Intro. [Recording date: May 1, 2024.]
Russ Roberts: Today is May 1st, 2024. My guest is philosopher and author Nicholas Bostrom of Oxford University. This is his second appearance on EconTalk. He was first here in December of 2014 talking about his book Superintelligence, about the dangers of artificial intelligence–of AI–and he was way early in worrying about that.
His new book, and our topic for today, is Deep Utopia: Life and Meaning in a Solved World. Nick, welcome back to EconTalk.
Nicholas Bostrom: Thanks. And, there it is, yes. No. 2014. It’s been a while.
Russ Roberts: Yep. And, I’m a lot smarter since then. I’m sure you are, too. But neither of us are quite super-intelligent; but there’s some things out there that are interesting. But that’s not our topic.
1:24
Russ Roberts: Our topic is your new book, and I want to say before we start, this is a wondrous book. It is mind-expanding. It is poetic. It is moving. It is funny. The writing is superb. Every page is full of ideas. And it’s about what you call ‘deep utopia’–the world we appear to be heading toward, a world of unimaginable material abundance, robots, AI able to do everything better than we do, which seems to lead to a world of nearly, if not literally, infinite leisure.
And, in such a world, you ask, among many things, what would become of us then? What would give our lives meaning and purpose in a ‘solved world,’ as you call it in? And, what would we do all day?
And I think, as you point out, thinking about these issues forces us to think about non-utopian issues–the question of how we live our lives now and what is for many of us near utopia; and for many, not so close.
But let’s start off by trying to say something very encouraging. I think many people are extremely uneasy with the idea of a solved world–of the idea of all the leisure we want all the time. Talk about why people worry about that and why you think those worries are not as worrisome as some think.
Nicholas Bostrom: We’ve built our conception of ourselves and our dignity to a large extent on the idea that we can make some sort of useful contribution to the world, whether it’s at a large scale or just in your family or in your community. And so, to the extent that that’s the foundation of your self-worth, if that foundation is removed, you might find a kind of vacuum underneath your feet that would be disconcerting.
So, in a solved world where all the practical problems have already been solved, there are no more problems for us to solve, or to the extent that there remain any practical problems they would be better taken care of by advanced AIs and robots. And so, either way, there would be nothing of practical utility that human work would be needed for.
So, that forces, then, a fairly radical reconsideration of what the foundation of a good human life could be. And, undoubtedly it would require jettisoning some treasured assumptions about what the life should contain. And, I think that’s part of where this sense of unease would come from.
There are also, perhaps, more mundane concerns that people would have in terms of if they couldn’t have a job, how would they make a living? But, those I kind of set aside in this book in order to be able to actually get to the point where we can ask about the fundamental values that are at stake, assuming we solve all the practical difficulties that lie between where we are now and this hypothetical outcome.
Russ Roberts: And, somewhat reminiscent of the conversations some people have had around the idea of Universal Basic Income [UBI] where many people have suggested that even if we could, quote, “afford” it budgetarily to take care of people at a fairly high minimum standard, it would be a mistake because people then have no reason to work, and then they would–I don’t know if you quote it or not, but there’s a certain, I think, cultural feel that idle hands are the devil’s workshop. But, you disagree, I think, and quite eloquently.
Nicholas Bostrom: Yeah. Well, I think it might be true for some people in that more leisure is bad and some people need, I guess, a lot of external pressure to retain their upright posture, psychologically speaking. And, I am, I guess, guardedly hopeful that if this were the only change–say, suppose there was some huge economic windfall like the way that some Gulf State or something finds in enormous wealth underground and then they can live off the rents from that. Sometimes it works out well and sometimes not so well. I’m guardedly hopeful that, at least in theory, we could have a cultural change that would allow people to be raised to have good leisure rather than to be productive workers. I think the school system right now is obviously, to the extent that it is aiming for any particular outcome, is aiming to produce disciplined workers. L,ike you’re told[?taught?] to sit at your desk. You are assigned little tasks to do that you have to do. Why? Because the teacher tells you to do them; and then you’re scored and graded and there’s a quality control stamp at the end of it.
And then, you can go on to work in an office building or a factory depending on what level of–this is I think hopefully by the light of the future a sad model of what human development could be. It’s kind of necessary now because there are all these jobs that need to be done, so we need to have workers that do them. But, if you imagine a scenario where there weren’t all of these jobs that needed to be done, then I think we could conceive of better ways of raising children, maybe cultivating the appreciation of art and literature, the fine art of conversation, a sense of humor, physical activities, appreciation of nature, all of these things that currently are relegated to the sidelines.
7:17
Russ Roberts: And, it’s–one of my favorite themes is that culture is emergent. It’s not under anyone’s control. It responds to all kinds of forces in the world, economic improvement and so on. And, we’ll talk more about this because such an interesting–you have some very interesting things to say about how things might change in a different world.
I think there is an interesting question always of the speed with which culture could change relative to the speed of technology. There could be a great deal of suffering and challenge if that culture responds slowly.
The example that I think of a lot is smartphones. When smartphones came along I, like others, would sometimes say, ‘Oh, well, our culture will change about what you can do.’ And, it has changed; and it’s changed that you can sit there by yourself looking at your phone and answer your phone in the middle of meetings and scroll through your social media feed when you’re at dinner and in all kinds of ways that don’t seem to me to be consistent so much with human flourishing.
And, I say that as someone who does some of those things some of the time. So, the cultural norms I think ultimately will change, but I think there is a question of speed–but maybe we have lots of time.
Nicholas Bostrom: Yeah. So, I’m not saying we would use the increased leisure for positive life affirming, and so I’m[?] saying that in theory there seems to be this possibility.
So, I think there are kind multiple layers of this onion, and what I see as the outermost layer–which is where most of this conversation both begins and ends–is we are considering some scenario of moderate increase in automation. Maybe some people get unemployed, and then maybe the answer to that is reeducation or flexible labor market so they can find new jobs.
Like, we have many jobs today that didn’t exist a hundred years ago when almost everybody were farmers, so, like, analogously to that. And then, combined with some kind of cultural adjustment to the new technologies. So, we hopefully encourage more positive usage. So, I think if the technology were basically frozen in time where it is now or a couple of years more advances and then just applications, I think that would be the right question and the right answer.
But, I think this is really only the first step on a process that will keep going until it ultimately reach its logical terminus, which is not just that a few jobs are automated, but that basically all economic labor–with a few exceptions that we could talk about–is done better by machine, and where you would have a post-work condition where humans no longer need to do any work for the sake of earning an income.
And so, that’s kind of reaches the second layer of the onion, if you want. It’s a slightly more radical conception: not just some reallocations where the labor is going, but the actual cessation of the need for human work.
But even that, it’s just an intermediate layer in this onion. Because, once you start to think it through, what it really would mean if AI succeeded, and then all the other technological developments that would follow in the wake of machine superintelligence, because then this super-intelligent AI would be doing the further inventing and innovating and discovery at digital time scales.
And so, ultimately after you have superintelligence–and I think personally not that many years after you have superintelligence–you would have a condition, I believe, that starts to approximate technological maturity, where most of those technologies that we can see are in principle physically possible have been developed. Especially general-purpose technologies.
And then, it’s not just that we have the current human conditions, but with a bunch of clever robots going around, getting our groceries delivered and so forth, but, a whole host of other things also become outsourceable to machines; and you’re beginning to get closer to this condition of what I call a solved world, which is not just that the difficulties have been solved, but also almost like a condition where previous firm boundaries are starting to get dissolved.
So, you think about what it would mean, for example, if we had this kind of complete automation and technological maturity: It’s not just that the AIs could do your economic labor, but a whole host of other things as well that people fill their time with when they don’t have to work.
12:28
Russ Roberts: Now, it’s interesting to think about one’s own job. I used to find it rather marvelous that I liked my job. And I used to–as an economist, I would wonder in the history of human wellbeing, how many people enjoy their work? And I used to suggest that the proportion of the world that enjoys their work today is probably something close to an all-time high. Plenty of people don’t. But your book forces me to think about: Why do I enjoy my work? I happen to be very well-paid. If I weren’t well-paid, would I enjoy it as much as I do?
If people didn’t read the essays and books I wrote, for whatever reason: because they didn’t have to work and they didn’t invest in whatever wisdom I hoped to provide–they didn’t read my work, but I still enjoy the practice of setting my thoughts down on paper? And, as Bruce Yandle likes to say–the economist–‘I love my work but if they didn’t pay me, I wouldn’t do it.’
And, yet there are many things we get paid to do–perhaps–that we enjoy sufficiently for themselves. Would that be true of you, Nicholas Bostrom?
Nicholas Bostrom: Um, yeah. I mean I enjoy my work, and especially the–not so much the part that was involving dealing with the university bureaucracy. I think that part I could happily have elided. But the other parts, yeah. But, it is possible that a lot of the reason for that is that I feel it’s worthwhile that I’m accomplishing something with the work.
In other words, the activity has a structure where you do something, X–putting in effort, concentration–in order to achieve something else, Y, that’s separate from the activity itself.
So, like, you write a book, and then the result is there now exists this book written by you that hopefully can be enjoyed by other people and maybe effect some positive change in the world.
And also, you yourself, maybe, by putting in this effort grow intellectually as well through the work you did.
But so, there’s this outcome separate from the activity. And the fact that the activity has this outcome, this causal effect, might contribute to your sense that the activity was worth doing and probably to your psychological satisfaction.
But, if we now consider this same scenario in this condition of technological maturity, it is no longer clear for most activities that this for the sake of which would still be there.
So, you can think through the different types of activities that people build their day. Like, some people enjoy going shopping–I don’t quite understand them, but that’s a common right. So, then that exercises various kinds of human faculties. You have to remember where the good things were. You have to evaluate different options. Maybe you go with a friend and you can develop an understanding of how the other person thinks, etc., etc.
But, and imagine if there were instead some recommender system that could give you–that could find for you something that would actually fit you better than what you would discover if you went and looked for it on your own: whether it’s like some object for your house or some piece of clothing. And, it could also just buy it automatically without swinging it by you if it were sufficiently good.
So then, you would have this situation where, yes, you could perhaps still go shopping, but you know the whole time that the only result of this is that you end up with something worse than what you would have gotten if you had just not bothered: You know, pay more. It’s worse. It doesn’t fit as well.
Now, in that scenario, it is possible that the kind of appeal of the shopping activity would start to come off a little bit. It would seem perhaps a bit pointless.
And similarly for other activities that we tend to fill our leisure time with: for many of them, they have this same structure of you do X for the sake of Y.
So, you might think: Well, right now, even if you didn’t have to work for a living, maybe you would pull yourself over to the gym a couple of times a week because you want to remain healthy and fit and you can’t hire a robot to do the StairMaster on your behalf. Like, that’s something you have to do yourself. But the technological maturity, you could pop a pill that would induce the same physiological effects and also the same psychological effect after a good workout, like the relaxed, energized, calm, focus that some people enjoy.
And so, that too, like–
Russ Roberts: Allegedly–
Nicholas Bostrom: we have points on struggling in the gym for an hour if you could just have popped a pill and had this actually the same effect.
So, you can go through and start to cross out the activities that fill our leisure, or at least put the question mark on top of them.
And, a lot, I think, would be crossed out or questioned. And, we have then a kind of post-instrumental condition. Not just a post-work condition, but a condition in which it seems like all instrumental effort becomes odious.
17:57
Russ Roberts: I’ve mentioned my granddaughter a number of times. She’s almost two. It’s extraordinary to watch her growth. I get to spend a few hours with her every week; and I like to think I’m part of her maturation, her growth in consciousness, her linguistic ability, her passion for owls–you name it. I’m part of her life.
And, Eric Hoel, a past EconTalk guest, had a post recently on a Substack saying, ‘You can teach a two-year-old to read.’ It’s not easy. Maybe not every two-year-old can read. But it can be done. And, he has a nice, thoughtful essay on whether that’s a good idea, a bad idea. And, I have to confess that I think my granddaughter is very bright. She might actually be, I don’t really know.
But, it’s an interesting thought I had: Could I teach her to read in a few months of diligent phonetic stimulation? I showed her the letter P. I explained ‘papa’ has the sound of a P, etc., etc.
And, again, it’s incredibly fun, by the way–unbelievably fun, exhilarating. And, you talk about parenting as one of the things that might be something people would focus on in a world without technological limits or without material shortages.
And then, I think: ‘Why do I really want to teach her to read?’ Is the goal of this kind of–I mean is in the back of my mind, even though it’s probably deep below my consciousness, some kind of instrumental goal here that is not actually the joy of just teaching her to read? I don’t know.
Nicholas Bostrom: Yeah. I think these instrumental reasons are so suffused throughout our whole motivation system and psychology that it is quite hard even to imagine what it would be like if all of these instrumental reasons were removed. It’s like, you know, a little bug, like, has an exoskeleton that, like, the hard shell basically that holds all the squishy parts together.
And, I think analogously, we humans, our souls, have this kind of exoskeleton of instrumental reasons that has always been there throughout our evolutionary history. Right? We’ve grown up with all this instrumental reasons: We have to do this for the sake of that. That, if we remove all of that, it’s like a question of whether we’d just become amorphous blobs or whether we would still retain some kind of structure. Human lives, would they still have shape after we no longer need to do anything?
And so, we come now to yet another layer of the onion where we’ve kind of seen how at technological maturity you would have this post-instrumentality.
But, there’s one more property I think that would follow if you really think what it would mean to have technological maturity, which is ‘plasticity,’ I call it. But, basically the idea that we ourselves, the human biological organism, our brains and our psychology becomes malleable.
We would have the tools needed for reshaping ourselves according to our wishes–reshaping our bodies, but also our minds and placing ourselves in whichever psychological conditions we preferred. Like, a crude way might–you could imagine some super-drug without side effects and without addiction potential and a wide range of different drugs that had very precise, tailored effects that you just pick and choose in a moment-to-moment basis, what your emotional state would be.
But more than that, you could imagine direct brain manipulation technologies and presumably uploading into computers where you could directly edit our neuronal connectivity matrix, etc.
And so, this removes yet some additional reasons for doing stuff. So you might say, ‘Well, the reason I’m going to the gym is to remain healthy. The reason to brush my teeth is so they don’t get tooth decay.’ And, these things would remain; but then you think, ‘Well, no; you could take a pill to do that.’
But then, what about other things people do? Maybe people play golf because it’s fun. Or, they study mathematics so that they will then know mathematics.
But, here with this condition of plasticity, you see that those also would go away. Like, if the only reason why you’re playing golf is to induce a subjective state of pleasure and well-being and joy, then that would be a shortcut with the pleasure drug. And similarly, if the only reason for spending hours poring over the mathematics textbook and doing exercises is so that your brain someplace down the road is in a position where it understands algebra and calculus, etc., there[?] would be a shortcut, again, like some direct brain manipulation that would rewire your brain into a condition where it has these skills and concepts without the preceding effort.
And so, then you really end up in this quite radical condition of what I call the solved world, where you really have to think about which things are worth doing, not for the sake of something else, but because you regard them as intrinsically worthwhile.
23:53
Russ Roberts: This does sound a little bit pessimistic. Listeners, I will assure you there’s some cheerier thoughts coming from our guest. But I want to say one thing about the shopping example or any of these examples, which is: Would I want the thrill of the hunt, to find the item that I would ferret out myself even though it wouldn’t be as good as the one I could, say, find on the Internet using this amazing app.
And, that’s an open question. But it’s obvious that, at least today, you would much prefer, I think, that I bake you a loaf of bread, Nicholas, that was inferior in some dimension–up to a point–to one I could buy in a store: the fact that I crafted it, that I invested the time for you. So, we can imagine these bespoke activities that enhance our human connection that might persist, I think, in a world as they persist now, despite our–I don’t think it’s irrational to bake bread even though it’s cheaper at the bakery down the street or even cheaper and higher quality. Sometimes it’s fun to do things oneself.
But of course, you could imagine that there’s a pill I could add to the bread from the store that would make you–I don’t know, feel like it’s homemade or that you would like it just as much? I am not quite sure. Do you want to rule out that personal touch aspect of these activities?
Nicholas Bostrom: No. No. I think we’re getting towards something. I mean it’s worth, again, making the distinction there between you might bake bread because you enjoy it. Then if the only reason for doing it is the enjoyment in the sense of subjective positive affect, then there would be a shortcut to that. It would seem, like, unnecessary to get the whole kitchen dirty when you could just have pressed a button and felt exactly the same amount of enjoyment.
But, the idea that you’re pointing to with the doing things for the sake of others who might just value it because you did it–I think this is an important potential source of purpose in a solved world.
So, the structure of the kind of inquiry in the book is a little bit like first tearing things down, breaking things into parts, and then seeing what we can rebuild from these basic building blocks of human values.
So, at one point, yes, we do confront this sense of dissolution as it were–like, everything we normally just assume to be worthwhile and nice, there seems to be no point in doing it. And, it seems quite drastically unappealing this condition that we would reach in a solved world, like a meaningless existence where we’re just floating through.
But, I think that rather than trying to paper over that or trying to hide it, I would rather: let’s actually dive into that and really think through the full implications of that.
And then, we can, on the other side of that, try to conceive of an existence that would actually make sense even in these[?] radically transformed condition. And, I’m somewhat optimistic about there actually being something, but it would be slightly maybe quite profoundly different from our accustomed human existences.
27:27
Russ Roberts: I was pretty pessimistic before I read your book. I’m much more optimistic, which is rare that a book can have that big effect; so I salute you.
Before we move on to some of those more optimistic things, I just want to mention Keynes’s famous essay, “Economic Possibilities for our Grandchildren,” which he wrote in 1930, which is–one of the things that’s interesting about it: It’s not a particularly good economic time, 1930; but he wisely foresaw that the standard of living would multiply quite dramatically over the next a hundred years, or close to 2030.
He expected that that would lead to a lot less work. It’s led to a little less work since then per hour–I mean per week, per lifetime. So, he’s on the right track. But it’s not a solved world, this world we live in now. And, a lot, I think, of what he–he imagined a world where we could enjoy the pleasure of the moment without having to save, say, or postpone satisfaction.
It’s a very interesting psychological essay. I would add, as I have in the past, that it has a anti-Semitic twist embedded in it. But, put that to the side. He didn’t quite get it right, and he imagined a very different–he didn’t have your imagination, I’ll just say that. Do you agree? Is that a fair summary of Keynes’s essay, the difference between you[?] and your book?
Nicholas Bostrom: Well, I mean, I think maybe at the time it was maybe quite a radical, imaginative leap over the status quo at that time. And now it’s easier to imagine this future now, because we’re much closer to it. We can see AI unfolding almost week by week, and there have been a lot of work done between his time and now. [More to come, 29:36]