Hang tight while we fetch the video data and transcripts. This only takes a moment.
Connecting to YouTube player…
Fetching transcript data…
We’ll display the transcript, summary, and all view options as soon as everything loads.
Next steps
Loading transcript tools…
Demis Hassabis and Veritasium's Derek Muller talk AI, AlphaFold and human intelligence | The Thinking Game Film | YouTubeToText
YouTube Transcript: Demis Hassabis and Veritasium's Derek Muller talk AI, AlphaFold and human intelligence
Skip watching entire videos - get the full transcript, search for keywords, and copy with one click.
Share:
Video Transcript
So I think one of my favorite parts of
the film is that part where the team has
just told you, well we could just find
the structures for all the proteins and
just release those. I and then you
release them to the world and you see
the the map of the globe light up as
people in real time are getting all of
those structures. What was that like?
Tell me what was the what was the
feeling that you had? I mean look there
there was so many amazing moments and
the team will remember this of um but
that was one of the big highlights. It
was very satisfying to see that this
sort of idea that we maybe if we crack
this really important problem you know
potentially millions of researchers
around the world will will make use of
it and um you know to see that sort of
lighting up all across the globe um is
really a kind of humbling and amazing
experience. I came here for the uh AI
for science forum which you held and I
think this the thing that shocked me is
that for 50 years the work of tens of
thousands of scientists revealed the
structures of 150,000 proteins. That was
the grand sum of human effort. And then
in a few years your team small 15 20
people was able to find the structures
of 200
million. Yeah. Well look I mean first of
all the first thing to say is we
couldn't have done it without the first
150,000 right. So that incredible, you
know, we need to thank the the
structural biology community, you know,
thousands of researchers painstakingly
putting together these structures using
very exotic and pretty expensive and
complicated equipment um over 50 years
like you say and the sum totals 150,000
but it was enough to uh kickstart us to
be able to create a system like Alphold
to learn from those 150,000 and then
actually uh learn further from its own
predictions, the best ones of its own
predictions and sort of feeding that
back into the system and then eventually
being good enough to kind of understand
something about protein, something
fundamental about protein structure. So
then eventually we could do all 200
million and and I think as John says in
the film, you know, it usually takes a
PhD student their their whole PhD.
That's kind of a rule of thumb of like
to find the structure of one protein. So
you know 200 million times 5 years a
billion years of PhD time which is quite
something you know to have done in in a
year. See like I feel like I didn't get
it before I came here and I heard those
numbers and I was like oh things have
fundamentally changed and I don't think
the world gets it yet. Um so I think
that's one of the exciting things about
this film. And I think, you know,
another thing that's really important to
keep in mind is you figured out all 200
million now they're out there, but the
discoveries and the breakthroughs that
are going to come from that, they're
going to take decades, but we are going
to be reaping the the rewards of that
for for decades, centuries, I think. So,
I mean, it's sort of opened up uh and
and this is why we put out there into
the world. We knew we could only think
of a tiny fraction of what the entire
scientific community might do with it.
And it's really gratifying to see the
whole range of things that people are
already doing. Over two and a half
million researchers from pretty much
every country in the world working on
their really important biology and
medical uh problems and making great
progress with that. And and right now I
think it's super well known in the
scientific community but as you say I
don't think it's it's appreciated yet in
the general public what this is going to
do. And I think that will come in the
next 5 10 years as we start getting uh
you know AI designed drugs that were
helped by things like Alpha Fold and
many many other amazing things for
society that will come as a downstream
consequence of us knowing what these
structures are. Now can you think of any
examples that have happened since the
film? Uh well there's many in fact a few
of them were were were mentioned in
those headlines you know these these
ideas of um designing enzymes which are
types of proteins you know that catalyze
certain reactions and maybe we could uh
modify some of these enzymes to help
deal with some environmental issues we
have like the amount of plastics in the
oceans or perhaps doing even carbon
capture things like this um I think
incredible opportunity and obviously the
main reason I was I I was interested in
doing uh protein folding was to
accelerate drug discovery
And uh and we spun out a company, a
sister company called Isomorphic Labs
that actually is developing other
technologies around AlphaFold and and
the newer versions of AlphaFold to
actually start not only do you know do
you understand the structure of a
protein, but then you could design a
drug compound to bind to the right part
of the protein surface once you
understand what it structure, what its
function is. And that's the beginning of
understanding disease and maybe trying
to cure some of these terrible diseases.
And we're working on, you know, uh,
cancers and cardiovascular diseases, all
sorts of things, you know, more than a
dozen drug programs. And one day, I
hope, uh, you know, we'll be able to
reduce drug discovery down from taking
like 10 years on average to go from
understanding a target to having a drug
in in the clinic to, you know, maybe a
matter of months, perhaps even weeks,
just like we did with the protein
structures. Yeah, that's extraordinary.
I wanted to ask you about your origin
story. um you know something that
occurred to me well here's my thinking
right AI in a way is not new dates back
to the 40s and maybe 50s and it went
through a series of sort of booms and
then busts or AI winters as people refer
to them um I think in the film you said
there's no point in being born you know
ahead of your time 50 years ahead of
your time so I think that my question
for you is when you were graduating from
Cambridge that was kind of an AI winter
Um, did you see something that other
people didn't see that led you to know
the time for AI was coming or were you
just obsessed with this idea of
intelligence and just ridiculously lucky
to be born in this moment?
Well, look, it's a bit of both, I would
say. So, and actually there's many
people in the audience, many of my
colleagues and friends who've been with
me that almost that entire journey. you
saw some of them, David Silver, Ben Copy
and Shane Le and um they'll remember
this very well and Tim Stevens and it's
um look I I have to be honest I would
have done it no matter what because um I
when I was growing up and you saw that
with the chess and other things I just
felt that intelligence and and therefore
artificial intelligence was the most
fascinating thing one could work on. I
always wanted my passion was was to try
and understand the universe around us
you know sometimes call it the nature of
reality all the big questions. So
physics was my favorite subject at
school and all the big physicists
Richard Feman and Steven Weinberg all
the great physicists Carl Sean. Um but I
sort of thought that we needed another
helping hand like a tool that could help
us help us as human scientists
understand the world better around us.
And uh and that for me was obvious to me
from the beginning as I was when I was a
teenager that um it would be AI and it
would be the you know not only the most
uh maybe most powerful tool to help us
do science but the most interesting uh
thing to develop in itself you know
interrogate what intelligence is and try
to understand what it is uh and while
you're trying to build something that is
intelligent. So I think I was always
going to do that. Um but also when you
look at these AI winters and you look at
the state of technologies you find it
you have to have a good reason why you
think you might be able to try it in a
new way those winters are are in a way
learning you know opportunities to learn
why did those methods not work those
deep blue methods that we saw that beat
Gary Kasparov amazing they could win the
chess but really was a little bit of a
dead end because they were hard
programmed hardcoded to only do that one
thing play chess so it wasn't some sense
was missing the essence of intelligence
in in many ways this this general
generalness and this learning cap
capability and we knew we had these
these techniques they were very nent you
know neural networks became deep
learning and then reinforcement learning
as you heard we we knew those techniques
um could potentially scale why did we
know that because actually the the the
human brain is a form of those you know
we're a neural network obviously that's
what inspired neural artificial neural
networks in the first place was was was
you know neurons in the brain
And reinforcement learning is one of the
main ways that animals including humans
do learn. You know the dopamine system
in the brain implements this form of
reinforcement learning. So you know in
the limit this must be possible using
these types of learning techniques. But
of course you don't know at that point
if you're 50 years ahead or not right
with your time. But I just want to be
clear on what you're saying. In essence,
you're saying that the AI models that
you're currently working with are in
some sense analogous to the human brain
or the human brain is analogous very
very loosely speaking they're inspired
by the same types of techniques and
approaches uh you know biological
learning systems use right that's the
key it's the learning and the generality
do you think then at some point AI will
be conscious well that's a that's a huge
question and and obviously you know we
have to you know they're not not
necessarily agreed upon definitions of
consciousness obvious Obviously there
are aspects of it like self-awareness
and things that are agreed upon. Um I
think that's part of the I always felt
actually answering that question was one
of the things that will come about being
on this journey with AI trying to build
artificial minds and then comparing them
to what we know about about uh uh the
human brain and then seeing what the
differences are if any and those
differences might tell us what uh and
certainly help us understand our own
minds better. things like dreaming,
emotions, creativity, and things like
consciousness, all the mysteries of the
mind uh and uh and then uncover help us
understand them and then maybe
understand how special they are to the
substrate that we're in. You know, we're
carbon based versus the silicon based
systems that we're building.
You started DeepMind here in London and
you had certain forces, investors maybe
trying to pull you to Silicon Valley,
but you resisted. Tell me what it was
about this place or the culture that
that made you want to stay here. Well,
look, I I I've been I I was born in
London. I've lived in London my whole
life and you know, I think there's a lot
of amazing things about the cultures
that I was immersed in. You know, you
saw me going to Cambridge and the sort
of golden triangle of Oxford, Cambridge
and Imperial as we're nearby and UCL,
all these august institutes. I think um
the UK has always been very strong in
science and innovation. We punch well
above our weight. There's also obviously
a rich history in computers with Charles
Babage and Alan Turing. So I feel we're
trying to carry on in that tradition.
But there was some practical reasons.
One is that uh uh I at the time when we
started in 2010, there was a lot of
talent trained by these top places that
um unless they wanted to go and work for
a hedge fund or something in the city in
finance, they wanted to do something
really intellectually challenging. There
weren't there aren't that many companies
doing that kind of stuff in the UK or
actually in Europe really. So I felt
that we could um gather a lot of talent
together very quickly that was probably
being underutilized in in Europe and
that that's how it transpired. But the
second reason was that I think AI is so
important. It's going to affect the
whole world. Obviously you've heard me
talk about in the film that you know I
think it's going to be one of the most
important things ever invented. I felt
that I do think it's needs the
international sort of approach and
cooperation around what we want to do
with this technology. how we want it to
be deployed, how we want it to um affect
our society. I it's going to affect
everyone in all countries. Um so I don't
I think it needs to be uh uh built with
more uh voices and stakeholders uh than
just sort of 100 square miles of um
California, you know, in Silicon Valley
and also beyond technologists and the
scientists just building it. think it
needs um social scientists, economists,
psychologists, you know, governments,
academia, all to be involved um in in in
defining how this this this enormously
transformative technology should go.
Yeah. Well, it's clearly going to be
very powerful and one of the issues that
the the film addresses is the morality
and ethics around that and I think
particularly the safety of it. What
keeps you up at night when you think
about AI? Well, many things and and um
you know, I don't get much sleep these
days, but I I for many reasons, but I
think um Shane and I, you know, will
remember this is that we we actually uh
when we started out 2010, um it's only
15 years ago. It's kind of amazing to
see how the world's changed. And in
2010, no one was talking about AI.
Nobody was doing industry. Um but we
knew that this was a, you know, this had
the kernel of something incredibly
important. And uh and we planned for
success. So, we thought it was going to
be a 20-y year journey and often when
you do that in technology and in
startups and and hard sciences that that
it always stays 20 years away, right?
So, somehow, but for us, it's it was
actually it really has been 20 years and
we're sort of 15 years in now. Um, and
we planned for success, but we knew that
success meant all these amazing things,
curing diseases, you know, um, solving
energy crisis, climate, using AI to
help, all of these things. Um but also
it came with these risks, risks of harm,
enormous risks of misuse. And so from
the beginning we've been very cognizant
of that responsibility. Um but also
trying to push that debate and be role
models about how to develop this
technology in a responsible way. Is this
potentially unstable in that you could
have a hundred companies who have the
utmost ethics and morality and they
think about safety to an extreme level
and you have one actor who doesn't.
Yeah. And then it ruins it for everyone. Yeah.
Yeah.
Well, that's the huge that's one of the
huge risks that I worry about today is,
you know, so-called race dynamics,
right? Race to the bottom. You know
there's many uh examples of this in
history right and even if all the actors
are good in that environment let alone
if you have some bad actors you know
that can drag everyone to to to to rush
too quickly to um cut corners these
kinds of things because in individual
it's a sort of tragedy of the common
situation for any individual actor sort
of makes sense but but as an aggregate
it doesn't and um and I've been saying
that for a long time and Shane and
others many others uh Helen and people
work on respons responsibility at deep
mind. We've been talking a lot about
this and that's why I was so pleased to
see some of these international summits
being set up. the first one in the UK,
Bletchley Park, and then just recently
in Paris uh that Macron hosted,
President Macron. And I think we need
those kinds of uh international debates
uh about where this is going and um and
one of the big problems is how do we uh
give access to these technologies and
you've seen with AlphaFold you know open
to the world open science obviously
that's better for progress than amazing
uh all the all the all the good
researchers and the good people around
the world can can can build on top of
that work and do amazing things with it
but at the same time you want to
restrict access to to that same
technology to would be bad actors
whether that's individuals or even rogue
nations and it's very hard balance to
get right like it's there's no one's yet
got a good answer for how you you know
do both of those things I think
initially I was encouraged by the amount
of effort required to develop AI so
there's many references in the film to
the Manhattan project and I think it's
one of the benefits of nuclear weapons
that in order to develop them you
actually need basically state
sponsorship or you know a huge huge
undertaking and initially AI looked the
same way. This is going to take you know
the huge tech companies or or states to
develop this but lately there's these
new developments like deepseek and
there's an Alibaba model and they look
much more sort of thrifty. Yeah. Which I
think there could be a fear that that
really democratizes the access to this
technology increasing the probability of
a bad actor. Yeah. So look that that you
you know you're exactly right and I feel
like this it's it's sort of it's very
good on the one hand you know more
people accessing these technologies um
you know hobbyists you know kids like I
was back when I was tinkering around
with theme park can now you know uh uh
work on some really interesting AI
systems and probably come up with
amazing new uh uh applications. Um but
yeah it's it's sort of uh it's you know
it's it's it's available to everyone and
it is worrying and I feel like you know
maybe we need some new uh uh approaches
you know where maybe uh the market
environment or something else is set
where it kind of incentivizes the right
behavior right so you know I was talking
to some economist friends of mine and
maybe they need to get involved now to
set up the right incentive structures so
that uh actually the players that and
the actors that are are are have the
right intentions, you know, backed by
government society are actually the ones
that that get successful and and those
AI systems are more powerful and and and
more productive. Um, and maybe we have
to start thinking about those kinds of
approaches to deal with the practicality
that we're in, which you know, I'd much
rather there be a a calm CERN-like
effort towards AGI, these final few
steps, but given the geopolitical
framework we're in, maybe that's not
possible. So, we have to be more
pragmatic about it. For sure.
In the film, you talk about how the
future will be radically different. So,
I want to ask for myself and for
everyone in this audience. Given that
you were one of the leaders at the
forefront of this, what do you think the
world will be like in 5 to 10 years? Do
you do you have an outlook on that? And
I guess further to that, I have four
kids. I'm like, what do I do with like
do do I send them to school? Is that
even worthwhile anymore? Like, so so you
are the guy that I want to ask this
question to more than anyone in the
world. Sure. Well, let's start with the
same question. For sure. send them to
school. I I I I I say my kids too. Look,
I think the next 5 10 years is going to
be um it's a bit you know what I would
say to kids these days is embrace the
new technologies and as parents I think
let your kids play with them that
they're coming and they're going to
increase productivity, creativity. I
think it's going to be amazing. It's a
bit like my era, my generation with
computers, the advent of computers, you
know, there was a lot of fears about
that too and even gaming. And then um
people work out, you know, if you're
growing up with that, it feels natural
to you, second nature. And then um
they're often the ones that can extend
it into new ways we couldn't even dream
of today. So I think a lot of that's
going to happen. So I still think it's
important to do maths and computer
science because you'll be best placed to
take advantage of these frontier
technologies and use them in new ways.
So um so that I don't the recommendation
I think is the same as it's always been.
um maybe just be prepared that things
are going to move even faster and to
learn you know about adapting and
learning to learn actually learning
quickly to adapt to a new technology
that's going to come out it seems like
almost every week uh in terms of like
society what I see happening is I mean 5
10 years is a long time in AI hard to
predict that far ahead but um what I
certainly imagine in the areas of
science is I think a new renaissance
almost a new golden age which I hope
alpha fold is just the beginning of um
of us understanding and making lots of
breakthroughs in many areas of science
uh and helping us with all the biggest
questions, you know, from curing all
diseases to helping with uh new energy
sources and and and climate. And I think
we're going to start seeing that all in
the next 10 years. That's extraordinary.
Well, I look forward to it. I hope you
do as well. Uh we're going to leave it
there. Um but yeah, congratulations on
all your great work and winning the
Nobel Prize and it's just tremendous. Seriously,
Click on any text or timestamp to jump to that moment in the video
Share:
Most transcripts ready in under 5 seconds
One-Click Copy125+ LanguagesSearch ContentJump to Timestamps
Paste YouTube URL
Enter any YouTube video link to get the full transcript
Transcript Extraction Form
Most transcripts ready in under 5 seconds
Get Our Chrome Extension
Get transcripts instantly without leaving YouTube. Install our Chrome extension for one-click access to any video's transcript directly on the watch page.
Chrome Extension