There's nothing that I understand better than the best chess engines. So, kind of the last frontier was something called fortresses, where generally one side has a significant material advantage but cannot make progress because the opponent's position cannot be attacked. And this is something that engines used to not have the horizon to be able to see, because it would just calculate, you know, 100 moves ahead and says, well, I have the advantage, I'm four points of material up, and the evaluation would be four points. But now it's remarkable how quickly it just drops to zero. This position is a draw, and they kind of spot it before the best humans do as well. And like I have to click and understand why is this a fortress, but this looks like a fortress. Oh, you can break it that way. That's been remarkable for me to see kind of how that has all changed. And now I'm just trying to learn, trying to understand why things that I have thought were truths for a few decades playing chess are now being challenged. And it's interesting.
But I think we're also coming to a point now where we're really bad at predicting the future, but I don't see how they can improve a ton. Like there will be some minute improvements, but I think chess has essentially been understood and perfected by modern engines, and that being both the kind of calculation-based older models and also the self-trained AI-based ones.
It's interesting just as you're speaking and answering that question. I was thinking that just to my point off the top, in terms of the theme today, talking about the relationship that humans have with technology. It feels to me like that's a relatively healthy relationship. I mean, there is a lot of fear out there about technology that people do have, but it seems like, you know, you're learning through that process as well.
The thing about it is, I kind of gave up a long time ago. Like it wasn't an issue that we were trying to beat the computers or anything like that. People have been asking me those questions for decades, and my thing has always been, well, I don't really like playing against them because, well, back in the days it used to be that essentially you would lose to someone who's really, really dumb, and that makes you feel dumb, and you don't like that, right? Whereas now they're just so good that there's no point. And there certainly have been points where it's been annoying to me as a professional that information, computing power, everything is so readily available to everyone now that you cannot really get competitive advantages at all. Like there's something called the Chess Olympiad that's being played right now, which is a team competition. And it used to be the case that if you played somebody from a decent chess country, let's say Colombia or Indonesia, countries like that that have a few grandmasters but no really top players, if you played against that team in the Olympiad, normally these players who were not chess professionals, you could sort of hit them early. You could hit them in the opening because they didn't have the capacity to practice the same way as the top professionals. Nowadays, you can have a full-time job and you can essentially prepare with your phone and be fine in the opening against even the best professionals. So it's kind of made the whole game, since information is so readily available, it's made the whole game a lot more egalitarian.
Aidan, just to follow up on that a little bit, how do you see the relationship between humans and technology evolving broadly speaking, knowing that there's probably a lot of optimism from people in the room, but a lot of fear out there as well.
Yeah, no, I'm a huge optimist with AI. I think it's going to contribute a lot. There's lots of fears about displacement. I actually think chess is one of the great, great examples. You know, like Magnus was saying 30 years ago, 20 years ago, whatever, decades ago, computers got to a point where they were either as good or better than humans at the thing that they were doing. And if you look today, chess has never been as popular as it is today. It's like the most watched it's ever been, the most played it's ever been. It's blowing up. It's extraordinary. And 30 years ago, computers were more competitive than us at it. And so I think it's a really fantastic example of this thing called Jevons paradox, which is this idea when like great efficiency-enhancing technology comes out, you would expect it would displace people. You would expect that, okay, it's more efficient, I need less people, so I'm just going to do this with less. And counterintuitively, the exact opposite happens. You end up seeing demand for the thing that is being automated or is being made more efficient rise, and then more jobs get created. Another good example of this is in software engineering. So when we were developing these LLMs, we actually sort of came for ourselves first, right? We automated software engineers first, and the number of jobs — people were super scared about this, by the way. Software engineers were super scared. They didn't know if they were going to have a job in a few years. And the CEO of Anthropic was saying 90% of coders will be automated away within the next 12 months. Well, it's well past 12 months, and there's more jobs. Unemployment is going down. There's more software demand than ever. There's more software engineers than ever. So I think it's playing out right in front of us. This technology is going to be augmentative. It's going to empower humanity to better choose how it spends its time. You will be able to automate things. You'll be able to automate the stuff that you don't want to do so that you can focus on the things that you love, like chess.
What about, in that context, the role of human judgment and critical thinking? Because there has been some data around the fact that especially with the younger generation, maybe they don't have the same level of skills that previous generations had. Do you buy into this narrative that human judgment will become more important? We should invest more time there in this age of technology.
For sure. Well, actually Magnus is a new dad, and so I would ask his advice about how to sort of bring a new generation up in this reality. You know, I'm getting married next week, and we're going to hopefully have a kid pretty soon, so I'd love to hear Magnus' advice.
Yeah, that's a very good question. I think since our son is turning one year old very, very soon, in a couple of weeks, we've not really come to a point yet where you can shape him too much when it comes to rules for technology use. Like obviously he's going to be curious about screens and all of that, but honestly, like he's curious about everything. So he's not on ChatGPT yet, and I haven't been able to crack the code when he kind of goes on the laptop either. So I think we'll definitely take an approach of being pretty cautious with those things. I think kind of our generation has benefited a little bit. I have benefited in chess certainly from being in a bit of a hybrid generation that you have the tools, but you also kind of remember being brought up where it wasn't as dominant as it is now. So you can kind of see the benefits of having some moderation there. But yeah, I think there are a lot of people who are in my position or who's in your position and are very definite about how they want to raise their kids, and when it comes to a certain point, you end up taking like a little bit more of a practical approach, but we'll see.
Yeah, I can certainly say that as having a teenager at home. You know, you have good intentions, but sometimes it doesn't always work out the way that you had planned. And maybe they use more of the tech than you initially anticipated. Maybe I can ask this question, and either one of you can grab this one, but I wanted to talk a little bit. I think this sticks with this idea of not just the next generation, but the world that we are in today. How do we ensure that technology doesn't create a dependency? Because you do see this happening sometimes. You know, you see it with writing, right, with ChatGPT. You see it with other use case scenarios with technology. So how do we ensure that, you know, humans continue to be sort of in control of that experience versus being totally dependent on the technology?
I can have a crack. I think, you know, I do worry about relying too much on outsourcing our reasoning, because that's such an important skill to hone. It's like a muscle. Like you need to flex it. And in the same way that we don't hand kids a calculator to learn, you know, long addition or long multiplication or these sorts of fundamental building blocks of mathematics, we give the calculator to them later, right? We're like, 'Okay, now don't waste your time. You know how to do it. Don't waste your time on that.' Right? You have to do the same thing with AI. So take away the tool, teach them the fundamentals, give them the skills, teach them to learn hard things themselves without a crutch, without being able to lean on the easy, quick tool. But then once they know, once they've proven that they know, give it to them because now they should just be able to outsource that and spend their time learning the next problem.
Thanks. No, I mean, I think we've, you know, we've opened Pandora's box. It's very hard to reverse that trend that we have. So I think, yeah, the next generation, the generation after that, our generation, we are going to get dumber and dumber. I think it's just the way it is. I find it very hard to believe that we can in a significant way kind of reverse the broader trend that is in society. Like obviously as individuals we can do better, but yeah, I don't see that changing.
