Will MacAskill of Effective Altruism Fame — The Value of Longtermism, AI, and How to Save the World | Transcription

Transcription for the video titled "Will MacAskill of Effective Altruism Fame — The Value of Longtermism, AI, and How to Save the World".

1970-01-01T06:21:35.000Z

Note: This transcription is split and grouped by topics and subtopics. You can navigate through the Table of Contents on the left. It's interactive. All paragraphs are timed to the original video. Click on the time (e.g., 01:53) to jump to the specific portion of the video.


Introduction

Start (00:00)

"Trig of action plan." That's what you need, Pao. Hello boys and girls, ladies and germs. This is Tim Ferriss. Welcome to another episode of The Tim Ferriss Show. My guest today is William McCaskill. That's M-A-C-A-S-K-I-L-L. You can find him on Twitter @willmccaskill. Will is an associate professor in philosophy at the University of Oxford. At the time of his appointment, he was the youngest associate professor of philosophy in the world. A Forbes 30 under 30 social entrepreneur, he also co-founded the nonprofits, giving what we can, the center for effective altruism, and the Y Combinator backed 80,000 hours, which together have moved over $200 million to effective charities. You can find my 2015 conversation with Will at Tim.blog/will. Just a quick side note, we probably won't spend too much time on this, but in that 2015 conversation, we talked about existential risk, and the number one highlight was pathogens, although we didn't use the word pandemic, certainly, that was perhaps a prescient discussion based on the type of research, the many types of research that Will does. His new book is What We Ode the Future. It is blurbed by several guests of this podcast, including neuroscientist and author Sam Harris, who wrote, "No living philosopher has had a greater impact upon my ethics than Will McCaskill." This is an altogether thrilling and necessary book. You can find him online, William McCaskill.com. Will, nice to see you again. Thanks for making the time. Thanks for having me back on. It's delight.


Exploring Personal Growth, Philosophy And Long-Term Thinking

I thought we would start with some warm-up questions to get people right into some details of how you think, the information you consume, and so on and so forth. We're going to begin with a few questions I often reserve for the end of conversations. We covered some of the other rapid-fire questions in the last conversation for people who want a lot on your bio, how you ended up being the youngest associate professor of philosophy in the world at the time of your appointment and so on. They can listen to our first conversation, but we spoke about a few books last time, and I'd be curious, what is the book or what are the books that you have given most as a gift and why? Or what are some books that have had a great influence on you? I know we talked already about practical ethics by Peter Singer and then superintelligence by Nick Bostrom last time, but do any other books come to mind when I ask that question? Yeah, so here are a couple. One is The Plessapas by my colleague Toby Ord, who I co-founded Giving What We Can With back in 2009, and it's on the topic of existential risks. So I see it as a compliment to my book What We Are The Future, and it details in kind of quite beautiful prose and also painstaking detail, some of the risks that we face as a civilization, from the familiar asteroids to the less familiar supervolcanoes and to the truly terrifying, which I also discuss in the book, and discuss how we might play and handle, like artificial intelligence, and, you know, engineered pathogens and engineered pandemics. And it also just talks about, you know, what we can do about them as well. And so I think it's just like absolutely necessary as I read. We'll talk, I guess, a bunch about some of those topics as we get into my work too. So I have, like, another kind of set of books, which are quite different, but they've had some of the biggest impact on just the background of my thinking over the last few years in very subtle ways. And that's Joe Heinrich's books, The Secret of Our Success and The Weirdest People in the World. And Joe Heinrich is a kind of quantitative anthropologist at Harvard, and his first book is just, "Why are humans the most powerful and ecologically dominant species on the planet?" And people often say, like, "Oh, it's our big brains." And he's like, "No." Our brains are several times the size of a chimpanzee's brain, but that's not the distinctive thing. The distinctive thing is that we work together, essentially. We're capable of cumulative cultural evolution, where I can learn something and then my children will pick it up for me, even if they don't really understand why I'm doing it. And that means that the way humans function, it's not like a single brain that's three times the size of a chimpanzee. It's tens of thousands of brains all working in concert, and now millions of brains over many generations. And that's why there's such a big gap between, like, chimpanzee, ability or intelligence and human intelligence. It's not a scale up of 3X, it's a scale up of 300,000. The hive mind of hominids. Basically, that's exactly right. So on this perspective, humans are not just another species that are like weird and not hairy and particularly sweaty and good at long distance learning. The start will commented that humans are the rational animal, and that's what made them distinct from other animals, whereas actually we're just very sweaty, and that's one of our most distinctive characteristics. And I like to think that I am therefore the most human of humans, because I'm the prettiest person I've met. So here's this book, and that alone just really blew my mind. It really made a big difference to how I understand humans. And he has this other book, The Weirdest People in the World, which is about the psychology in particular of weird people, Western educated, industrialized, rich and democratic, which are the subject of almost all psychology experiments, but they're not representative at all of most cultures. And in fact, they're very unusual among most cultures, much more individualistic, much more willing to challenge authority, even perceive the world in slightly different ways. And the overall picture you get from these two books is an understanding of kind of human behavior that's very different from the kind of economic understanding of human behavior, that's where all these just like self-interested agents going around to then kind of maximize profit for ourselves. Whereas on this vision, it's like nowhere, these cultural beings, we have like a vision for the world, and we go and like try and put that vision into the world. And that's what like the big fight, like the kind of big fights are about. And I think it has a much better explanation of history. Anyway, I'll stop there. When you said quantitative, I think you said quantitative anthropologists. Am I hearing that correctly? Yeah, that's exactly right. What is a quantitative anthropologist? I know those two words separately, and I can pretend like I understand what those mean together, but what does a quantitative anthropologist do? So you might know kind of evolutionally biology has these formal models of how kind of genes evolve over time. And you can, I don't know, it's hard to make predictions within this field, but at least you have these like precise formal methods that you like can start to kind of understand what's going on in terms of how organisms evolve. Now it turns out you can do the same thing but applied to cultures. I mean, Dawkins made this word meme very famous. And that kind of gets across the idea, although it's not quite right, because it's not like there's a single divisible unit of culture. But nonetheless, you can think of different cultures, kind of like different species. And some are like more fit than others, so some are going to win out over time. And you can apply the same sort of like formal methods that evolutionally biologists use to study evolution of genetics to the evolution of cultures as well. And Joe Henlek does that at least a little bit. All right, well, I'll have to take a look now.


How Dostoevsky’s Crime and Punishment changed Will’s life. (08:00)

We were talking about using the royal we you. We're talking about, I suppose I mentioned in passing, existential risks and threats. And I have a number of questions related to this, not surprisingly, but I want to touch upon first, perhaps an unexpected insertion. I have in my notes here, Crime and Punishment by Dostoyevsky, as a book that was important to you. And I would like to know why that is the case. So that book is actually what got me into philosophy, originally. Back when I was about 15, I read it. And I was at a time, very interested in literature, I wanted to be a poet and an author. And it was via that that I learned about this word philosophy. And I realized that like, oh, actually, you can just tackle the big ideas directly. You don't need to go via fiction. But I was also particularly interested in the time in existentialist philosophy. And this is something that honestly, like I kind of still bear with me, I'm kind of a bit unusual in that I often think to myself, could I justify my life now to my 15 year old self? And if the answer is no, then I'm going to get a bit like, or what are you doing? You're not living up to what earlier Will would have wanted for present Will. And the key thing, I think, for 15 year old Will, who was inspired by existentialism, was living an authentic life. And I still find that very liberating and empowering and inspiring. So some of the things I do, so for example, I give away most of my income, which is like a very unusual thing to do. And you might think, oh, that's like a sacrifice. It's making my life worth. But actually, I find it kind of empowering because it's like, I am making an autonomous decision. I am like, not merely kind of following the dictates of what social convention is telling me to do, but I'm like reasoning about things from first principles and then making a decision that's like genuinely authentically mine. And that was something I hadn't particularly pegged it to kind of acting model when I was 15, a load to some extent. But that was something that really moved me then. And yeah, honestly continues to move me today. So how would you just for, I often say for the listeners out there who may not be familiar, but if I'm being honest with myself, I have not studied existentialism. And I hear certain names associated with it, so I can kind of fake it until I make it and create the illusion of hope you're like, ah, you know, a kirker guard, I think maybe this person, that person, but what is existentialism as it is portrayed in crime and punishment or conveyed? Well, I think one of the things I liked about crime and punishment and dusty fcs work in particular is that it kind of, at least is wrestling with existentialism. So, and that can, word can get used in various ways. But here, one way of thinking about it is just the world as it is, like has no intrinsic meaning. Yeah. And yet we are like placed into it and have to make decisions. And that's this like absurd position to be in. And you can create your own meaning out of that through like radically free act, like authentic genuine act. And Dostoevsky in his work kind of wrestles between like free positions, I think. One is this like existentialist position. A second is just pure nihilism, which is just like actually literally, if you take it seriously, like, and there's no guard, then everything is permitted. There's no reason to do anything, not even like reason created or, you know, from yourself. And then third is this religious position, which I think he actually ultimately endorses. And it's almost like nihilism as like a proof, like the rejection of nihilism, like therefore guarantees that you should be like the religious. QED, God. Yeah, exactly. Yeah. Well, it's like life is meaningless unless God exists. And I'm now describing it slightly pascally in terms, but like you may as well act as if there's a God that is giving meaning to life.


Maintaining optimism in the age of doomscrolling. (12:33)

We're not going to spend a whole bunch of time on it now, but in our last conversation, we talked about Pascal's wager, but also Pascal's mugging, if I remember correctly. Yeah, very good. Yeah. Something along those lines. So we won't take a side alley down into Pascal's mugging just now, but I said I had two things I wanted to ask you about. The first was crown punishment, which I think we've covered. The second before we jump into our longer conversation, which will go all over the place, and I may ask still some of the shorter questions. When people hear existential threats when they hear super volcanoes, AI, man-made pathogens, etc, I think that there will likely be an apprehension, perhaps, a little seizure of the breath for some people listening who might think to themselves, "My God, this is just going to be audio doom scrolling. This is just going to come away from this conversation with higher blood pressure, more cortisol." My impression of you in the time that we've spent together is that you are not nihilistic, you are not apathetic, you are not pessimistic, you are quite the opposite of all of those things in some respects. How do you do that? Is that just will out of the box? And that's just how you came programmed. Is there more to it? And this is I think a crooks question because I don't see in the people, say in my audience, including those who are very competent, effective action, if they don't have some degree of optimism or belief that they can exert change. Right, so could you just speak to that because I know I also succumb to just getting water-bordered bad news all day from around the world? I'm like, "Shh, I can't do. I cannot put a salve onto all of this for all of these people, and it can be overwhelming. So how would you respond to that?" Great. I think there's two things that motivate this. One is just the desire to actually make the world better, and then second, I'll call low standards. So, on the first side, when I first, so age 21, and I'm like, "Man, I'm about to really start my life. I'm trying to look for like, I wanted to act morally. I'm trying to look for different causes." I bounce into a lot of kind of classic, the sorts of classic causes that you'd find for someone socially motivated in a college campus, like vegetarian society, some left-wing politics, climate change stuff. I found there was very little in the way of action. There was an awful lot of guilt, and an awful lot of talk about the problems, but not that much in terms of like, "Hey, here are the solutions. This is how you can actually make the world better, and this is what we should do." But if you actually care about what making wanting to make the world better, and that's the key motivation, the size of a problem, and really thinking about suffering, it can be important, especially if it's motivating you. But the ultimate thing is just what do you do? Something could be the worst problem in the world, but if there's nothing you can do, then it's just not relevant for the purpose of action. And that, therefore, really makes me think, in the first instance, always about, "Okay, well, what's the difference we can make?" Not like, "How scary are things, or how bad are things?" But instead, like, "How much of a difference can we make?" And there, it's like very positive. So in the last podcast, we were focusing in particular on global health and development. And what's the difference you can make there? Well, if you're a middle-class member of a rich country, it's on the order of saving dozens, hundreds, maybe even thousands of lives over the course of your life, if you put your mind to it. That's huge. Now we're talking about existential risks, and the long-term future of humanity. What's the difference you can make? You can play a part in being pivotal, in putting humanity onto a better trajectory, for not just centuries, but for thousands of millions, or even billions of years. The amount of good that you can do is truly enormous. You can have cosmic significance. That's pretty inspiring. And so, yeah, when you think about the difference you can make, rather than just focusing on the magnitude of the problems, I think there's every reason for optimism. And then the second aspect I've said was low standards, which is just, you know, what's a world that you should be sad about, what's a world you should be happy with? Well, in my own case, I think, like, look, if I came into the world, and when I leave it, the world is neither better nor worse. That's like zero. I should be indifferent about that. If I can make it a bit better as in virtue of my existence, hey, that's pretty good. And the better, the better, the more good I can do on top of that, the better. And I think I have made it much better. Like, you know, I'm not zero. I'm like, positive. And so all of the additional good that I potentially do feels like a bonus. And so similarly with humanity, when I look to the future, what's the level at which I'm like, ah, it's indifferent. That's where just like the amount of happiness and suffering in the future kind of cancel out. And relative to that, I think the future's going to be amazing. Like, already, I think the world today is like much better than if it didn't exist. And I think it's going to be a lot better in the future. Like, even just the progress we've made over the last few hundred years, people today have like far, far better lives. If you extrapolate that out just, you know, a few hundred years more, let alone thousands of years, then there's at least a good chance that we could have a future where everyone lives, not just as well as the best people off alive today, but maybe tens, hundreds, thousands of times better. Yeah, I mean, Kings, Kings a few hundred years ago didn't have running water, right? No, they didn't. They didn't have air conditioning. They didn't have an aesthetic. No antibiotics. Oops. If they were gay, they had to keep it secret. They could barely travel.


What is effective altruism? (18:41)

Yeah. Lots of things we easily take for granted, which we can we can come back to because it may be related, but why don't we take 60 to 120 seconds just for you to explain effective altruism. Your name is often associated just so we have a definition of terms. And people have some idea of the scope and meaning of effective altruism, since you're considered one of the creators or co creators of this entire movement. If you wouldn't mind just explaining that briefly, and that way people will have at least that as a landmark as we go forward. Effective altruism is a philosophy and a community that's about trying to figure out how can we do as much good as possible with the time and money we have, and then taking action on that basis. So putting those ideas into practice that I actually tried to make the world better as effectively as possible, whether that's through our donations, with our careers, with how we vote, with our consumption decisions, just with our entire lives. What have been some of the outcomes of that? Yeah. So I've been promoting these ideas along with others for over 12 years now. We've raised or moved well over the billion dollars to the most effective causes. So that means if we take just one charity that we've raised money for against malaria foundation, we've protected over 400 million people, mainly children, from malaria. And statistically, that means we've saved about 100,000 lives, or maybe a little more, which is the size of a small town, about the size of Oxford. And that's just one charity. There's several more within global health and development, I think, in terms of other cause areas that we've focused on within animal health and welfare. Hundreds of millions of hens are no longer than cages because of corporate cage for the campaigns that we've helped to fund. And then within the field of existential risks, there are, you know, it's not as easy to say, "Oh, we've done this concrete thing. This thing would have killed us all that we avoided it." But we have helped make AI safety a much more mainstream field of the search, people taking the potential benefits, but also the risks from AI, like much more seriously than they were. We have also invested a lot in certain pandemic for paidness measures. Again, it's kind of still early stages, but some of the technology there are things I think have really promising potential to action, at least making sure that COVID-19 is the last pandemic we ever have. One of the many things I appreciate about you, and also, broadly speaking, many people in the effective altruism community/movement, is the taking of a systematic approach to not just defining but questioning assumptions and quantitatively looking at how you can do good, not just feel good, if that makes sense.


Resources for maximizing the impact of your philanthropy. (21:22)

And it seems obvious to anyone who's in the community, but the vast majority of philanthropy or charity, broadly speaking, is done without that type of approach from what I can tell. And it's really worth taking a closer look for those people listening. Just a few URLs you'd like to mention for people who'd like to dig into that, and then we can move into some of the more current questions. For sure. So if you're interested in how to use your career to make the world better, then 80,000hours.org is a terrific place to go. I'm a co-founder of that organization, gives in-depth career advice, and one-on-one career coaching as well. If you're interested in donating some of your money, then giving what we can.org encourages people to take a giving pledge, typically 10% of one's income or more. It's a great way to live. If you're interested in donating to effective charities, then givewell.org is the single best place for donating to global health and development charities. Let's givewell.org. There's also the effective altruism funds or EA funds that allow you to donate within animal welfare and existential risks and kind of the emotion of these ideas as well. All right. A few more calisthenics, then we're going to go into the heavy lifting, the max squats of long-termism.


How adopting a check-in system has most improved Will’s life. (23:13)

All right. Here we go. In the last, say, five years, you can pick the time for him, but recent history, what new belief behavior or habit has most improved your life? I think the biggest one of all, and this was really big, doing writing the book, which was this enormous challenge. It was like my main focus for two years over the course of the pandemic, was evening check-ins with an employee of mine who also functioned a bit like a productivity coach. Every evening, I would set deadlines for the next day, both input and output. Input would be how many hours of track lighting I would do, where going to the bathroom did not count. A really big day would be six hours, sometimes very occasionally I'd kind of get more than that. And also output goals as well. So I'd say I will have drafted this section or these sections or I will have done such and such. I would also normally make some other commitments as well, such as how much time do I spend looking at that on my phone? Can which caffeine am I allowed to drink? Do I exercise? Things like this. And Laura Pomerius, who is doing it, is wonderful and the nicest person ever. She just never beat me up about this. But I would beat myself up and it would make me, it was incredibly effective at making sure I was just like actually doing things because I, like many others, find lighting. It's hard, it's hard to get motivated, it's hard to keep going. And sometimes, I don't know, I'd have gotten drunk the night before, let's say, and it was a Sunday. And normally, you just, it would be right off for the whole day. I think like, oh no, I'd just be so embarrassing at 7pm to have to tell Laura, like, yeah, I didn't do any work as I got smashed. And so instead, I would feel hung over and I would just keep typing away. And that was just huge. I mean, I think it increased my productivity. I don't know, it feels like 20% or 25% or something, just from these like 10 minute check-ins every day. So these were 10 minute check-ins, seven days a week? What was the kids? I was working six days a week, so, but yeah, issue was doing something else at the weekend. We can check in. Ooh, right. So the format would be, walk me through 10 minutes, would be the first five minutes. Here's how I measured up to what I committed and here's what I'm doing next. Exactly. So it would be a review of the day. So yeah, did I hit my input goal, my output goal, how much caffeine did I think, did I exercise? And then also, like, was I getting any migraines or back pain, which are two kind of ongoing issues for my productivity? Next would be a discussion of what I would try to do the following day. And interestingly, you might think of a productivity coach as someone who's like, you know, really putting your nose to the grindstone, whereas with Loyl, that's kind of the opposite because my problem is that I beat myself up too much. So she's like alluring ET out of the closet with the Reese's Pieces candy. Yeah, so I would be like, oh, I got so little done today. So I'm going to have to just have a 12 hour day tomorrow or something. Or like, I'll work through the night or something like that. And she's like, that doesn't make any sense. Like, you know, we've tracked this before. And when you try and do this, maybe you get like an hour of extra work where you feel horrible days afterwards. And so she would be very good at like, countering bullshit that my brain would be saying, basically. It's a couple of things. Caffeine, what were your parameters on caffeine? Like what were the limitations or minimums?


Caffeine limits. (26:43)

I don't know how you said it on caffeine. And then how did you choose this employee specifically for this and why? So caffeine, I mean, I think a big thing is just if I do think too much, I'm likely to get a migraine. So I set my limit at three asbestos worth. So about 180 milligrams of caffeine. And I'm very sensitive. So it's like, under nadys legitimate for a sensitive person. Exactly. So that's like the max that I do, whereas double asbestos was fine. But then it's like shading in between, I'll be like very cautious about. And then how did I choose this person? I think it's like a very subtle thing, the kind of the poor or personal fit you have with someone who can be a good coach, where she kind of knew me well enough that she knew the ways to like push me around and was like, yeah, kind of the combination of like, maybe I call it friendly pushiness or something was like perfect. You know, it could be very easy to go along on either side of that line. All right, sounds like any evening check in. All right. Who was my victim going to be? All right. Maybe we can spy it. Well, I know, I know, I know it's, I know it's four in the morning, but I had to call you for my evening check in. We're in different time zones for people who may not have picked up on the, that is not a New Jersey accent that Will has.


Effective back pain relief. (28:08)

Okay. Comment sidebar on low back pain. I know this came up in our last conversation. Have you not found anything to help? And I may have some suggestions if you would like suggestions, but have you found anything to help? Actually, I've almost completely fixed it. So it was just, I mean, I was working, you know, sitting in a chair, especially, you know, pandemic and like a book for like eight hours a day, but there's actually only one period that I started getting lower back pain. At the moment, our conversation knew they commended me these boots so they could hang up, sit down. And I did buy them and I confess I never used them. So I'm sorry. Family adherence failure. No, that's a failure. My recommendation, if it's not going to be used, it doesn't make any sense for me to recommend it. But what I did do, in the end, I just developed my own workout routine, where I got advice from physios and so on. I talked to like loads of doctors. In general, people just aren't really engaging with what your problems are and like self experimentation, I think was just better. The other thing is just all of this takes loads of time. And like, if you're a time fast individual, so I mean, firstly, the advice is often like geared towards like old people. So it's like very like easy stretches or like basic work movement that most people aren't doing. And then secondly, it's like, man, you want to do all of this. It's like two hours or something. Like, I can do this more efficiently. So I developed my own routine, which involves standing on a Bose ball. So it's all on a Bose ball. I've got two free weights. I do a squat. I'm sitting in like squat position as the resting position. That's very good because it stretches your hip flexors. And for those people who can't see, well, he's got his hands in front of his chest. Yeah, imagine like a prairie dog. But really, I think what that symbolizes is he has the dumbbells in front of his chest, like a goblet squat, if people know what that is. Yeah, exactly. With my legs wide kind of knee elbows in between your knees so that your legs are kind of slayed out like that and you'll feel a stretch on your hip flexors. So cultures that squat to sit actually experience lower rates of back pain. So that was the kind of inspiration there. And then from there, standing up squat, do bicep curl up into a shoulder press, go down, then dead left going into an up-fight row. And then that's kind of what, and that's all in a Bose ball. And the thoughts here are the like strengthening your entire kind of anterior pelvic chain. So I think my hypothesis was like, why was I getting this? Because I was one of an idiot young male who was like, why would you work anything out apart from your beach muscles? What would be the point of that? And that like majorly distorted my posture. And so I would do that. So kind of one of them every 20 seconds in like two sets of 10 minutes. And then that combined then also with core work. So plank in particular, really just, I think, sorted things out because it's all about just like, you know, had just bad posture for 25 years made worse by like very poor focus at the gym. And so it's like this long process of reconfiguring your body so that it's like, it makes more sense. And in particular, as we talked about, I had anterior pelvic tilt. So my gut stuck out, my pelvis was too far forward. And so then it's like your glutes tearing that back and then stretching out your hip flexors. Oh, and I invented my own stretch as well. So for the listeners who don't know, I actually, I was previously married and I took a different name, took my wife's grandmother's maiden name. So my name wasn't always McCaskill, it used to be couch. And so I named this stretch the will couch in honor of my former self in roles standing and hooking your, you stand up, you hook your foot into your two hands, and then like press out, like kind of extend your leg, but pushing against your two hands. And that stretches out this muscle that is goes all the way from your kind of pelvis up your back. And I've not found any other stretch that stretches that particular muscle. And that was the one that was like really causing all the pain. So you do this standing? Yeah, it's standing. That's right. Yeah. It's like a kid and play dance move. Okay, so people may just, I'll put my liability hat on, I'll just say maybe start on the ground to try this. So you don't get your foot stuck and topple over like a, you know, that an army figurine onto your head. But yes, I can see how that would work. Anyone would say that I'm not a professional workout coach. I can't wait for the Willcrouch YouTube instructional fitness series. So I did take on this role in the early stages of the pandemic, the house I was in, I would go outside every, every lunch at 1pm and put on my best Scottish accent. And I'd be like, right, you, you, you, you, you, you, you never made it to YouTube though. Well, you know, it's never too late. So a couple of things real quick. The first is these exercises, did you do them every day in the morning? Did you do them midday? How many days a week at what time of day? So I almost always work out just after lunch. People always complain to me that's like, oh, you'll get a sore stomach or something. I'm like, but I don't never happens. But I deliberately time it because I have a real energy dip just after lunch. And so doing something that's just not work makes a ton of sense. Yeah. Yeah. Plus after sitting for a few hours, you can break up the two marathons of sitting. Exactly. And I'll make one, one other recommendation for folks who may also suffer from occasional or chronic low back tightness, which has been an issue for me also. If I sit a lot and it ends up affecting my sleep most significantly and can cause that type of anterior pelvic tilt and lordosis. So if your gut is sticking out and you look like you're fat or pregnant, even though you are not, perhaps that means your pelvis is, is pouring forward. So if you think about your pelvis as a goblet or a cup full of water, if you're pouring that water out the front, you have interior pelvic tilt. And one of the causes of that or contributing factors can be a really tight iliapsoas or iliakus that, that in some fashion connects to the lower back of the lumbar. And so you get this incredible tightness slash pain. For me, it can cause tossing and turning at night and really affect my sleep. And the device that was recommended to me a few times before I finally bit the bullet and got it with something called the Sohright PSO-RITE. It is the most expensive piece of plastic you'll ever buy. But worth it at something like 50 to $70 for self release of the SOHAS, which is incredibly difficult to achieve, I find incredibly difficult to achieve by yourself otherwise. And most a lot of soft tissue therapists are not particularly good at helping with it, nor is it practical really didn't necessarily have that type of work done every day even if you could. So the Sohright is helpful. All right, so let's move from personal long-termism, making sure that you're able to function and not be decrepit when you're 45 into the broader sense and discussion of long-termism.


What is longtermism, and why did Will write What We Owe the Future? (35:49)

What is long-termism? And why did you write this book? Well, long-termism is about three things. It's about taking seriously the sheer scale of the future that might be ahead of us and just how high the stakes are in anything that could shape that future. It's then about trying to assess what are the events that might occur in our lifetimes that really would have impact not just for the present generation, but they could potentially shape the entire course of humanity's future. And then third, then I figure out like, okay, how do we ensure that we can take actions to put humanity onto the right path? And I think you're exactly right to talk about personal long-termism and the analogy there, because in the book, in what we are the future, I talk about the analogy between the pleasant world and humanity and an included teenager, like a reckless teenager, where there are things, you know, what are the really high stakes decisions that a teenager makes? It's like not what you do at a weekend. Instead, it's the sort of, it's the decisions that would impact the entire course of your life. So in the book, I tell a story where I was quite a reckless teenager, I nearly killed myself, climbing up a building. That was one of the biggest decisions, dumbest decisions, like I ever made, because if I had died, then it would have been 60, 70 years of life that I would have lost. In the same way, if humanity dies now, you know, if we cause although an extinction or the end of the coverable end of civilization, such as by a worst-case pandemic, then we're losing, well, not just 70 years of life, it's thousands, millions, even billions of years of future civilization. And so similarly, if, you know, I made decisions as a teenager that affected the kind of whole course of my life, like whether to become a poet or philosopher or something, you know, could have become a doctor. And similarly, I think in becoming essentially, in our lifetime, humanity potentially makes decisions about, you know, how is future society structured? What are the values we live by? Is society a liberal democracy around the world? Or, you know, is it a totalitarian state? And how do we handle technologies like AI that I think could impact the very, very long run? So I want to read just a paragraph that you sent me, which I found thought-provoking because it's a framing that I had not heard before.


Future generations matter. (38:21)

And here goes. Imagine the entire human story from the first homo sapiens of East Africa to our eventual end, represented as a single life. Where in that life do we stand? We can't know for sure, but suppose humanity lasted only a tenth as long as the typical mammalian species, even then more than 99% of this life would lie ahead. On the scale of a typical life, humanity today would just be six months old. But we might well survive even longer for hundreds of millions of years until the earth is no longer habitable or far beyond. In that case, humanity is experiencing its first blinking moments out of the womb. And I appreciated this framing because my feeling, at least with my audience of listeners, is that there's a small percentage who are rushing headlong into battle with some vision of long-termism and feel committed to fighting the good fight. And a non-trivial percentage have decided it's too late, right? They've decided that the end is nigh. We are the frog in the water, slowly heating. That will be boiling before we know it. And I find this at least, whether we put aside for the second how people might find fault with it or pick at it, a useful counter frame, right? Just to even sit with for a few minutes, why do you think it's important to at least consider that something like this is plausible? Maybe it's not 90% likely, but let's just say it's even 10%, 20% likely. Well, it's so important just because future generations matter, future people matter, and whatever you value, whether that's well-being or happiness, or maybe it's accomplishment, maybe it's great work of art, maybe it's scientific discovery, almost all of whatever you value would be in the future rather than now, because the future just could be vast indeed, where, you know, again, if you look at like what has been accomplished since the dawn of humanity, well, dawn of humanity was the end of a thousand years ago. Agriculture was 12,000 years ago, industrial revolution was 250 years ago. And yet, even on the scale of a typical mammal species, we have 700,000 years to go. Now, we're not a typical mammal species, we could last only a few centuries, we could last 10 years if we really do ourselves in in the short term, but we could last much longer, and that just means that all of what we might achieve, all of the good things that humanity could produce, they're basically in the future. And that's really worth thinking about taking seriously and trying to protect and promote. You know, one thing that you and I were chatting a bit about, I brought it up before we started talking, is the question of if it is possible to make, let's just call it altruism, or in this case, long-termism, right, investing in a future we will not necessarily most likely see ourselves.


Finding the line between apathy and fatalism that spurs action toward ensuring there’s a future. (41:30)

Can you make that self-interested? How do you position it such that it appeals to the greatest number of people possible since our collective destiny depends on some critical mass of people taking it seriously, right? It can't, probably isn't one person. We're not going to get say nine billion people. So, how many do we hope to embrace this philosophy? And is it possible to position it as self-interested? And this is going to be a bit of a ramble, so I apologize in advance. But when you were talking about Dostoevsky and nihilism almost as a proof and him ultimately landing on God, like, yeah, you kind of need to something resembling God to make sense of this sea of uncertainties so that you can maybe stabilize one's self and feel a sense of meaning. It brought to mind something I read very recently, and I apologize, this is again going to be a bit of meander, but this is something that Russ Roberts included, so famous for econ talk podcast. In an article he wrote called My Twelve Rules for Life. Now, he is not sure this is the best descriptor, but culturally, and I would think religiously Jewish, so he has that as a sort of lattice work of sorts. But number two in his Twelve Rules for Life was find something healthy to worship, and I'm just going to take a second to read this. He quoted David Foster Wallace, and I'm going to tie this into what I just said in a second, because here's something else that's weird but true, and the day-to-day trenches of adult life, there's actually no such thing as atheism. There's no such thing as not worshiping. Everybody worships. The only choice we get is what to worship, and the compelling reason for maybe choosing some sort of God or spiritual type of thing to worship, be it JC or Allah, be it YH, WH, not sure what that is, or the wicked mother goddess or the four noble truths, or some inviolable set of ethical principles that is that pretty much everything else you worship will eat you alive. If you worship money and things, if they are where you tap real meaning in life, then you'll never have enough, never feel you have enough. It's the truth, worship your body and beauty and sexual allure, and you'll always feel ugly. And when time and age starts showing, you will die a million deaths before they finally grieve you. On one level, we all know this stuff already. It's been codified as myths, proverbs, cliches, epigrams, parables, the skeleton of every great story. The whole trick is keeping the truth up front and daily consciousness. Okay, and then dot dot dot, but the most important thing to remember is not to worship yourself, not as easy as it sounds. So I'm wondering if long-termism in a sense doesn't need to be spun to envelop self interest. If it is basically something to worship, it gives you purpose when there is so much uncertainty and chaos and entropy around us. Anyway, long TED talk, thank you for coming. But what are your thoughts on any of that? And the over-arching question is, how do we make long-termism catch to have some critical mass of people who really embrace it? I think there's a really important insight there, actually one made by John Stuart Mill in a speech to Parliament at the end of the 19th century. And he asks this question like, what should we do for posterity? After all, what has a posterity ever done for us? And then actually he makes the argument like, "Posterity's done a lot of things for us because the projects we have only have meaning in so far as we think that they might contribute to this kind of relay race among the generations." Here's a thought experiment, this is this film, "Children of Men" and "In It" people just unable to reproduce. And so it's not that anyone dies, there's no catastrophe that kills everybody, but there's no future of human civilization. How would that change your life? And I think for many, many people and many, many projects, it would just rob those projects of meaning. I certainly wouldn't be as nearly as interested in intellectual pursuits or like trying to do good things or so on. I mean, maybe I would to some extent, but for a lot of things it seems like, "Oh, they have meaning because, you know, take scientific inquiry." There is this same eye-built cathedral of knowledge that has, I've inhabited from all of my ancestors that has then been passed to us. And it is incomplete. So we've got general relativity and we've got quantum theory and they're amazing, but we also know they're incomplete and like maybe we can work harder and like see farther and build the cathedral a little higher. But if it's like, no, actually, it'll just get torn up. It's kind of like, "Oh, you're painting an artwork and you can add to the painting a bit and like it's going to just go in the shredder the day afterwards. You're not going to be very motivated to do it." And so one thing I think that a lot of people find motivating is, yeah, this thought that you're part of this grand project, much, much grander than yourself, have then a build, a good and flourishing society over the course of not just centuries, but thousands of years. And that's one way in which our lives have meaning. So what do you hope the effect will be of people or on people who read what we know the future?


What Will hopes readers take away from What We Owe the Future. (47:22)

What are you hoping some of the things will be that they take from that? The number one thing is just a worldview that's what my colleague Nick Boslin calls "getting the big picture of Uffleivite." So there are just so many problems that the world faces today, so many things we could be focusing on and paying attention to. But there's this question just, "Well, what's most important? What should be taking most of our attention?" And this, the ideas in the book, I hope, give a partial answer, which is, well, the things that are most important are those that really shape the long-term future of the human project. And that really now is things down, I think. So that's kind of broad, kind of worldview. More specifically, though, I would like it to be something that guides the decisions people make over the course of their lives. So I think the biggest decision people make are what could the early pursue. So do you go and, you know, become a management consultant or a financier and make money and live in the suburbs? Or do you instead, like, pursue a life that's really trying to make the world better? And if so, then what problems are you focusing on? Where it seems to me some of the bigger problems are the development of very advanced, or the biggest issues or events that will occur in our lifetime. Are the development of advanced artificial intelligence, in particular artificial intelligence, that's as smart as humans or maybe considerably smarter? I think that has a good claim to being one of the most important technological discoveries of all time once we get to that point. And that point, very good chances in the coming decades. A second is the risk of worst of, you know, very catastrophic pandemics, things that make the far worse than COVID-19, which again, I think are just in the horizon because of developments and our ability to create new viruses. And a third is a third world war, which again, if you look at history and look at leading scholars and delaying models of war, I think it's like a really pretty good chance we see a third world war in our lifetime, something like one in three. And I think that could quite plausibly have just unparalleled disruption and misery on the world in the limit just being the end of civilization, whether that's because of nuclear warheads scaling up 100 fold and being used in an all-net nuclear war, or because of the use of bio weapons. And these are all things that like, like, smart people who read this book could go and work on. I'm aware that that again kind of sense, kind of bleak, but perhaps the final thing is like, there is this positive vision in the book too, which is that if we avoid these threats or manage these technological transitions well, we really can just create a future that's truly amazing. And I, you know, this is present kind of throughout the book. I did feel like I hadn't fully given it as due. So there's a little Easter egg in the book as well, about in the final page QR code that sketches a little vision of a positive future in an extraordinary form. But maybe that's the final thing of all in terms of this worldview, is appreciating there's so much at stake that are enormous risks that we face or threats that we face that we need to manage. But if we do, then we can create a world that is flourishing and vibrant and wonderful for our grandkids, for their grandkids, for their grandkids. What is Value Lock-in and could you give some historical examples?


Analyzing Potential Threats And Opportunities Over Next Decade

What is value lock-in? (51:08)

So Value Lock-in is when a single ideology or value system or kind of set of ideologies kind of takes control of an area or in the whole world and then persists for an extremely long time. And this is one thing that I think can have very, very long lasting effects. And we've already seen it throughout history. And so in the, in what we have a future, I give a story of ancient China. So during this period that's known as the Hundred Schools of Thought, the Zhu dynasty had fallen and there was a lot of fragmentation, like ideological fragmentation in China. And wandering philosophers would go from state to state with a package of kind of philosophical ideas and moral views and political policy recommendations and try and convince like political elites of their ideas. And there were four main schools. There were the Confuitions that we're kind of most familiar with. The Legalists, which are kind of like Machiavellian, like political realists, just, you know, how do you get power was the main focus of them. Taoists, who are these kind of more, somewhat more spiritual, like acting in accordance with the way with nature, like advocating spontaneity, honesty. And then finally, the Moists, which I read and I'm like, wow, they were kind of similar to the effect of Altruus, except in ancient China, where they were about promoting good outcomes and good outcomes impartially considered. They forewent much like fancy kind of spending on luxury or ritual. So their funeral rites were very modest. They wore very modest clothes. And they were just really concerned about them to make the world better. And so they created a paramilitary group in order to defend cities that were under siege, where the reasoning being that if defensive technology and defensive strategy was so good, then no one could ever wage a war, because no one could ever win. And so there was this great diversity of thought. But what happened? One state within China, the chin, influenced by legalism took over and tried to essentially make legalism state orthodoxy. And the Emperor Chin declared himself a 10,000 year emperor wanted this ideology to possess indefinitely. It actually only lasted 14 years, because there was a kind of counter the billion. And that was the start of the Han dynasty, which then successfully did, basically over the course of a while, quell kind of other ideological competition, and instead implemented Confucianisms like this is the official state ideology. And that persisted for 2000 years. And that's kind of just one example among many. Over and over again, you see what's the kind of ideology or belief set of ruling power, whether that's Christians or sorry, I mean Catholics or Protestants, or is it communism of the Khmer Rouge of Stalin or national socialism of Hitler. Once that ideology gets into power, people with those ideology get into power, they quickly try and stomp out the competition. And the worry is that that could happen with the entire world. So again, I spoke of a risk of Third World War. Well, what might happen as a result? One ideology could take power globally after winning such a war, implement a world government or world state, or at least dominant world ideology. Then within this situation, with as much less ideological competition, and at least one reason why we've gotten model change and model progress over time, which is in virtue of having a diversity of model views that they're able to like kind of fight it out in the ideal circumstances, the kind of the best argument wins, we would no longer have that. And so if there was a single kind of dominant ideology in the world, that could persist for an extremely long time, I think. And if it was long, which is quite likely to be wrong, because I think most of our model views are probably wrong, that would be like very bad indeed. Yeah. I mean, just to give you an idea, I mean, this is not exactly ideological, but you mentioned the Han dynasty. And Mandarin Chinese, one way to say Mandarin Chinese is Hanyu, which is the language of the Han people. Like, Hanyu pinging is the romanization system used, which most people see in with the diacritical march for tones for Mandarin Chinese. So these things can last a very long time indeed. Do you have any other examples of a value lock in that could be past tense historical examples or attempts that are being made currently that you think are worth making mention of? Could be either. I mean, historically, one particularly salient example was when the Khmer Viyush took power in Cambodia, Paul Pot, just very systematically, like anyone who disagreed with the kind of party ideology would be, I mean, generally would just be executed. So 25% of the population were killed in Cambodia. And he mean, again, it's like very transparent what's happening. He has this quote, "pure the father party, pure the fire the army, pure the fire the cadres." So it's just very clear that's what going on is, you know, almost like a species or virus can be taken over and like other competitors get wiped out. This one ideology takes over, competitors are wiped out. Similarly, if we look at like British history with different times Catholics and Protestants taking power, there was one act passed called the act of uniformity with Protestants saying like, okay, Catholicism is now banned in this country. And again, it's like very boldly named. And in general, just if you have a particular model view, then you are going to want everyone else in the world to have that particular model view as well. So of AI pathogens, let's just say bioweapons, we can include in that World War III, how would you rank those for you personally in terms of concern over the next?


Most concerning threats projected over the next 10 years. (57:13)

Let's call it, let's call it 10 years. Over the next 10 years, I'd be most concerned about AI. Over the next 50 years, let's say my lifetime, I'd be both most concerned about AI and war. Development and AI, the reason I say that is like, wars are most likely when two countries are very similar kind of military power. And the kind of historical rate of like one country going to war with and like one major power in the world, like one of the big economies of the world going to war with another when it kind of gets overtaken economically or militarily. It's pretty high, some ways of modeling, but size like 50%. But I think that's more likely to happen not kind of within the next 10 years, though it's definitely possible that there would be some kind of help that you have war such as between the US and China. Even though the risk of war between the US and Russia is definitely higher than it has been in the last, I guess, 30 years, potentially. I still think the odds are like quite low, thankfully. With AI on the other hand, I think the chance of very rapid and surprisingly rapid developments in AI within the next 10 years are higher than any 10 year point after that. So as in 2020s, it's more likely there'll be some truly transformative development than 20 30s or 20 40s is my kind of view. And that's for a couple of reasons. One is that if you look at how much computing power different brains use, and you compare that with how much computing power the kind of current language models use, or the biggest kind of AI systems use, the biggest AI systems use computing power over approximately the brain of a honeybee. It's kind of hard to estimate exactly, but that's kind of where we are, which is a lot smaller than you might think as much smaller than I thought. And you might think like, okay, it's the point and time where you've got AI systems that are about as powerful as human brains. That's like a really crucial moment in time, because that's, you know, potentially the point and time at which AI systems just become more powerful than us, or at least approximately then when we start to get overtaken. And that, again, it's like very uncertain, but there's a decent, pretty good chance that happens in something like 10 years time. And now it's very hard to do technological prediction. I am not making any confident predictions about how things go down, but to least something we should be paying attention to, just from a kind of outside perspective, if you think, oh yeah, and then we're at a point where we're like training these AI systems that are like doing as much computing as the brain is. That's like, okay, well, it means maybe they're going to be just, you know, of a similar level of power than ability as human brains. And then that's really big. And that's kind of big for the few reasons, I think. One is because it could speed up rates of technological discovery. So historically, we've had like fairly steady technologically, they have an economic growth. That's actually over a couple of hundred years. But that's because of two things happening. One is that like ideas get progressively harder to find, but we throw more and more to searches. So we have like a bigger population, we throw a larger percentage of the population at them. If instead we can just create engineers and research scientists that are AI systems, then we could rapidly increase the amount of R and D that's happening. And what's more, perhaps they'd be much, much better at doing research than we are. Like, you know, human brains are definitely not designed for doing science, but we could create machines that really are. And in the same way that, you know, go, the best AI systems are far, far better than even the very best human systems now. The same could happen within science. And if you plug that into like the standard economic models, you get the conclusion that, okay, suddenly things start really moving like really very fast. And you might get many centuries worth of technological progress happening over the course of a few years or a decade. And that could be terrific. That's what a lot of the, you know, in a sense, I think both the kind of the optimists and the doomsayers are correct, where that could be amazing. If it gets handled very well, then it could be radical abundance for everyone. We could solve all the other problems in the world. If it gets handled badly, well, the course of that tech development could be dangerous pathogens. Or it could enable us to lose control to AI systems, or it could be involved like misuse by humans themselves. So I think, and I can go into that more, but I think there's just a lot of things going on that could be extremely important from the long term perspective. Well, let's go into it more. I mean, I as a as a simpleton assume that pretty much any new technology is going to be applied to porn and warfare first. And that those two will also sort of reciprocally drive forward a lot of new technology. I'm actually only 10% joking. But go ahead. Well, you know, dial E2. I do actually. Yes, I'm going to be using it a bunch this week. Oh, fantastic. Well, for listeners, you don't know, it's a failure of essence in AI system. And you can tell it to produce a certain image using text. So maybe that image is an astronaut riding a unicorn in space in the style of Andy Warhol. And it will create a near perfect rendition of that. And you can really say a lot of things. You can say, oh, I want a hybrid of a dolphin and a horse riding on a unicycle. And it will just create a picture of that. It's really like, in a way that really makes it seem like it understands the words you're telling it. And at the moment, it can create like non faces of imaginary people almost picture perfect. Again, if you pay close attention, you can see like weird details. So they kind of give it away. When you say imaginary people, what do you mean? So if you type in like a picture of Boris Johnson, then it will not give you a picture of Boris Johnson. And I don't know this for sure. But my strong guess is that because it's been deliberately restrained so that it does not do that. Right. Because well, so it doesn't deep fake everything. Exactly. Yeah. Because you were mentioning porn like with that technology, like you could well fill in the blanks. I'll let you think of your own text bumps you could put in involving, you know, Joe Biden or Tim Ferris or whoever you want. Joe Biden and Tim Ferris. Oh, exactly. Boris Johnson too. You're all in the fame. Who knew you were such good friends? Oh, God, the horror, the horror. Yeah. So to my knowledge, that's not been used for porn yet. But I think the technology would make it completely possible. And then is it going to be used for warfare like absolutely? I mean, there'll be a point in time when we can automate weaponry. So at the moment, part of the cost of going to war is that your people, your part of your population will die. That's also a check on dictatorial leaders as well. You need to at least keep the army on your side. Otherwise, there'll be a military coup. Now imagine if there's a world where the army is entirely automated. Well, dictators can be much more reassured because their their army can be entirely loyal to them. It's just coded in. Yeah. Also, the costs are going to war much lower as well, because you're no longer sustaining casualties on your own side. And so that's just one way in which technological advances could be hugely disruptive via AI. And it's far from the biggest way. So let's take just a short intermission from Skynet and World War three, just for a second. We're going to come back to exploring some of those. But what are some actual long-termist projects today that you are excited about?


Most promising developments happening now. (01:05:41)

So one that I'm actually excited about is investment in and development of a technology called far UVC lighting. Far UVC is just a very specific and quite narrow spectrum of light. And with sufficient intensity, just put into light bulbs, it seems like that just stabilizes over. Now we don't, we're not confident in this yet. We need more research on its efficacy and the safety. But if this was just installed in all lighting in every house around the world, basically in the same way that we do for fire regulation, like every house at least in a relatively well-off country has to meet certain standards for fire safety. It could also have to meet certain standards for disease safety, like having light bulbs with UVC light as part of them. Then we could would make a very substantial progress to never having a pandemic again, as well as as a bonus eradicating all of us spiritually disease. And so this is like some extremely exciting technology. There's a foundation that I've been spending a lot of time helping to set up over the last six months called future fund. And this is something that like, yeah, we're donating to and investing in, because yeah, it just could make an absolute transformative difference. So that's one other things that are very concrete within the biotech space include kind of early detection of new pathogens. So just constantly sampling wastewater or constantly testing healthcare workers and doing like full spectrum diagnostics of like, just all the DNA in the sample excluding human DNA. Like, is there anything there that we just looks like a pathogen and what we don't understand so that we can kind of react to new pandemics very quickly? Also more boilingly, just like better PPE, where you could just have, you know, you put on your super PPE herd and you're now just completely protected from any sorts of pathogens that could enable society to continue even if there was an out-the-equivable, really bad pandemic. So that's very exciting within biotech. Within AI, there's a lot of work on technical AI safety, where the idea is just using methods to ensure that AI systems do what we want them to do, where, you know, that means even if they're very powerful, not planning to seek power and disempower other people who created them, not being deceptive, not causing harm, and there are various things you can do there, including with these kind of, you know, not the sophisticated models that we're currently using, like tests to see if they are acting deceptively what are structures you can use to make them not act deceptively. Can we have better interpretability so that we actually understand what the hell is going on with these AI models? Because at the moment, they're very non-spentent, and we really don't know, like, what's, how do they get to a particular answer? It's just this huge, like, huge computational process where we've like trained it via learning over what in computer time is like extremely long time. So maybe it's like tens of thousands or even millions of games of Go that's played, and now it's like very good at Go, but like, what's the reasoning that's going on? We don't really know. We could keep going as well. There's just like many, many things within technical AI safety. And then there's the governance side of things, both for AI for other technologies, for the juicing of the risk of World War III. Here, I kind of admit it gets tough. It's like very hard to measure and be confident that we're doing stuff that's like actively good, and we have to hope a little bit more that just having smart, thoughtful, competent people in positions of, you know, political influence where they're able to like understand the arguments on both sides and put in policies and regulation in place, such that like we more carefully navigate these big technological advances or such that we don't go to war or face some sort of race dynamic between different countries. That is also just like actually important to me in my view. Okay, let me take a pause to jump back to never the questions I have next to me. When you feel overwhelmed or unfocused, or if you feel you've lost your focus temporarily, what do you do?


How Will refocuses during periods of overwhelm. (01:10:09)

Yeah, other questions you ask yourself, activities, because I think it is easy, maybe I'll speak for myself to feel like there's so much potentially coming down the pike that is an existential threat. It's easiest just to curl up into the fetal position and just scroll through TikTok or Instagram and pretend like it isn't coming. Not saying that is where you end up, but when you feel overwhelmed or unfocused or under focused, what do you do? For me, it's most often driven by, you know, dropping my mood. Yeah. So I've had issues with depression since forever, basically, although now it's just far, far better. I think I normally say I'm something like five to 10 times happier than I was a decade ago. It's a bit good. That is good. Don't be happy about that. Yeah. And so I have a bit of, have you heard of the term of a trigger action plan? Say that one more time. I'm not sure if this is the word or the Scottish accent. I know it's trigger action plan. Oh my god, I've shracked on the podcast. This is amazing. All right. Yeah. Trigger action plan. Go ahead. Exactly. So no, I don't know what that is. The idea is there's a trigger like some event that you hit that happens and it's just like when that event happens, you immediately just put into place some action. So a firewall arm goes off. Then it's like everyone knows what to do. There's the fire drill. You follow the fire drill, you like stand up, you walk outside, you leave your belongings. And it's like, it's so that you don't have to think in complex situations. I do that. But for when I have like low mood, the thing that has been very bad in the past is when it's like, oh, I've got low mood. So I'm not being as effective and productive. So I'm going to have to work even harder. And therefore, I beat myself up and it makes it even worse. So instead, what I do is I'm just like, if I notice my mood is really slumping and therefore it's hard harder to work, I just bump like fix my mood to the top of my to-do list becomes the most important priority where the crucial thing is not to let it spiral. The number one thing I do is just I go to the gym or go for the run. Because then I'm like, look, I wanted to do this a certain amount of time a week anyway. It's something I enjoy a friend like a recreation. So worst case, I'm just moving time around. Similarly, I'll probably meditate as well. Then at the same time, in terms of how I think, I have again certain kind of cashed thoughts that I found very helpful. So one is just thinking about how it's like, yeah, this happens before, and it's not the end of the world. It's being okay. Like if I've, you know, gone through this before, it's just I'll be able to get through it again, probably. Second is just thinking about no longer assessing like the individual day that I'm having, but instead some like larger chunk of time, where it's easy to beat yourself up if you like, look, I've had just the shittest day and I've done nothing. What a loser. Whereas if you like, okay, well, how have I done over the last three years or 10 years or like my whole life? And at least assuming you feel kind of okay about that, which I do, then it's that's very reassuring. It's like, okay, I've had a shit day, but like, if someone were to like a history of my last few years, they probably wouldn't talk about this day. They talk about the other things. And so it's kind of like, I've got a little bit in the bank there. So even if I just take the whole day off, like in the grand scheme of things, it doesn't really matter. And so some of these thoughts then, you know, that combined with like taking a bit of time away from whatever's making me plunge. And then like exercise I just find has a mood boost as well. So also get, but then also gives time for these thoughts to really percolate and sink in. Generally means that I can just then like come back a couple of hours later and be a bit of a fast. But the key thing is just like, once this happens, you just, you just do the thing and you stop thinking. It's like, look, this is what my plan is. Can you say trigger action plan one more time and a heavy Scottish accent? Trigger action plan. That's what you need power. Gonna put that right at the top. Oh, so good. So good. Thank you. Thank you. Nate, bother pal. You know, if this long term is an effective altruism, philosophy thing doesn't work out for you. I think you have a future in voice acting. So you always have that. Effective altruism is about doing the most country can. Not being a weak old blows. I don't know why if you speak in Scottish accent, suddenly you've got to be somewhat aggressive and like insulting someone. Otherwise it doesn't quite work. You can't whisper an aggressive Scottish accent. It's very hard. Very, very challenging. I'm not even going to try that. It would be embarrassing. But let's hop back into AI for a moment. So you hang out with a lot of the smart, cool kids and very technical people who really understand this stuff.


Perils of AI considered plausible by the people who create it. (01:15:02)

When they talk about robots gone bad or just the plausible scenarios that would be very bad, what are they? Like what are the two or three things that they would see as an event or a development that would be the equivalent of the trigger action plan, right? Where it's like, oh, this is life before in life after. What do they say two or three or one to three scenarios that they've honed in on? From my perspective, two extremely worrying scenarios. One is that AI systems get just much more powerful than human systems and they have goals that are misaligned with human goals and they realize that human beings are standing away of them achieving their goals and so they take control. And perhaps that means they just like wipe everyone out. Perhaps they don't even need to. So an analogy is often given between like the rise of Homo sapiens from the perspective of the chimpanzees where Homo sapiens were just smarter. They were able to work together. They just had these advantages. And that just means the chimpanzees just have very little say in how things go over the long term. Basically no say. It's not that we made them extinct. Although in a sense, they're kind of lucky. We made many. In fact, I think we made most of large animals extinct due to the rise of Homo sapiens. But that could happen with AI as well. We could be to the AI systems, what chimpanzees are to humans. Or perhaps it's actually more extreme because once you've got AI systems that are smarter than you and they're building AI systems that are smarter again, maybe it's more like we're like ants looking at humans when we're looking at advanced AI systems. So give me the second one and then I'm going to come back to the first one with just a sci-fi thought experiment. For sure. And then the second one is like, okay, even assume that we do manage to align AI systems with human goals. So we can really get them to do whatever they want. Nonetheless, this could be a very scary thing. Where if you do think the AI systems could lead to much faster rates of technological progress, in particular by automating technological discovery, including the creation of better AI systems. So we've got AI writing the code that builds the next generation of AI that then writes even better codes to build the next generation of AI. Things could happen very quickly. Well, even if you manage to have AI systems do exactly what you want them to do, well, that could concentrate power in a very small number of hands. Could be a single country, could be a company, could be like an individual within a single country who wants to instill a dictatorship. And then once you've got that power, it's kind of similar to what happened during the industrial revolution and earlier. So you've got more and more powerful technology over that period. And what did it do? It used it to colonize and subjugate a very large fraction of the world. In the same way, it could happen, but even faster, that a small group gets such power and uses it to essentially take over the world. And then once it's in power, well, once you've got AI systems, I think you're able to have indefinite social control in a way that's like very worrying. And this is value lock-in again, where at the limit, imagine you're kind of the dictator of a totalitarian state, like 1984, the Handmaid's Tale or something. And that that's a global totalitarian state. And you really want your ideology to persist forever. Well, you can pass that ideology on to this AI successor. It just says, "Yep, you got all the world now." And the AI has no need to die. It's like software. It can replicate itself indefinitely. So unlike dictators who will die off eventually, causing a certain amount of change to occur, well, this is not true for the AI, it could replicate itself indefinitely. And it could be like in every area of society. And so then when you've got that, it's like, why would we expect moral change after that point and it's like kind of hard to see? So in general, I think there can be these states where you get into a particular state of the world and you just kind of can't get out of it again. And this kind of Orwellian perpetual totalitarianism is actually one of the things I really worry about. Okay. First question. Again, this is a happy book. It's an apopecan. Yeah, yeah, yeah. So yes, so within the context of a discussion of the happy book, you talked about, I think it was the mohists. I can't remember the term you used, but you mentioned they were similar to effective altruism and they formed a paramilitaric group. When are you forming the effective altruism paramilitaric group, counter AI insurgency squad? Is that in the works? Well, there is an analogy between, we haven't yet got our own army. Probably that won't happen. I think things are going pretty weird if they have and I might need to intervene at that point. But there is an analogy where they built very powerful and created very good defensive technology. So you got Chvebuche is very powerful for attacking. You say Trebuche? Trebuche is? Oh, Chvebuche. Yeah, it's like a catapult. What is that? It's got a sling on it? Is that what it is? It's like a catapult with a sling. So you get the, it's like an adult adult, but for throwing much bigger things. Anyway, the physics involved here, I think the same. Yeah. For sure. But also walls are defensive technology. If you had just really good walls, really good defenses, then... But you said wolves for a second. I was like, wow, did not see that coming. We're going to have to resurrect the wolf if we want to have any hope of defensive wolf technology. All right, walls. Yeah, good tea. They are training eagles to attack drones. So it's not like, it's not so insane. All right, wolves. We have wolves to attack the robot overlords. I don't back the wolves, I've got to say. But we can think in the same terms of, look, there's certain technology that has kind of offensive advantage. So the ability to design new pathogens, there's certain technology that's like has a defensive advantage like this far UVC radiation. And so one of the things we're doing is really trying to develop and speed up kind of defensive technology. And so similarly, when you look at AI, there's some things that just pure capabilities. It's just AI getting more and more powerful. And then there are some things that are helpful in making sure that AI is safe, like understanding what's under the herd of these models. Just means that like, okay, we know what's going on a bit better. We can like use it better. We can predict what sort of behavior that we'll have. Okay, so let's talk about defensive capabilities. I'll just give another example of an asymmetric offense defense situation, which would be drone warfare, right? So the ability to create weaponized, potentially lethal drones and swarms and so on is much lower than the cost to defend against them, generally speaking, right? I mean, certainly that becomes true. If you start to combine, say, targeted bio weapons with drones, things get really expensive at best to defend against. But let's talk about my sci-fi scenario. So when you use the analogies of humans and chimps, or the analogy of the industrial revolution and the technological gains in western Europe, predominantly, which then allowed the physical, and that's the word all underscore sort of subjugation and colonization and dominance of a significant percentage of the world's population. I suppose there's part of me that on a simplistic level, it feels like even though this would not be easy to do, because it would be sort of like a homicide suicide for a lot of folks, the more interdependent we become, but it's like, all right, if AI is constrained to a physical infrastructure that is dependent upon power, would not part of the defensive planning or preemptive planning go into trying to restrict AI to something that can be unplugged to put it really simply. But how are people playing out this hypothetical scenario, right? So with the AI, presumably, if it's as smart or smarter than we are, foresee this and then develop sort of solar powered extensions of itself, so it can do A, B, and C. I mean, how are people, I'm sure this is part of the conversation, I've just never had it. So what are smarter people exploring with respect to this type of stuff? Yeah, your nose is pointed in a good direction on this one. And it's this sort of thing that makes me among my peers on the more optimistic end of thinking that advanced AI would not kill everybody, where, yeah, you could have like air-gapped computers, so you can't access the internet now that don't have other ways of kind of controlling the world apart from kind of text output. And they've been trained to act as kind of articles, so they just use ask them a question and they give you ideally like very justified kind of true answer. And perhaps you have many of these as well, you've got like a dozen of them, and they don't know that the others exist. And then you just start asking them for help. So you're like, okay, we're going to start building these like incredibly powerful AI systems that much, much smarter than we're generally able than we are. Like, what should we do? What should our plan be? And so that's a pathway where it's like, you're using AI to help solve the problems that we will face with even more powerful AI. And what's the response? I mean, some people would say, oh, well, if the AI systems are really that powerful, they're like far better than humans, they will then trick you. So and they'll be able to do that just by outputting text and telling you like, oh, do this thing or do this thing. And that will all be this like long deceptive play. And I just think that's unlikely. That seems to be speculative for me. I don't have like so long reasons to think that we would have that. The current AI systems we have are more like, you know, they just output text. It's not like they're an agent that's like trying to do things in the world. You just put in the text input, it gives you a text output, at least for language models. And potentially, we can scale that up to the point where they're like these kind of sages and boxes. And I think like, so I think that's a significant pathway by which like we make even more powerful AI systems that are kind of agentic are kind of acting, you know, have a model of the world and it's going to do things in the world that we make them safe. But that's exactly a good example of yeah, again, kind of deferential technological progress where an AI system that's just like this article in a box, you know, separated from the rest of the world, that seems very good from a defensive perspective, whereas an AI system that's been trained on like war games, and then it's just like the least into the open seems like potentially very bad. Oh, yes, indeed, robots gone wild. All right. Are you going to create new subnet? I don't know if I'll create a subreddit. I mean, I'm tempted to start digging spider holes in my backyard and learning how to hunt with bone arrow. But you know, all things in due time. What are the most important actions people can take now? What are some actions for people who are like, I want to be an agent of change, I want to feel some locus of, if not control, at least free will, I don't want to just lay on my back and wait for the cyborg raptors to descend upon me, or to become a house pet for some dictator in a foreign land who overtakes the US as a super or whatever it is, right?


Longtermist-minded resources and actions we can take now. (01:27:24)

I want to actually do something. What are some of the most important or maybe impactful actions that people can take? So I think there's kind of two pathways for such a person. One, you might be motivated to help, but you don't really want to reject your whole life. Ideally, perhaps you just don't really want to have to think about this again, but you do want to be doing good. Then I think the option of donations are just particularly good. So you could take the give them what we can pledge, you make a 10% donation every year of your income, and then you can give to somewhere like the long-term future fund as part of EA funds, and it'll get redistributed to what some kind of domain experts think are the highest impact things to do in this space. That's the kind of like baseline response or something. And I think it's important to emphasize you can do an enormous amount of good there. You know, there's a lot of ways we could spend money to do good things in the world, including from this long-term perspective. The second is if you're like, no, actually, I want to be more actively a kind of agent of change. Then I think the first thing to do is to learn more. I've tried to pack as much as I can into a book, but I think there's a lot to engage with. I mean, the book is what we are the future. It's talking about some big philosophical ideas. It's also just covering a lot of broad ground of different disciplines, different issues. We talked about AI and technology and iOS and World War 3. There's plenty of other issues I didn't talk about. We haven't talked about nukes. We haven't talked about technological stagnation, which I think is particularly important as well. We also haven't talked even just about promoting better values as well, kind of all more, but odd ways of making the long-term better. So all of these things are things that I think we can learn. And therefore, I had encouraged leading The Presspus by Toby Ord, which I mentioned in my recommended books. Also, 80,000 hours.org as well has enormous amounts of content. Openflantheb.org. Also has just a lot of really interesting content. They're a foundation, but they've done some really deep research into some of these topics, such as this issue, which we didn't get a touch of when we should expect human-level intelligence to arise with some arguments that, you know, we should really put a lot of probability mass, like maybe more than 50% on it coming in the next few decades. And then following that, I think the most important single decision is, how can you either use or leverage your career or switch career in order to work on some of these, you know, most important issues? And again, we've really tried to make this as easy as possible by providing like endless online advice and also like one-on-one coaching, such that people can, yeah, get advice. And then the final thing would be getting involved with the effect of autism community, where this stuff is hard. It can be intimidating. One of the big things that just is a fact when we start thinking about these more kind of civilizational scale issues compared to the kind of original seed of EA, which is, you know, funding these very well-evidenced programs that the multiple-ly-improve health is like it can be very overwhelming, and it can be hard to know exactly how to fit in. But we now have a community of thousands or tens of thousands of people who are working together and really keen to help each other. And there are many conferences like EA Global Conferences, places like London, D.C., San Francisco, all kind of independently organized conferences, EA Global Exes. In many places, there'll be one in India, for example, in early January, as well as like hundreds of local groups around the world where people get together and, you know, can often like provide support and help each other to try and figure out like, okay, what is the most impactful thing you can do? So yeah, that would be my kind of laundry list of advice. With respect to say ditching the news, or at least going on a lower information diet with the most manufactured urgency that we get flooded with, instead spending time looking at big picture trends, or trying to get that big picture roughly right, as you put it, both from a historical perspective and a current perspective. Would you still recommend, for podcasts, in our time hosted by Melvin Bragg, I believe discusses history, philosophy, science, with leading academics, and the 80,000 hours podcast, would those be two you would still recommend? Yeah, I would still strongly recommend them. There's also another podcast here, this idea by Finn Morehouse and Luca Rogatti. I also particularly like, rationally speaking, by Julia Gaylef as well. It's like fairly good. And then in terms of websites, if you want like big picture, like beyond the websites, I've already said, to just have like the best big picture understanding of the world, I don't know of a single better source than our world and data, which is just, I mean, it was very influential during the COVID pandemic, but it has, you know, if you want to learn about nuclear war, or long-run economic growth, or world population, it's articles that are presenting both data and like the best understanding of the data in just this like timeless evergreen way, with exceptional rigor and exceptional depth. It's just, it's just amazing. I use it very heavily to like, orient myself for the book. So Will McCaskill, people can find you on Twitter at will, McCaskill, M-A-C-A-S-K-I-L-L on the web, william McCaskill.com.


Conclusion

Parting thoughts. (01:33:22)

The new book is What We Ove the Future. I recommend people check it out. Is there anything else you would like to add any requests to the audience? Anything you'd like to point people to, any complaints or grievances with this podcast process that you would like to air publicly, anything at all that you'd like to add before we wrap this conversation up? The main thing to say is just like, as we've said over and over again, I think like we face truly enormous challenges in our life. Many of these challenges are very scary. They can be overwhelming. They can be intimidating. But I really believe that each of us individually can make an enormous difference to these problems. We really can significantly help as part of a wider community to putting humanity onto a better path. And if we do, then the future really could be long and absolutely flourishing. And your great, great, grandkids will thank you. Well, thank you very much. Well, I always enjoy our conversations and I appreciate the time. Ditto, thanks so much, Tim. Absolutely. And everybody listening, I will link to all the resources and the books and websites and so on in the show notes as per usual at Tim.blog/podcast. And until next time, be just a little bit kinder than necessary. And thanks for tuning in.


Great! You’ve successfully signed up.

Welcome back! You've successfully signed in.

You've successfully subscribed to Wisdom In a Nutshell.

Success! Check your email for magic link to sign-in.

Success! Your billing info has been updated.

Your billing was not updated.