Episode 120

Science in a Post-AI World w/ Aragorn Meulendijks

Watch on YouTube

About this episode

In this episode of Untangling Web3, futurist and returning guest Aragorn Meulendijks explores how artificial intelligence and robotics are redefining science, labor, and the future of human society.

With developments like humanoid robots entering mass production, the reversal of aging in primates, and AI-driven scientific discovery accelerating, this conversation focuses on the transformative—and potentially disruptive—technological shifts underway.

Aragorn unpacks the exponential acceleration of innovation and its potential to trigger post-scarcity economics, reshaping the social contract as we know it.

Key highlights

  • Robotics and Post-Scarcity Economics. Humanoid robots are advancing rapidly from prototype to consumer product, with companies like Figure and Tesla aiming for mass production within the next two years. Robots are already deployed in factories and are now entering homes to accelerate learning through real-world complexity. Aragorn projects a scenario where 1.5 billion humanoid robots could exist by 2035, disrupting manual labor across the economy. This raises fundamental questions about capitalism, productivity, and whether society must begin preparing for a post-scarcity world.
  • Scientific Discovery in a Post-AI World. AI is revolutionizing the scientific method by reducing years-long research timelines into days through simulation and modeling. Projects like AlphaFold have predicted over 200 million protein structures, effectively accelerating biochemical understanding by centuries. In fields such as drug development and fusion energy, AI is proving to be not just a research assistant but a generative force for discovery. However, Aragorn raises concerns about verifying AI-generated insights and the increasing noise in scientific literature from low-quality, AI-generated papers.
  • A Philosophical and Emotional Reset. As AI and automation disrupt productivity and labor, humanity may undergo a broader philosophical shift. Aragorn speculates about a post-work society, where meaning and purpose are decoupled from employment. Referencing Asimov’s Foundation series, he suggests that humanity may need to evolve emotionally and cognitively in parallel with technological progress to avoid dystopian outcomes and fully embrace a new model of civilization.

The future that Aragorn sketches is both exhilarating and daunting. Mass robot deployment, scientific automation, and redefined human purpose all lie within reach, but navigating the transition will require a collective reimagining of economics, ethics, and governance.

Whether society evolves toward utopia or faces instability will depend not just on technology, but on humanity’s ability to adapt, redistribute value, and embrace new social contracts.

This episode is sponsored by the VeChain foundation. Learn more about VeBetterDAO here: https://vebetterdao.org/

Transcript

Auto-generated from the episode audio.

Welcome back to the Entangling Web 3 podcast. At the risk of sounding like a broken record, AI and automation are everywhere and the tech is advancing by the day. Lately, we've been discussing its impact on the digital realm, social media, the internet, and desk-based jobs. But, have we truly considered its impact on the physical domain? How are these technologies set to change how we do science, for example? Or, how we discover, produce, and distribute medicine in a post-AI world? Today, we're delighted to welcome back a friend of the show, Aaron van Mulendijk.

Today, he'll be helping us answer all of these questions and more. Aaron, welcome back to the show. Hey Jack. Hey Alec. Wi-Fi's up and running. Ouch. I don't even want to talk about it. I don't even want to talk about it. Yeah, if anyone's watching the video of this, you may have seen a quick a quick clothing change, a wardrobe change from the three of us because we initially started trying to record this episode. Alex's Wi-Fi didn't want to do so. Yeah, so we're back here a couple of weeks later.

I think, you know, it's interesting. There's lots that's happened in that time. Alex, do you think it's going to hold out? Well, that was 2 weeks ago and then we tried to record last week and Aaron just forgot about the slot. So, I messed up. Aaron messed up. Let's see if Jack messes up today and we get the whole holy trifecta. No chance. We've got to have We've got to have one of us being professional, right, guys? But, Aaron, it's it's genuinely it's so good to have you back.

We obviously, you know, we had a great first episode with you. I honestly can't remember how long ago now, at least a year, I think. And then, we also had you join us in London for our event in back in March, I think. Yeah, I don't know where the time's gone since then. When we talked a lot about, you know, the intersection of kind of the metaverse, AI, and obviously a bunch of other stuff because it was a bit of a runaway train of a conversation.

What have you been thinking about since then, Aaron? What's been on your mind since those six kind of five, six months since we last saw you? Well, yeah, what have I been thinking about? Um In all fairness, most of what I've been talking about has been how AI will kind of snowball into robotics, right? Because everybody's talking about AI and I always feel like most of the people talking around it, the most of the AI experts out there, they're they're always just talking about, you know, the latest feature.

Yesterday, we got Agent Builder from OpenAI, which is obviously massive. And and now I can get like everybody's posting about that, right? But, I always try to kind of look, you know, to to the to the next thing. And one of the the biggest next things, I is robotics. And funny enough, 15 minutes ago, Figure actually launched their first a kind of preview video of Figure 3, which is a third iteration of their humanoid robot. And they got a billion in funding this year. So, we've been talking about that a lot.

I've been making kind of I've been thinking about, not really talking about, thinking about where things are headed with other fields of science as a result of AI. One of those is is quantum. Another one is longevity, which is going extremely fast, much faster than I even anticipated. And I was already anticipating it to go, you know, so fast that people thought it was crazy. Because China literally 2 weeks ago released a paper proving that they reversed aging in macaques, which are primates. So, we moved from mice at the start of the year to macaques at the end of the year.

And that is pretty much bonafide proof that we can reverse aging in humans as well. So, that's crazy. And then finally, the next for compute is also relevant, which is something I've been thinking about. So, where are we going to go because we've hit the upper limits of traditional silicon chips. Everybody's talking about how ASML is the most powerful company in the world. And now I already saw somebody who made a video about a different company is the powerful company in the world because it's the company that supplies the parts to ASML.

So, obviously, it's also the social media thing like it But, but but it is it is an interesting question because we're definitely in well, I I I I I sometimes refer to it as as the second Cold War. Right? We're now in in a massive arms race between China and America. And Europe is not a part of it. Because we managed to we literally managed to maneuver in in the worst possible way at every Our own worst enemy. Right, Aaron, you talked about six different topics there.

Let's dive into some of them. Like, I want to start with robotics cuz that's obviously my background, one of my You guys moderate me. I don't know what I'm doing. Yeah, we've got a moderator. You You know you're moderating. Don't think we have a mute button on you, but maybe we should try and get one. That could be a a feature request for the next time. Um let's go start with robotics. Like, obviously, all anyone's been talking about, like you said, for the past 2, 3 years has been ChatGPT, how to automate workflows, how to create digital content, all this kind of stuff.

But, not many people have been talking about robotics. And I think, as you said, that has been massively overlooked. We had Colin Patrick on here like maybe like a year ago. And one of the lines he said which really stuck with me is like, "You think the market for autonomous vehicles is big. Wait until you see the market for manual labor." And, you know, we're starting to see these humanoid robots. You and I were talking about the foundation earlier when that was kind of imagined like a hundred years ago.

But now, there are videos. There are actual demonstrations of this potentially getting to mass manufacture, to everyday lives. Like, what is that going to mean for humanity? Yeah, what is that going to mean? Well, how philosophical are we allowed to go today because the thing is that yeah, ultimately, I really I really believe that this is the point in time where we need to start thinking about post-scarcity. And that sounds crazy to people. the whole idea that we might have to kind of start considering what Star Trek economy really meant because it's never really defined.

Um it's something that a lot of people just can't wrap their head around. And whenever you try to bring it up, they'll be like, "Yeah, I mean, come on. Come on." But, the idea that capitalism and and a consumption economy as we have in in the modern world is the end-all of everything is of course absurd, right? recently, Chinese research has shown that a skull or or research has shown that a skull that was found in in China in the '90s is actually a million years old and it's a Homo sapiens skull.

Um so, before we thought Homo sapiens was 200,000 years old. Now we know we're 1 million years old. and in that time, literally, the the smallest sliver of that time was spent in in democracy and capitalism. So, we really have to seriously consider what that means. And if we bring robotics into the game, that will change everything because based on my estimation, so I looked at I looked at smartphones. I looked at EVs from other companies than Tesla. I looked at electric bicycles, electrified bikes, which is quite popular in the Netherlands, for example, right now.

And some other products. And basically, I selected them based on either their kind of return on investment, how valuable is it to get one in your life, and what do they cost? And I tried to get a mix that on the one hand represents the potential cost of having a humanoid robot. And we know that, for example, Elon wants to sell his Optimus said around 40,000 US dollars when they come out. And the same goes for Figure. And the same goes for for X1. And and then on the other end, you need to consider that an electric vehicle is is is in the same price range, but is much less convenient.

The return on investment is significantly lower because it can only be used for one thing and that's transportation, right? So, it's more a humanoid robot, especially one that is can get upgraded, is far more in line with the potential return on investment of a more versatile device like a smartphone, for example. And so, if you bring all that data together, and I'm simplifying now, but you can kind of draw a comparison in how fast that adoption adoption will go. And what most people do not want to believe and they come up with all kinds of excuses why it's not possible is that even though we only have a few hundred thousand humanoids robots today, right?

just launched Figure 3, but their Figure 2 has been working in in the BMW factory in Spartanburg for months now. It's going very well. Now they're launching robots in the in the that you can buy at home because they found that these robots learn much faster when they're in a human home because it's messy. So, they're going to start rolling out. And they expect to roll out about 100 to 200,000 in the next 24 months. So, you think, "Well, okay, maybe we'll double the amount of robots we have." But, if you extrapolate that kind of trend and you do that for all 12 companies that I know of that are working on humanoid robots right now and the investment going into that, and then the return of investment that that brings, then you come to 1.150 million in 2030.

And that brings you to 1.5 billion in And then, the robotic experts say, and I know this is long winded. I'm going to cut it off and then you guys can ask me a question. But, the robot experts say, "Yeah, but we they can't go that fast because there's all these technical issues and hands are complicated and blah blah blah blah." But, they always say this about every technology. And the point is that once the return on investment starts to kind of materialize in the future on the horizon, then what you see is that there's a convergence of effort and finance.

And then suddenly, that problem solving for those issues accelerates. And then suddenly, it's there. Exactly like, you know, with the longevity. I talked about the macaques now, right? 5 years ago nobody believed we would ever reverse aging. A year ago we they did it on a mouse and they said, well, that's kind of interesting and now we're at macaques because, you know, suddenly people are like, oh, we can do this and then they're just going for it. And I think people really overlook that kind of factor of of when brains and finance and human focus come together and how that accelerates these But you kind of you've talked about like numbers there and I agree like the numbers are staggering and they all going to happen like it's inevitable.

But like what does that actually mean for society? Like you talk about this is going to create a world of abundance, right? Like the fact that it's going to cost $45,000 or $40,000 and that's going to be something that can basically in some ways work 24/7 round the clock for 5 years. That reduces the cost of labor significantly. But the people buying these aren't going to be maybe you or I maybe they will be. It's going to be big corporations. So at the end of the day that might mean everyday users potentially lose out on jobs even if there is a world of abundance.

What is that going to mean for society, do you Well, I think that it's just very simple economics and and if you put one and one together, you come to the conclusion, okay, there's going to be an increase in job loss from AI itself, not even robotics, right? I mean Sam Altman yesterday he announced his agent builder and today people are posting, oh, he just killed another thousand startups. And that's just the startups. Deloitte this week is is paying back the government I think 400 million because they admitted using AI and it wasn't perfect.

Um McKinsey's letting go people. Like there's all these industries and people that claimed, you know, they would be last and and they are first to reduce their um workforces. We're going to see a lot more of this. So that means that we going to see a rise in job loss and that might be offset by something. For example, you know, aging populations um whatever, but but ultimately we'll see that increase and increase. And so I I would say argue that I agree with the paper from the World Economic Forum and the University of Amsterdam which was published in January this year which states that by their estimate in 2030 up to 70% of all jobs in the job um will either be already replaced with AI or robots or will be so completely different from what they were before that they would need complete retraining of the human to do them.

And if you consider that, that's a massive amount, right? Now let's say that they're wrong. Let's say that it's 30%, right? Less than half. But even then if that were true, then we'll see an increase of 15 to 10, 15, maybe 20% in in job job loss, right? In Those are massive numbers. I mean every other time before in recent in the last 200 years in recent history when you see countries or economies that have, you know, more than 10% or 15% unemployment, they're in a real crisis.

20% is is massive is a real problem. So what we're going to see is over the next 5 years we're going to go there and by the time we get there we can't governments will start to realize that something is going on that they've never experienced before. They will need to start thinking of solutions, maybe universal basic income, maybe something But at first they'll just try to kind of find ways to kind of slow things down or fix the problem temporarily. But the problem will just exacerbate.

It will just get worse because then robots come into the picture. So it's no longer just, you know, the the white collar jobs, but then it starts being the blue collar jobs and then it will get worse. And so at some point they'll realize, okay, we there's going to be a point where no job is needed. Exactly like Elon Musk already said to to your former Prime Minister Rishi 2 years ago I think at the AI summit he said there will come a point that which no job is needed.

You can have you can have a job if you want to have a job, but ultimately the the AI can do any everything and that's exactly what's going to happen over the next 10 to 15 years and by the time we get to the end of that some people will have had a really really rough time of it. Other people will be have been a little bit more lucky staying in their job a little bit longer, but ultimately the global governments will have to come together with a solution on how to share out wealth and the the means that the the result of production through AI and robotics because they will basically do everything.

And then people say, yeah, but in the past new jobs were created. Yeah, but when we created the spinning wheel or the steam engine, it wasn't so intelligent that it would learn every new job that was created out of that as well, right? So the only limiting factor here is going to be production and when it comes to software, there's basically, you know, sky's the limit. So the the only thing slowing this down in my opinion will be the the speed at which humanoid robots and other robots can be produced.

Yeah, I think that people often use previous industrial revolutions or revolutions as like markers to be like, ah, you know, it all worked out in the end. But like the last industrial revolution took place over 100 years with a much smaller population. This is going to take place like you said from zero humanoid robots to hundreds of millions over the course of the next 5 to 10 years. So it's going to be 10 times more impactful in 10 times less time. So it's going to be 100 times more impactful in a time where, you know, bureaucracies are far more bureaucratic and decisions don't get made and people understand less and less about what the technology is.

So yeah, I do think it is going to be a scary time in a lot of ways, but also maybe exciting. Unsure. Let's see what happens. Like if you trust the capitalist elites, then maybe you're in a good position and everything's going to be fine. If not, who knows? This is kind of a conundrum always for me because on the one hand, you know, I I introduce myself as a futurist or you know, which I think is a kind of a little play on on the the psychohistorians of our foundation.

Obviously if you hadn't realized yet, so now you'll extra appreciate this. And I would like to think that maybe this is the start of of kind of a discipline like that where instead of always looking back only we start to more seriously look forward and not just as futurists, but more as as kind of discipline. But but on the one hand that means that a continuation of uh of the past or repeat a repetition. But on the other hand, I also very strongly believe that we're experiencing something that at least in known history has never happened before to humanity.

And therefore we really have no No way to to really understand what's happening to us and where things are going to go. And of course Ray Kurzweil etc. they call talk about the singularity and how, you know, the acceleration of technology is at such a point that we can't even kind of comprehend what comes next and I I do also believe that. By the way, I also I started to put a lot more stock into Peter Thiel's which is not a completely, you know, fleshed out, but he has this stagnation thesis.

Um and at first I thought he was a um but I started to think about it recently and we didn't want to go down that path. I just realized I actually didn't want to talk about end of empire, but we definitely Yeah, but we're definitely seeing I I I do believe that you know, we we're seeing stagnation in terms of our ability of developing ideologies and progressing, let's say, the humanities as as opposed to the stem stem technologies that underpin the development of for example AI and robotics and and all of that stuff.

Today's episode is sponsored by the Vechain Foundation. Now we often talk about how emerging technologies like blockchain have the potential to revolutionize sustainability and Vechain is at the forefront of this green movement with initiatives like VebetterDAO. Let's take a closer look. Imagine tracking small actions like recycling at low cost making them valuable assets. These verified records build a massive collective impact benefiting people through their own actions. Sound a bit complex, but Vechain made it fun. Explore our apps at Vechain's goal is to power a new era of sustainable solutions.

So if you've got an app idea or just want to earn rewards for small sustainable actions everyday, go check out vebetterdao.com to find out And now back to the show. Yeah, I've got it. This is obviously everything you say is kind of bringing a lot of questions into my mind. Um not least about what comes next, right? But before we get to that, I'm kind of interested in the path from here to that end state that you mentioned in however many, you know, 10, 20, 30 years it will be where we have hundreds of thousands, millions of humanoid robots because you know, if you take I saw recently, maybe this week that there was a lot of graffiti in I think New York defacing advertising around one of these AI wearable pins.

I think it was the friend one. Um because there's there's obviously some resistance to the idea of AI replacing human connection, it seems. So what what do you think the the likely path is? Because I agree with you. It can feel like maybe it's inevitable, but there's also a part of me that thinks, you know, is this going to have to be driven purely by uh you know, corporate adoption and they it being forced down people's throats because it's in their interest or do you think that as a society we will genuinely come to accept that, you know, this is just the most convenient way of doing things going forward.

I'm curious to to know what you think about that. Well, like I said, we don't really have any historical precedents for this in in in a real way. I mean obviously you could refer to to like the stories like the one Mustafa Suleyman starts with and ends with in his book The Coming Wave, you know, the burning of the the the what are they called? The the dollies? The No, what's a beautiful English name for them? The weaving machines that Anyway, you know what I'm talking about.

The machines that are made to replace the weavers initially in the first Industrial Revolution. Which gave rise to the Luddites and the Ned Ludd leading them to burn these So, well, there you have it. I mean, at the very least I think that we'll we'll have to expect that some of that will occur. Right? Not everybody likes change. I literally right before I got into this recording saw somebody that I highly respect in the Netherlands, a thinker, a researcher, writer who wrote that we just invented the end of boredom.

But the way he describes it is that it's like the end of humanity. So, there's a lot of doomers out there. And some some of these doomers, they will resist. And they will take up arms. They will feel that they're being replaced. But ultimately, I think that again, coming back to store history for the majority of the 1 million years that humanity has been around, at least for the 200,000 years that we've studied more thoroughly. We know that humans didn't have a 9-5 job. We weren't going to schools and universities and then defined ourselves through our role and our economic value and our agendas.

So, our calendars. So I think that there is a definitely a back towards a more wholesome human existence in which we spend our time on things that feel very important, very purposeful, and give us purpose. And But that are not a job or an economic to see that kind of future requires a lot of humans because we're just not built to look that far ahead. So, many many people will struggle with that. Of course, any transition of this type is extremely hard. We've seen in every Industrial Revolution that the transition was hard.

There were always people that lost jobs before new jobs were created. And in this case, it will be everybody losing their job before we, you know, figure out a system that allows people to be jobless and still kind of prosper. So And then of course, there it's not impossible that we're heading towards a more dystopian future. I I think that the chances of it are not nearly as high as people would like to believe because we're prone to kind of, you know, see the risks and the dangers and the negative potential first and then overvalue that.

It's a well-known cognitive bias. But Yeah, we like to catastrophize everything as a society, don't we? Yeah. Yeah, exactly. And so, we're, you know, we got the new alien series on television and you're like, "Oh, four corporations ruling the world. This is exactly what's going to happen." People are sitting there thinking like they're they're just going to take power from the governments because they have way more money. And you can see it now because Elon Musk and Oh, no, Elon is was not invited, but Zuckerberg and Nadella and and they're all having and Bezos, they're all having lunch with Trump and yeah, it's it's going on.

They're taking over the world. Um We're all idle creatures, right? We've been adapted over the course of millions of years to survive. So, fear is a a far more easy button to press and stimulate us than hope and curiosity and all this kind of stuff. So, it makes sense that we kind of are driven by fear. Um I guess yeah, one thing I I want to mention, I completely agree with you. Like there's going to be pushback. I you know, in the Foundation series and the early I Robot books, he talks about the idea of like rather than left and right, we're going to have humanists and futurists.

And some people will have to lose out short term for the benefit of society overall. Like we look at productivity in the UK, you look at the NHS. Sure, if you carpet ask someone, "Are you okay with robots replacing people?" Most people will probably say no. But if you say, "Okay, well, what if we could bring wait times and a doctor's from 2 days down to 30 minutes by introducing a new type of nurse robot?" Then they say yes. It's all use case dependent, right? But like on that kind of note, like I I want to talk about the scientific method and this is something you're very interested in.

I saw an interview with Demis Hassabis, the CEO of DeepMind. And his definition for AGI is actually quite interesting. He says it's defined by AI being able to innovate, run a you come with a hypothesis and then run a scientific method to kind of justify and prove that new hypothesis that hadn't been kind of theorized before. Is this an area you think is underappreciated? Like science and research specifically. What what kind of role can AI and robots play in this space? Oh, 100 100%. It's it's not just even the role that it can play, it's a role that it's already playing.

I there's a guy in the Netherlands called Jarno Duursma and he back here in the Netherlands, he's much more famous than me. He's he's a he's a very often booked futurist. He's a great guy. I talked to him a lot. But he just made a post and he said, "Yeah, somebody in my audience has asked you know Chat GPT, why should I pay for it?" And the conclusion of that question was that most people, the vast majority of people that are now kind of, you know, starting to hear about AI and starting to consider using it they don't even know the difference between, you know, the different models.

They don't know that Chat GPT has like GPT-40 and now it is five and before it was just a large language model and now it's a reasoning model and it doesn't understand what the difference is between They don't even know that there's a Claude, let alone that there's a Claude 4.5 Sonnet and then a Claude 4.1 Opus and a Claude 4.5 Sonnet thinking. And you know, so they don't even know what the difference is between all this. And I think that is really the the the thing here.

Um is that people people just have have have no idea about that and let alone that they know what's going on in other fields of science. If AI progress would stop today, right? If if we would suddenly be in the Three-Body Problem or in, you know, the Remembrance of Earth kind of universe and aliens send these four-dimensional um things over here so that we can no longer progress our science. Maybe I'm making this too complicated for people, but that's that's another sci-fi theory. Then even with no progress being made, the implementation of the technologies that were created in the last, you know, 8 years would completely revolutionize our world over the next decade.

Like I said, longevity. We're now seeing the first basically customized cancer vaccines being rolled out. A lot of that is to do with the proteins structures and and the capabilities of AI of finding ways to turn this into medications that are now, you know, starting to make their way into actual consumer products. Right? All of this has been going on. So, our medicine is going to be revolutionized over the next 10 years. People are going to be significantly Healthier, right? So, what are we going to do with pensions?

Even if we did keep if we didn't have the robots, what are we going to do there? Although this might actually offset the This might offset the problem in the job market, of course, because well, or would it? If people live no, it would be the other way around because people are living longer, they're staying healthier. So, technically, they could remain working and then at the same time we'll have robotics. Anyway, so that's changing. I think that at some we're going to see AI make its entry into other fields.

Maybe even like politics. But for now, we already seeing it being used in science more and There's also problems with that. I think that the recent research showed that of the papers being the peer-reviewed papers being released every day now is actually increasing quite rapidly, the amount of them. But what's also increasing is the amount of them being written by AI and therefore there being, you know, a lot of papers that have serious problems in them because there is complete bogus input in And people don't even know anymore.

I hope that this will be like an early peak as a result of of bad AI and people having bad etiquette and hygiene in using them and then it will, you know, that will that will plateau out or or, you know, drop down again. But But the speed with which it's growing Cuz I feel like it's going Is that likely? I guess like the model's going to improve, but like I've seen a lot of studies about people kind of becoming dependent on LLMs. Like even I, like my ability to write is definitely degenerated because of my overuse of LLMs and Chat GPT.

Is that a bad thing? Like I don't know. It means I have time to focus on say more creative things, focus on things that are more productive rather than writing out an email and spending 10 minutes to summarize what I've done in the day. I'm not sure. You can forget about your for your for your writing a book with the skill you just I um I think that as long as we live in the world where we are the way we measure everything is based on how much we can get done and the value of that work in the shortest amount of time is is the way of measuring it.

Things would initially just get worse, right? Um and it's a race to the bottom. Um but I think that and and this whether this takes 5, 10, 15, or 20 years, I I can't say. But I guess once we start to realize that the world has changed forever that we will live significantly longer that we in that time that we live, we will have significantly more time to ourselves. And we're not rushing to keep some kind it it growing economy going, which by the way is impossible.

Well, it's just a capitalism. I think that we'll have a complete kind of um maybe maybe even spiritual reset. Um at which point we realize that there's no reason in rushing through things, right? And and I mean if if you're into Buddhism or anything like that, you're already kind of you you you might already be able to identify with that kind of mindset of, you know, letting go that kind of uh those kind of um drivers in in in in life. And I think that that's where we need to go.

If we as a species by we keep coming back to the foundation. So, in the series, they actually pretend that the second foundation, and this is spoilers by the way, if anyone is listening to this and you've never read the books or watched the series, I'm going to spoil it for you. But in the series, they pretend the second foundation is just these mutant kind of psychic people that come together. But in the books, as you know, it's completely different. They're not just some random kind of mutations.

No, they're people that saw a completely potentially different future for humanity where we really develop both our emotional and our cognitive abilities to the point where they take us to a next kind of level of communication and and and and thought processes. And I think that um you know, something like that could occur once we're placed in completely different circumstances. Of of course, the problem we're facing, the three of us and everybody listening to this, is the transition phase. We're the guinea pigs. I mean, okay, well, speaking of guinea pigs and mice and things, um I want to dive a bit deeper into this this topic of the use of well, the automation of the scientific method essentially.

Because you've hinted at that you've hinted at that in the context of, you know, aging, right? And how we can potentially stop that. But I think it'd be interesting if you can kind of lay the ground of how how are we automating science currently and, you know, potentially in the future, but even just the state of play. Because I think it can mean so many different things, right? We've talked about robotics, we've talked about AI, which, you know, can mean like deep learning, I think when we talk about protein folding and things like that.

We're talking about a very different kind of flavor of AI to the LLM uh gen AI type stuff that we're talking about with um you know, the what's happening on the internet. So, yeah, do you do you I mean, you must be following these things quite closely. What are the kind of big developments in in the field? Um well, f- f- f- for the for biological sciences, it's I mean, yeah. On the one hand, uh these are different flavors of AI, you're correct. And in fact, to some extent, they're completely different models.

But a lot of the um And I'm trying to yeah, the lot of the learnings, right, that are occurring can be applied in in both fields, right? AlphaFold and and what's going on at Google DeepMind is definitely benefiting from the massive rapid development of interest in AI in general. Mhm. And because of the the massive investment in AI, this is also a thing. Like the finance bros don't understand the difference between the large language model and the reasoning model, right? So, if you and in of obviously, there's some VCs out there that do, but a lot of them don't.

So, you know, they get pitched like, "Oh, we got this great idea for an AI, and if we do this, then we're going to be able to, you know, generate new medicines like um like Ozempic, but we're going to be able to do it in half the time, and uh we're going to do a lot more, and we can do it a lot cheaper. So, we're going to we're going to make billions, boys." And they're like, "Oh, yeah, let's go. I want to get rich." Um and uh there's so much investment going into AI that a lot of these other models um are getting a lot of love in that sense.

And so, like you said, with the the the the current models, they, for example, did that AlphaFold uh gave us, you know, 200 million proteins. It's a billion years of PhD time saved. As Demis Hassabis said, you probably know this. Um And so, this is on the one hand, it's creativity, right? Because being able to predict these protein protein structures is is some kind of creativity, right? And also uh in some sense, a very deep understanding of beyond what humans are capable of of what comes into play um when these these uh structures are folded.

But it's not quite on the level yet of a human, you know, coming up with a radical new idea for how the, you know, how how physics uh or anything works. Like it it's not yet on the level of where Albert Einstein comes up with general relativity um And so, the question there is what is required for an AI to do that? And the skeptics right now say that AI cannot do it. But when they put mathematicians up against some of the best AI models recently to solve some of the more complex problems, the AI did ridiculously well to the point where uh the math um geniuses were like, "Holy We never expected this." But they also said something interesting, and this is I think where there's a danger.

They said AI was also really good at cheating. Right? It was It was better at than any human. And because you're talking uh of complexity at a level that is beyond most humans, um if you're really good at bullshitting at that level, you can get away with a lot. So, there's there's a little risk and a little challenge there, of course, um because if even the the best geniuses in the world don't understand your math math anymore, and you say, "Well, yeah, but it it's it's correct." Mhm.

How are they going to prove you wrong, right? Um I think that we're just very close to that level where AI truly breaks through in in that real kind of human that kind of creativity that we've always considered human. But it's just hard to say when uh exactly that will happen. Um and then it's it's it's still harder to create an AI that then can go out and make, for example, completely new theories on physics, where we know that it's reliable. And I think that that last part might actually turn out to be the hardest part.

How do we verify the things that it's finding for us? I think that's where the real real challenge is. And this is where it kind of bridges into the world of like the physical world, right? Robotics and being able to test things in reality. I think like talking about the different models is quite important. Like it's something Google DeepMind does very well. Like they have very specific applications like AlphaGo for scientific research, for modeling. Like it's kind of taking what we know, but applying it in a kind of much more granular way.

So, imagine if you had a perfect model, or you could do it or run a perfect experiment on a set of cells, you're always going to be limited in polynomial time. So, it comes like degenerative diseases, you've got to test them for 30 years before you can really see what happens. But if you have a a perfect simulation of those same cells, say run by AlphaGo now, you can reduce that time of 30 years down to a simulation that takes minutes or seconds or something like this.

Like a really good example of this is in silico's drug time, I think they they reduced something like testing from 500 days to 30 days just by using like simulations. I think that this this is powerful, right? And but like the new innovation stuff is kind of like I don't think it really matters that much right now. There's still so much value to be had just from automating, simulating, and then taking things to the real world. Maybe the innovation stuff is going to come soon, though.

Actually, you're really touching on a super interesting thing there because yes, on the one hand, I I 100% believe that in in in the next, you know, 10 years, we'll have AI that will be capable of maybe even, you know, finally combining general relativity with the quantum physics uh in in kind of unified unified framework or or theory. Uh but at the same point, at the same w- we already have a like literally we're we're we're we're returning into a backlog world, where we have a massive backlog of of stuff that AI suddenly allowed us to do.

Like a billion years of PhD time in terms of proteins, and we don't even know what to do with ourselves to turn into medicine, right? You know? So, that's the other that's the other challenge here. And then obviously, a lot of this will you know, more rapidly get automated um um etc. But yeah, it's going to take a while because all the human systems that are still in place are so incredibly slow, right? Uh even just the bureaucracy, even if AI would not only be able to come up with a new protein uh saying I keep bringing coming back to that example because it's easy, but um and and if we could have turned it into a medication instantly as well with AI, right?

That they even then they would still have to go into trials and and policies that would have to be passed, and then they would have to be allowed on the market, and then Americans would try to keep it off the market so they can come up with a different version that they can ask more money for for a while. You know, so and I think that that that is where we're really going to run into the realization that all the human systems that we built have all become outdated.

And instead of moving us forward, they're now massive roadblocks that we either need to overcome or we need to, you know, get them off the road so that we can actually move forward. Um in the past, unfortunately, now I'm I'm definitely not hoping that that's going to happen now. But in the past, I think that a lot of this kind of forced fire kind of transition time was usually also hand in hand or side alongside, you know, some some war or some destruction or something like that, which actually made it easier.

It's not fun for humans, but it does make that kind of process ultimately easier when everything is destroyed and you kind of have to build from scratch. You have to start a new country or something like >> Well, I was just going to ask you, and I'm kind of, you know, move towards the future. But, you know, on that point, I I really like the bringing up the point that we don't know how to apply these technologies necessarily and what to apply them to. Like what what is the problem space that we should be applying these two like you know you know what what can we apply it to and I think that's where science is really interesting as a as a field because you could argue that it's one of the better fields to be using AI because you always test things empirically, right?

So you have you have some quite you know when you talk when you talk about physical theories you tend to have good ways of you know verifying them, right? By experiment. >> Yeah. The problem is that empiricism advantage kind of goes away the bigger the problem becomes, right? The bigger the problem you're trying to solve. Like if we were if we were talking about the Large Hadron Collider experiments having been you know we're always going to test out this AI theory then I think people would be a lot more worried when when the experiments going on.

So I think yeah it's it's a really interesting to think about applying these things to to Yeah. I think the thing with AI here is that we we still again people just are looking at it and it's like oh it's a toaster. And if my toaster told me to build this Large Hadron Collider I wouldn't trust But if I ask my toaster to toast my bread and I press a button it comes out after exactly 4 minutes and I always get the same level of toast depending on whether I put it in frozen or you know But that's just not how it works.

I think that people really really and this will take a little time people need to around to the idea that a lot of these AIs should be treated more like another human being. And when we deal with each other we have a kind of developed sense of who we can trust to what extent on which topic. Right? And even there we humans are quite you know how do you say this? Um not perfect because we like to think that for example I like to think that Alec is a really nice and trustworthy guy but when we really look into the science we know that that really depends on the circumstances.

So Alec might be one way with a certain set of people in certain circumstances but he will be completely different in another set of circumstances with different people which I will never But we know that there's actually a lot less consistency there than we think there is. But for humans to be able to successfully interact with each other we kind of need to have these set beliefs. But now when it comes to AI we kind of completely in the blue and we just don't trust it because we're not used to it and we don't know how to deal with it.

And so I think that you'll see two movements. On the one hand you'll see the growing ability of people getting used people getting used to AI starting to trust it more treat it more like another human being kind of get this intuitive kind of ability to to to decide on when to trust it when not when to double check it and and on the other hand we'll have that technological development of increasing methods of really making sure that what the AI gives us is reliable. And Google just um what's it called?

I think it's um I forgot the name but they released released this or was it Microsoft? Man so much information sometimes I mix it all up but one of the two basically released a whole new platform that is that is basically an a special AI kind of set of experts built to continuously fix and update things in your software infrastructure. So it can completely autonomously do things because they made it they built it so that it would never do something that can harm your system because there's all these checks and balances built into it where it double checks its own conclusions double checks its own solutions etc.

etc. And that is what we'll see more and more because there's a massive demand for that. And so then we'll get to a point where you know you can trust the AI 99.999% of the time and for all intents and purposes that's perfectly fine. and then we can you know allow it to build a new Hadron Collider or we can allow it to one we can allow >> the nuclear weapons, right? Yeah well like AI is also the reason that we went from only a nanosecond of fusion running a fusion power plant to running it for 22 minutes straight with positive energy output, right?

This is another field where AI is literally propelling us forward centuries in terms of technological advancement. Um which puts a really clean energy within within reach. We won't even have to go you know fission technology for that which I also think is perfectly fine but that's another conversation. I agree with you there. I think AI is a tool and a tool you know like fire can cook food or burn a house down. I actually disagree I think that most people trust AI. They kind of don't question it at all.

Like we've kind of grown up in the era of Google is correct. I don't know something I search it on Google I find out from a trusted custodian what the answer is and I trust the answer. That's the the definitive answer to all questions. And GPT is being used like that, right? Most people are now using ChatGPT over Google and when they search it they trust the answer. I actually think the problem is more we overtrust it right now and I think it does make mistakes and it's kind of becomes a question as these AI models become more autonomous they require less human intervention they act on our behalf they transact and interact with each other how do we propagate that trust?

And I think as AI becomes more pervasive how do we maintain and manage that trust? I will propose that we're both right. In that now yeah you know this saying the problem with the world is that that stupid people always think they're right and intelligent people always doubt This is this is exactly what you just basically we have that same conundrum. So on the one hand you're right. People that have you know no real understanding of AI I don't mean this in a bad way but people that have no real understanding of AI are prone to once they've had a few good experiences then completely trust it.

Right? To the detriment to their own detriment because they also trust it when it tells them you know I just you know do this or do that and when it's really not so smart to do yet. Um at the same time you have these super super intelligent people that are building these AIs and some of them are super enthusiastic and you know that they're pretty trustworthy but then there's a lot of other people out there that are I I we can't trust it it's still a machine, right?

They're basically species and they're like you know can't trust machine. Um I think both of us are right. Yeah. Yeah I think this is this is one of those things I think it is too early to whether one or both of those answers is correct. Um but I like the arguments on either side as as a neutral arbiter. Yeah it'll probably be a foundational aspect of life over the next decades. Like just like you know the the struggle between left and right politics has really defined the last the last half century after the Second World War and and our modern world I think this and of course the struggle between the West and you know the second or first second and third world countries this development this is going to be another really kind of pivotal struggle.

obviously there's going to be a point where people will be like oh do they need to does AI and robots need to get rights? Right? And people some people will argue in favor and other people will argue against. The famously Elon Musk and Larry Page were friends and they lit one of the things they fell out over I don't know if this is true but this is what I've I've heard and been told is that Larry Page called Elon Musk a species because Elon said that machines will always have to serve humanity, right?

So so this this argument has already Oh yeah. I is is because obviously that argument has been thought about in you know the likes of Foundation and pretty much all the AI robotics literature that we've ever mentioned. Yeah. Yeah I think it's it's just going to rage on for the And Aragon so we've got a couple of minutes left. I want to end on the future and really just ask your outlook because I I don't know about you but like I find it hard to gauge you Aragon when when you're talking the way you put across these arguments is very it's very beguiling but I can never quite get if you're excited or scared about this future.

Maybe maybe a mix of both but what is your general outlook for let's say that future that you described at the beginning does become a reality. Are you are you are you leaving the planet? What are you I think your gut feeling your gut feeling is is is quite quite good there. I wouldn't necessarily say that I'm scared because I I don't necessarily think that being scared or afraid or is useful beyond the point of being more aware of both bad good and bad outcomes. Um so I would argue that I'm I'm I'm generally optimistic but I'm very aware the incredible ways all of this can go And I I would argue that I'm probably far more aware than most people are of the scale at which it can go off the that means that although I wake up pretty optimistic every day and I dream of a future of having my own robot companion and spending more time in virtual realities and simulations having these novel experiences and seeing humanity move into the stars

but also into the and and the fact that I believe that I have a real possibility of being there for it for the next you know 150 years maybe even still at the same time and I hate to say this I kind of don't want to say it at the end of the show I'm just kind of waiting for something horrible to happen as well as a part of that transition. Um and uh that kind of fills me with a little bit of dread sometimes. And I think that that's what you're sensing in in in when I'm talking.

But I believe that the reality that will be is the reality that we choose to and I hate this jency phrase but manifest. So obviously it's not so simple as that all of us deciding that this is going to happen and then poof it's going to happen. But I do really believe that and there's a lot of signs to back this up that what we focus on, the things that we really want to see happen, um are ultimately much more likely to happen than the things uh than anything else.

And so if we start massively focusing on all the things that could possibly go wrong and in today's world uh that might be a little bit more likely I don't know than 100 years ago. I'm not sure how to judge that but that maybe it will go wrong. But at the same time as long as there's people like like you guys inviting me on shows like this to talk about it they can help people see all the wonderful potential utopian things that uh be in store for us then maybe we can tip the the scales a little bit just enough so that we can make it into this wonderful new world.

Oh thank god you ended on ended on a nice hopeful note there. I thought that was going to be really sad. That's a great way to end. We love it like I love the technology. I get it. There's a lot of scary things that could happen. But there's also so many exciting things that are happening now and probably will continue to happen. I do overall believe in humanity and I think we will be fine. So let's roll those dice baby. Aragon it's always a pleasure. Thanks for coming on.

This has been an epic conversation with the AI robotics. Kind of got the automation of science in there. Lots of things about the foundation obviously and manifest your destiny. I was not expecting that. But thanks Aragon. It's been an absolute pleasure. We'll have to get you on again ASAP. And thanks for those listening wherever you may be. Join us next time to untangle a little more of Web3.

Topics: AI & Robotics

Related episodes

All episodes