foundation models.

So, I think if you're looking for things like that,

that's one of the more defensible areas I would say.

And I've always loved deep tech,

so I'm kind of biased towards deep tech things.

I think nothing that's really long-lasting and worthwhile is easy.

And so, I'm always been drawn to to deep technologies.

Obviously, AI was like that back in 2010 when we started out, right?

It was It was thought to just we know we know it doesn't work

kind of thing is what I was told by investors

and even in academia it was considered to be a very niche subject

that we sort of tried in the '90s

and we know doesn't work.

But, if you, you know,

if you have belief

and conviction in your idea why it's different this time

or what special combination from your background

that you had,

ideally you're expert in both those areas,

both the machine learning

and the other area you're applying it to

or you can create a founding team with

that expertise,

I think there's huge impact to be made there

and huge value to be built there.

>> That's of important message.

I mean, even I mean, it's hard it's easy to forget.

Like, basically, once you've done it, you've done it.

But, before you've done it, people are arrayed against you.

>> Oh, sure.

I mean, no one believes in it,

which is why I think you got to you've also got to work in

things that you're genuinely passionate about.

Like, for me, I would have worked on AI no matter what happened.

I just decided from a very young age it was the thing

that um could be the most consequential thing I could think of.

It's turned out that way, but it might not have.

Maybe we would have been 50 years too early.

And it was also the most interesting thing I could think of working on.

And so, I would have still be working on AI today even

if we were still,

you know, in a little garage somewhere and it still wasn't quite working.

I would have still been trying to find Maybe I'd have been back in

academia or something,

but I would have found some way of of continuing to work on it.

>> So, I mean, AlphaFold was like an example of a spike

that you pursued and it worked.

You know, what makes a scientific domain ripe for an AlphaFold style breakthrough?

And is there a pattern, a certain objective function?

And like >> The way I I'm I should write this up at some

point when I have 5 minutes spare,

but the lesson I've learned from all the Alpha projects we've done,

specifically AlphaGo and AlphaFold,

is um I think the techniques we have

and the problems I look like to look for are great in

if this if the situation can be described

as massive combinatorial search space.

The more massive the better in some ways.

So, no brute force or special case algorithm will will solve it.

And that's true of Go moves and of, you know, different configurations of proteins,

far more than the atoms in the universe, both of those.

And then, um you have a clear objective function.

So, you know, you can think of it

as minimizing the free energy in the proteins

or,

you know, the winning the game of Go.

So, you need to be able to you need to specify your objective function

clearly so you can hill climb.

And then, um enough data

and or simulator that can generate you uh lots of uh in distribution uh

synthetic data.

If those things are true, then I think um with today's methods,

you can go a long way into tackling

and finding the kind of needle in the haystack

that you need uh to for the solution

that you're trying to look for.

And I think of just drug discovery, by the way, in the same way,

right?

There is a compound out there

that would solve this disease

if one could find it,

if one could only find it, right?

And that wouldn't have any side effects and so on.

And as long as the laws of physics allows it,

then the question is how do you find it in an efficient way,

in a tractable way?

I think we showed for the first time, actually, with AlphaGo,

that these systems could uh find those kinds of needles in a haystack,

in that case, you know, the perfect Go move.

>> I guess uh to get a little meta, I mean,

we've we're talking about humans using these methods to create AlphaFold,

but then there's a meta level,

which is humans using AI to explore the space of possible hypotheses.

How close are we to AI systems that can do genuine scientific reasoning,

not just pattern matching on data?

>> we're close.

Um we're working on these general systems like

that like I think we have this system called Co Scientist,

and we have other algorithms like AlphaFold

that can go a little bit beyond what the basic Gemini will do.

And obviously, all the frontier labs are experimenting in this way.

I've yet to seen anything so far,

and we we all tinker with the same things, you know,

some math problems that are a little bit harder than IMO and so on.

I haven't seen anything yet um that is a true genuine, you know,

massive discovery.

That's my personal opinion.

I think it's coming.

I think it may be related to uh this earlier this thing we discussed

about creativity,

and and actually going on beyond the bounds of what's known.

So, clearly, that's just not pattern matching at that point,

cuz there is no pattern to match to,

and it's a bit more than extrapolation.

It's some kind of analogical reasoning, and I don't think these systems have that,

or at least we're not using them in the in the right way to

do that.

So, the way I often say

that in science is can it come up with a hypothesis that's really interesting,

not just solve one.

When I say just,

we're not talking about just like solving the Riemann hypothesis or something.

This would be obviously amazing, or one of the Millennium Prize problems,

and maybe we're a couple of years out from doing that.

Um but, I'd like to solve P equals NP.

That's That's my favorite one.

But, can you But,

even harder than that would be to come up with a new set of

of Millennium Prize problems

that were regarded by top mathematicians to be

as,

you know, deep and meaningful

and worthy of lifetime of study

and effort to solve.

Right?

I think that's another level harder.

And uh we don't have um you know,

I still don't think we know how to do that.

I don't think it's it's magical, though.

I do think these systems will be eventually be able to do that.

Maybe we're missing one or two things.

And then, the way we would test that is, you know,

I sometimes call it my Einstein test, which is, you know,

can you train a system with the knowledge of cutoff of 1901,

and then will it come up with you know, what Einstein did in 1905,

including special relativity, you know, his annus mirabilis.

Can Can it do that, right?

Uh and then, I think we could run that test.

May Maybe we should just run that test and keep seeing if that's possible.

And once that is,

then I think we're on the verge of these systems being able to invent

something new,

truly novel.

>> So, last last question.

For the people who are deeply technical in this room who want to work

on something,

you know, even close to the scale

that what you have created with you know,

it's one of the largest AI efforts in the world,

and you've been a pioneer for all these years.

So, for that, I think everyone in this room thanks you

and the folks at DeepMind very,

very deeply from the bottom of our hearts.

Thank you.

What's the thing that you know now about building at the frontier

that you wish you'd known at 25?

>> I think we covered some of it in terms of actually you you

work out that going after hard problems

and deep problems um is no more difficult in some ways than than going

after a shallower,

simpler, more superficial problem.

They're they're they're just differently difficult.

There's different things that are hard about each of those things,

but I think given life's very short and you know,

you only have so much time and energy,

you might as well put your life force into something

that will really make a difference

if you hadn't done it,

if you hadn't been there to push it.

So, I would just think of it through that lens.

And then the other thing is

if you're if you are

and then we talked about deep tech

and I love interdisciplinary uh work

and I think that's going to be even more prevalent in the next few

years in combinations of fields

and uh finding the the the the connections between those fields.

And it's going to be even easier to do that with AI.

And then the only other thing I would say is if, you know,

if you have your depending on what your AGI timeline is, you know,

mine's like 20 30 or something like this,

then if you start off on a deep tech journey today,

usually that you're talking about a 10-year journey for for true deep tech in

my opinion.

So, then now you have to just consider AGI appearing in the middle of

that journey.

So, what does that mean?

It doesn't it's not bad necessarily, but you have to take that into account,

right?

To will it be able to leverage it?

What will the AGI system do with it?

And it goes a little bit back to what you said earlier about AlphaFold

and general AI systems.

So, one thing I can think see happening is Gemini, Claude,

or one of these general systems making use of AlphaFold like specialized systems

as tools.

I don't think we're going to have it just in one giant brain cuz

it will have too much regression in

if I put all the proteins into,

you know, Gemini, that wouldn't make sense.

We don't need Gemini to do protein folding.

Going back to your information efficiency,

it will definitely affect its language skills or something like that, right?

In a bad way.

So, much better I think is to have really good general purpose tool usage

models that will then maybe they could even train those specific tools,

but they would be in a separate system.

So, I think that's kind of interesting to think through the implications of

that and then what you might build today.

Also, physical things too like what kinds of factories would you build,

what sorts of you know, finance systems and so on.

So, I just think you need to really take

that seriously and in in in on the one hand is like an imagine

what that world would look like

and then build something

that would be useful

if that comes in halfway through.

Demis Hassabis everyone.

>>