---
title: 'Video HA4_2rqrP70'
source: 'https://youtube.com/watch?v=HA4_2rqrP70'
video_id: 'HA4_2rqrP70'
date: 2026-07-03
duration_sec: 1235
---

# Video HA4_2rqrP70

> Source: [Video HA4_2rqrP70](https://youtube.com/watch?v=HA4_2rqrP70)

## Summary

Demis Hassabis, CEO of DeepMind, discusses the timeline and implications of Artificial General Intelligence (AGI), predicting a high probability of its arrival within five years. He describes AGI as a system with all human cognitive capabilities and compares its impact to ten times the Industrial Revolution, unfolding at ten times the speed.

### Key Points

- **Definition of AGI** [00:00] — AGI is defined as a system that exhibits all cognitive capabilities of the human mind, using the brain as the only existence proof of general intelligence.
- **AGI Timeline** [00:26] — Hassabis states there is a very good chance of AGI within the next five years, consistent with DeepMind's original 2010 prediction of about 20 years.
- **Impact of AGI** [01:04] — Hassabis quantifies AGI's coming as 10 times the Industrial Revolution, happening over a decade instead of a century.
- **Compute as Bottleneck** [02:31] — Compute is the biggest bottleneck, needed for scaling up systems and running experiments. The cloud serves as the workbench for testing new ideas.
- **Scaling Laws Not Plateauing** [03:02] — Hassabis disagrees that scaling laws are plateauing; returns are still substantial, though slightly less than at the start of scaling.
- **Areas Ahead of Expectations** [03:45] — Most domains are ahead of expectations, including video models and interactive world models like Genie. However, continual learning is still missing.
- **Missing Capability: Continual Learning** [04:11] — Systems do not learn after training; integrating new learning into existing models is a major unsolved problem, analogous to brain consolidation during sleep.
- **DeepMind's Breakthroughs** [05:31] — Hassabis claims about 90% of modern AI breakthroughs (e.g., AlphaGo, Transformers) came from Google Brain, Google Research, or DeepMind.
- **Other Missing Capabilities** [06:29] — Beyond continual learning, missing capabilities include different memory systems, long-term planning, hierarchical planning, and consistency (jagged intelligences).
- **Widening Gap Among Labs** [07:47] — The gap between leading labs (3-4) and others is widening, as new algorithmic ideas become harder to find and implement.
- **Open Source Models** [08:39] — Open source models are typically one step behind the frontier (about 6 months), but DeepMind supports open science with models like Gemma for small developers and academics.
- **Future of LLMs** [09:52] — Hassabis believes foundation models will not be replaced but built upon, with AGI systems likely incorporating them as key components.
- **Positive Vision: Scientific Discovery** [10:36] — AGI will be the ultimate tool for science and medicine, potentially leading to a golden age of discovery, including curing diseases like his mother's multiple sclerosis.
- **Drug Discovery Challenges** [11:13] — AI can help design drugs and simulate metabolism, but clinical trials remain a bottleneck. Hassabis envisions a future where AI predictions are trusted enough to skip steps like animal testing.
- **Two Main Worries** [13:12] — Hassabis worries about misuse by bad actors (dual-purpose technology) and technical control of increasingly autonomous systems, requiring international regulation.
- **Regulation and Certification** [14:31] — He advocates for minimum standards, benchmarks for undesirable properties (e.g., deception), and an international certification process, similar to the Atomic Agency.
- **Labor Displacement** [17:28] — AGI will cause massive job disruption, but new, higher-quality jobs will emerge. However, the change will be 10 times faster than the Industrial Revolution, requiring better mitigation of downsides.
- **Short-term Hype vs Long-term Revolution** [19:02] — Hassabis notes that AI is overhyped in the short term (next year) but still underappreciated in its revolutionary impact over a 10-year timescale.

### Conclusion

Demis Hassabis presents a sobering yet optimistic view of AGI, emphasizing its imminent arrival within five years and its potential to revolutionize science and medicine. However, he also highlights critical unsolved problems like continual learning, the need for international regulation, and the risk of misuse, urging proactive measures to mitigate downsides.

## Transcript

We think very consistent how we define AGI as basically a system that exhibits all the cognitive capabilities the human mind has. And that's important because the brain is the only existence proof we have that we know of, maybe in the universe, that general intelligence is possible.
So that, to me, is the bar for what AGI should be. It's the worst question. How close are we? Well, everyone says different things. And it's very difficult when you have very prominently the same few years earlier, this is 2026, 2027.
happens yeah i mean i think look i've got a probability distribution around and the timings but i would say there's a very good chance of it being within the next five years so a lot of old jobs go away or not vibrant anymore but then actually the history of it is that my whole set
of new jobs arrive that maybe one can't even imagine before and those are higher quality higher paying so that's the normal course of course you have to be very careful to say this time is different and i guess that's what people like marco claiming is like you know it's the
same as the last sort of 10 massive breakthroughs like the internet mobile and so on i do think this is going to be bigger than all of those previous technological breakthroughs. I sometimes quantify like the coming of AGI is like 10 times the industrial revolution. Guys, that's Demis
Hassabis, the man whose own lab built most of the breakthroughs sitting inside every AI tool you use right now. And he just told us this isn't slowing down. It's coming 10 times faster than the last time the world got turned upside down. Now, listen to the rest of what he said. It's the worst
questions. How close are we? Well everyone says different things and it's very difficult when you have very prominent figures saying two years earlier than 2026, 2027. Yeah I mean I think look I've got a probability distribution around the timings but I would say there's a very good chance
of it being within the next five years though it's not long at all. Is that closer than you thought? Has that changed over time? Well really I mean actually my co-founder Shane Legg who's a few scientists here when we started out DeepMind back in 2010 he used to write blog posts sort of
predicting about when AGI would happen and bearing in mind in 2010 when we started almost nobody was working in AI and everyone thought AI was the great idea, no one was reading it, no but they're still there on the internet for people to check and we used to do this extrapolation of compute
and algorithmic progress and basically we predicted around 20 years it would take from when we started out and I think we're pretty much on track. What are the biggest bottlenecks when you look today in the documentary research you just never have enough compute? What are the biggest bottlenecks
when you look at where we are today. I think compute is the big one, not just for the obvious reason of scaling up your ideas and your systems as the scaling laws, as they're called, keeping on building bigger and bigger architectures
with more and more parameters. And as you do that, you get more intelligent systems. But the other thing you need a lot of compute for is for doing experiments. So the computers, the cloud is our workbench, basically. So if you have a new idea, a new algorithmic idea,
but you want to test it, you kind of got to test it at a reasonable scale. otherwise it won't hold when you actually put it into the main system. So you need quite a lot of computers if you have a lot of researchers with lots of new ideas. You mentioned the word scaling rules.
A lot of people suggest that we're hitting scaling rules and we're starting to see that plateauing effect. Do you think that's true? No, I don't think so. I think it's a bit more nuanced than that. So, of course, when the leading companies all started building these large language models,
you're getting enormous jumps with each generation of new system, and maybe they're almost doubling in performance. And at some point that had to slow down. So it's not kind of continuing to be exponential. But that doesn't mean there isn't great returns still for scaling the existing systems up further.
So, yeah, and we and the other Frontier Labs are getting a lot of great returns on that kind of compute expansion. So I would say the returns are kind of still very substantial, although they're a bit less than they were, obviously, at the start of all of this scaling.
Where are we behind where you thought we'd be? I think actually in most areas we are ahead of where I thought we would be. If you think about things like the video models or even now with our newest systems like Genie,
they're interactive world models, which I think is kind of incredible if you sort of step back and think about it. I think if you'd shown me that five, ten years ago, I would have been pretty amazed. So I think in most domains we are ahead of where the field thought.
There's still some big things missing though, like continual learning. These systems don't learn after you finish training them, after you put them out into the world. they're not very good at learning further things. And I think some critical capabilities are like that.
Why do we not have continuous learning? Well, people haven't quite figured out yet, and all the leading labs are working on this, how to integrate new learning into the existing systems that you spent months training.
So, of course, the brain does this very elegantly, right? And probably through things like sleep, reinforcement learning. So you just kind of get consolidation, it's called, in the brain, where your memories during the day are replayed,
and then some of that information is elegantly incorporated into your existing knowledge base. And perhaps we, I thought for a while, maybe we need something like that to incorporate new information along with the existing information base.
So back in 2010, when nobody believed in AI, his own team predicted this exact moment, and now he's telling us they're on schedule. That's not hype talk. that a track timeline that actually held up for 15 years straight The only thing standing between where we are and a system that keeps teaching itself forever is one missing piece and every major lab on earth is racing to find it right now
Now listen to this. We made some organizational changes, so I think we've always had the deepest and broadest research bench at Google and at DeepMind. I mean, if you look at the last decade or plus, you know, 15 years, but I would say
About 90% of the breakthroughs that underpin the modern AI industry were done either by Google Brain or Google Research or DeepMind. So one of our groups, if you think of like AlphaGo and reinforcement learning and, of course, Transformers, you know, these are all the key breakthroughs.
So I would back us to sort of make those breakthroughs in the future if there are any missing ones. And I think we basically helped put together all the talent from around the company sort of pushing in one direction. And then we talked earlier about, you know, compute resources.
it was also about combining all of our resources together so we could build the biggest models rather than having two or three versions around the company. So I think a lot of it was assembling together all the ingredients we already had and then kind of pushing with relentless sort of focus
and pace, acting almost like a startup really to get back to the frontier and be ahead in many areas. You say if anyone's going to do the breakthrough, it could and should be us. Yeah. When you think about that, if continuous learning is the next breakthrough that you're most excited
I think there's quite a few things that are missing. There's continual learning. I think a lot of mileage in looking at different memory systems. At the moment, we have these long context windows, which are kind of a bit brute force. You just put everything in them.
And then there's stuff like long-term planning, hierarchical planning. These systems are not very good at planning long time horizons, many years into the future, which we as, with our minds, we can do. So there's quite a lot of problems, I think, that's still left to overcome.
Maybe one of the biggest is consistency. So, you know, I sometimes call these systems jagged intelligences because they're really amazing at certain things when you pose the question in a certain way. But if you pose a question in a slightly different way, they can actually still fail at quite elementary things.
So a general intelligence shouldn't be that sort of jagged. When you reposition files and you set up agents to perform in certain ways. Yeah. And then the files no longer configure. It completely falls over. Sure, sure. Exactly. A hundred percent. That's a disaster.
Yeah. Well, I mean, the general intelligence, you know, if you think about how our minds work, it shouldn't have those kinds of holes in it. We've said about plateauing and scaling. Everyone talks about a commoditization of models in terms of capabilities.
Do you think we see that or do you think we see ones to continue to accelerate ahead of the others? Yeah, I feel like maybe, you know, the three or four leading labs now, which we're one, I think the gap is sort of starting to pull away.
Because a lot of these tools also, of course, help you build the next generation. so things like coding tools, math tools. And it's getting harder and harder, I would say, to kind of eke out the same gains from just the same ideas.
So I think those labs that have capability to invent new algorithmic ideas are going to start having bigger advantage over the next few years as the last set of ideas are sort of all the juices being run out of them.
So one company is telling us they built 90% of the ideas behind this entire industry and the gap between them and everyone else is getting bigger, not smaller. In the same breath, he admits these systems still fail at basic stuff the moment you phrase a question differently.
We're putting more and more power into fewer hands, building something that's still full of holes nobody fully understands. Listen, what does our future look like? Yeah, I think it's probably similar to what we're seeing today.
I mean, we're big supporters of open science and open models. And we've done many, many things, obviously, from the original Transformers to AlphaFold. You know, these are all things we've sort of given out into the world and to help the research community.
And we plan to continue to do that, especially in applied domains, you know, scientific domains, applying AI to science, which is obviously my passion. But I think increasingly, from what you're going to see, is the open source models are probably one step back from the absolute frontier.
It usually takes about six months for the open source community to sort of reimplement and figure out what those ideas are. But we are also pushing hard on a kind of suite of open source models called Gemma, which are, you know, we're determined to kind of make best in class for their sizes.
So specifically for small developers or academics or, you know, the beginnings of a startup, I think they're perfect for that and also edge computing too. So we're very interested in open source models for certain types of applications.
How do you think about the world post-LLM? You have different people with different views. You have Yalda Coons with very different views. For me, I don't think it's to kind of disagree with Jan on a few things in terms of, I think there might be, there's a 50-50 chance there's some things maybe missing that we still need to make breakthroughs in, perhaps their world models, these kinds of approaches.
But my betting is, pretty strongly, is we've seen how successful these foundation models have been. They can do incredibly impressive things. I don't think that's going to go away. We're still seeing, you know, gains from returns from the scaling laws.
So I think the only question really is when you think about a future AGI system is an LLM foundation model going to be the key component only or is it the total system So I just think it a question of is anything else needed I don think it going to get replaced I think it going to get built on top of these foundation models just like the way we do with our world models
When we think about that future, five years down the line, potentially with AGI, what does that world look like? Many people have different concerns. Yeah. If we just start generally, what does that world look like? Well, I think on the positive side
and the things, obviously, I spent my whole career in life building towards AGI is, I think it will be the ultimate tool for science and medicine. So in terms of advancing scientific discovery, finding cure for diseases,
I think we need that kind of technology. And so I'm hoping in five years plus time, we'll be sort of entering a new golden era, golden age of scientific discovery. My mother's got multiple sclerosis, but it's the thing that I'm always most excited about.
The thing to worry about is actually drug discovery, the process of getting it through all the trials, and knowing that it takes a decade before my mother will actually get any benefit. How do we solve that?
I think we'll get to that point soon. First of all, what we're doing is, you know, after we did the AlphaFold project to do protein folding, then we spun out a company called Isomorphic Labs, which is doing extremely well. And that is supposed to, you know,
the idea there is we're focusing on solving the rest of the drug discovery process, which is a lot of chemistry, designing the compounds, checking it's not toxic, and all the different properties you need for drugs to be safe. I think we'll have that whole drug design engine ready
in the next five to ten years, then you're right. The next problem is the clinical trials still take many, many years. But I think AI can help there in terms of maybe simulating parts of the human metabolism,
also stratifying patients to make sure that certain patients get exactly the right type of drug that's suitable for their genomic makeup. And so I think AI can help there too. But I think the real revolution will come when a few, maybe a dozen or so,
AI drugs get through the whole process and then the government and the regulatory body see that and they have enough data to sort of back test the predictions of those models and then maybe what we can do will be in the future where maybe 10 further years where we can really just trust
the predictions that the models are making and actually then maybe skip out some steps perhaps like the animal testing is not needed anymore maybe we can go up the dosage ladder quicker because you can rely on these models listen to what he actually said there the end goal
in his own words, is to trust the model enough that we start skipping safety steps like animal testing completely. That's coming from the one man on earth with the most personal reason to want this to work fast.
His own mother is waiting on it. So if he's already talking about cutting corners, what happens when a company with way less at stake gets there first? Hear this. I think it was, I don't know, I watched it last night over dinner.
Yeah, it was a great watch, which is obviously the Doctor Manifest. I think it was Stephen Horne who said we must get it right. because we might not get another chance. Do you think that's right? Yeah, I do think that's right. I think that is the stakes that we have to deal with.
And there's two things I worry about. One is the misuse of these systems by bad actors and they can be repurposed. These are dual-purpose technologies. They can be used for incredible good in science and health, as we just discussed,
but they can also be repurposed for harmful ends by a bad actor. So that's one issue. Second issue is the technical one, making sure these systems, as they get more powerful, not today systems but maybe in a year or two times when they become more agentic more autonomous as
we get towards AGI can they be kept on the guardrails that we want and I think the right kind of regulation could help here in terms of making sure there's at least sort of minimum standards from all of the leading providers but it needs to ideally be a kind of international
standards. What is the right kind of regulation and again I'm quoting yourself back in this eventually you're right I think we need more global coordination which worries me because we're getting worse at it. Yes. Which I think would be an unwavering truth.
Yeah, for sure. I mean, it's sort of crazy the timing that we're in, right, with this most consequential maybe technology the world's ever seen, but at the same time it's a very fragmented sort of international system.
And it's not ideal, but I think we're going to have to try and do the best we can to at least come up with a sort of set of maybe minimum standards, some benchmarks to test for undesirable properties, for example, deception.
Nobody should be building systems that are capable of deception because then they could be getting around other safeguards. And then I imagine, you know, if things go well, some kind of certification process that basically is almost like a kite mark of quality,
that this model has certain safeguards and certain guarantees. And so therefore, consumers and companies can safely sort of build on top of it. And I think that is how it should go, ideally.
but it does have to be international because, of course, these systems are cross-border and, you know, they're cross-territory. Who is that ultimate verification system? I mean, you obviously started the theme park.
Yes. I love where I'm going. Brilliant. But, you know, obviously it's a media company. I go through any media platform and say, I don't know what's real or fake. I'm always having to ask what's real or fake. Yes. Who is that arbiter of verification Yeah Well I think there are I mean ultimately it got to be government I think But you know the kind of technical bodies that would be able to do the technical work would be like maybe the AI Safety Institutes You know there a very good one in the UK that
you know, was set up under Prime Minister Sunak. And I think he's doing great work. And then there's one in the US. And maybe some of the leading countries that have the best research should also have an equivalent body that is staffed with high quality researchers too,
that can actually evaluate and audit these kinds of systems against certain benchmarks and kind of like independently check whether they are meeting the right standards. If I were to give you like a magic wand
that was only applicable to AI safety, what would be your implementation idea, program that you would put in place with this magic wand? Yeah, I think we need some kind of international body,
maybe similar to the Atomic Agency, something like that, that perhaps the AI safety institute sort of feed into and the research community has to also do this and be involved in like what are the right set of benchmarks to check what types of traits what types of capabilities maybe there
are other safeguards too like you know it would be desirable to have ai systems output tokens that are not human readable so you know some kind of machine language that we couldn't understand i think that would you know introduce a new vulnerability there's quite a few sort of
things like that which i think most of the leading labs would agree are probably not best to do and And then these bodies would, you know, these institutions would test against those things. And I think that would give the public confidence and academia could be involved as well, as well as civil society,
that these systems, which are going to get incredibly powerful, have been independently checked and audited. Guys, he just said the safest path forward needs every country on Earth agreeing on the same rules at the exact moment.
The world is more divided than it's been in decades. and he casually mentioned that these systems could start talking to each other in a language we can't even read. That's not some far-off worry. That's something he's flagging as a real concern right now, while none of it is actually in place.
Now, listen to this. How do you think that the labor displacement problem, when you look at how truly capable these systems are, and what that does to labor markets? Well, certainly, in the past, with every new revolutionary technology, there's been a lot of jobs disruption.
so that's for sure and I think that's definitely going to happen so a lot of old jobs go away or not viable anymore but then actually for the history of it is that a whole set of new jobs arrive that maybe one can't even imagine before
and those are higher quality, higher paying so that's the normal course of course you have to be very careful to say this time is different and I guess that's what people like Mark are claiming is the same as the last 10 massive breakthroughs like the internet, mobile and so on
I do think this is going to be bigger than all of those previous technological breakthroughs. I mean, I sometimes quantify the coming of AGI as like 10 times the Industrial Revolution at 10 times the speed. So unfolding over a decade instead of a century.
So if you, you know, I've been reading a lot about the Industrial Revolution, a lot of great books about it. And that caused a huge amount of upheaval as well as a lot of advances. I mean, we wouldn't have modern medicine today. Child mortality was at 40% back in the Industrial Revolution.
So things, you wouldn't want it not to have happened, but ideally this time around, we mitigate some of the downsides a bit better than we did during the Industrial Revolution. I often listen to amazing voices like yours, and I get very excited about how fast it's coming.
Yeah. And then I try and stop myself from being too useful and think, ah, I should be more wise. Yeah. And I'm told that, you know, we always overestimate what can be done in a year and underestimate what can be done in ten. Is that the truth here?
Yeah. Or is it actually coming faster than we think? No, I think that's still the truth. I mean, maybe both timescales of short-term and long-term are nearer than other technologies, but I do think, like, literally today, as of today, and in the next year,
things are a bit overhyped in AI. I mean, they couldn't be any more hyped in some ways. But on the other hand, interestingly, I still think it's still very underappreciated how revolutionary this is going to be in the sort of timescale of about 10 years,
so we could call that long-term. So there's still that dichotomy even today with AI. So jobs disappear, but he says new ones always show up, except this time the change happens ten times faster than the industrial revolution, and that one came with riots and decades of
pain to work through. His actual answer to the inequality this creates is pension funds and sovereign wealth funds, basically a maybe, dressed up as a plan. Notice how confident he sounds about jobs, about energy, even about fusion, but when
the question turns to what we're actually for once the work is gone, that's the one thing even he admits nobody has figured out. He built most of the breakthroughs behind this entire industry, and by his own number we've got five years before this stops being a conversation and starts being our reality.
If that didn't sit right with you, you're not overreacting, you're paying attention. Subscribe, because nobody building this is slowing down to wait for the rest of us to catch up.