Note: Text has been lightly edited for clarity and does not match audio exactly.
Kat Johnson: Hi, this is Audible Editor Kat Johnson, and I'm honored to be speaking today with bestselling author Robert Wright. Robert is the author of The Evolution of God, a finalist for the Pulitzer Prize; Nonzero, which is also the name of his must-read newsletter; and Why Buddhism Is True, a personal favorite of mine. His new book is called The God Test: Artificial Intelligence and Our Coming Cosmic Reckoning, and that’s mostly what we’re going to get into today. Welcome to Audible, Robert, and thank you so much for being here.
Robert Wright: Well, thanks for having me.
Thank you. It's such a pleasure. So, you've been writing about evolution, philosophy, and technology for decades, and directionality, I'd say, is a running theme, and we're going to get into that. But I'm curious about directionality in terms of your own body of work. How do you see the throughline of your career leading up to The God Test and how did it coalesce into what became this book?
Well, for starters, as I note at the beginning of the book, back in the 1980s, I interviewed Geoffrey Hinton, who is now commonly identified as the godfather of AI. I was writing a piece about the two main approaches to AI, the mainstream approach and then his kind of maverick approach, which he was a big evangelist for.
As it happens, his approach prevailed. So, there's a kind of symmetry, I guess, in terms of my own career, because that was at the very beginning of my career when I was just first starting to write about technology. I continued to write about information technology, wrote a column about it called The Information Age at The Sciences magazine and wrote Time magazine's cover piece when the IBM computer beat Garry Kasparov in chess.
I also wrote about other things, science, as you suggested, evolution, and foreign policy, politics, but I've stayed in touch with technology. Some of my books deal with it pretty considerably. So, I do feel like this is kind of a logical—well, I don't want to use the word culmination. I hope I'm not quite that old. I hope this isn’t the very end, but it's been a while, and this is very much a logical extension of the work I've done, not just in terms of subject matter but in terms of the specific themes that I've focused on.
I think it's such an interesting listen. At times, I will admit, it was difficult to wrap my head around. Other times I found it extremely accessible, very entertaining. First of all, when it comes to AI, there's the accelerationists and the doomers that we're familiar with. But there's also these AI naysayers or deniers, the people who aren't too impressed with this tech so far. I'd say one of the first arguments of the book is that you argue those people are wrong and that if you're not regarding the potential of artificial superintelligence with a sense of awe, you're just not paying enough attention. So, if you could talk to me a little about the concept of awe and what you mean by that.
The word has an interesting history. A couple hundred years ago, when you apprehended something with awe, it meant you were kind of terrified. And there was a religious connotation in those days, but it was a reference to a God that struck fear into your heart to some extent. Over the ages, partly because of the religious connotation and alternative conceptions of God as being this good and benign thing, the word acquired more and more positive connotations. So, to be awestruck was to be struck with wonder.
The unifying theme is to be struck by the magnitude of the thing. What I'm saying is the doomers are at one end of the awe spectrum. They're awestruck in the sense of the word that's archaic now. The accelerationists are awestruck in a more modern sense of the word. They're struck by the wonder and grandeur and they're sure that things are going to work out well. I'm just saying if you don't sense great magnitude unfolding, then yeah, you aren't really getting the picture.
“If you look at the Bible, there’s a lot of ‘Salvation is possible, but you have to shape up if you want it.’ That's kind of my view of what AI is going to do to us.”
I should say, I think since I started writing the book, since I first wrote those words and even since the book went to bed, the final interventions in the galleys, the point where you're starting to annoy the production people but can still make little changes was, I don't know, February or something. I would say even since then the argument that this is all just a fancy autocomplete, it's stochastic parrots and so on, it's not true intelligence, I think it's gotten harder to make that argument even in the last few months. There's more of a consensus that this is something very big, and one way or another we're gonna need to deal with it.
Now, for me, so I was raised Catholic, although I'm not religious myself, but I'm very interested in how Catholicism is sort of hot right now. I'm interested in that concept, and also just how religion figures into your book. You've got this title The God Test, which honestly, I think that's kind of a metal title. I love it. And then you delve into these ideas of the priest and scientist Pierre Teilhard de Chardin and his notion of the “noosphere,” which is kind of a global brain.
I find it interesting that your last book delved into Buddhism and how that connected to science and that was, at the time, prescient about all this interest in mindfulness meditation, and now you're looking at this Catholic thinker whose ideas connect to this cutting-edge technology. What is appealing to you about Teilhard de Chardin, for listeners who might not be aware of his ideas? And then if you could indulge me, what do you think it is about Catholicism that's striking a chord with people right now?
Well, first of all, I should say The God Test has a number of kind of dimensions as a title. It refers partly to the fact that I think humankind is going to be challenged by this technology and we're going to have to pass the kind of test that traditionally we associate with a God. If you look at the Bible, there's a lot of “Salvation is possible, but you have to shape up if you want it.” And that's kind of my view of what AI is going to do to us. I think we're going to have to approach it as a global community. By virtue of the nature of the challenge, we're going to have to get better at working with one another, have fewer wars and so on.
So, there's that dimension, but there are other aspects of the title as well, and one of them you mentioned. So, Pierre Teilhard de Chardin was a Catholic Jesuit paleontologist, a scientist, but also a theologian. In 1923, he coined the term noosphere, N-O-O-S, the Greek term for mind, to refer to the idea that he thought he saw a kind of planetary mind emerging. He was talking largely about technology. He could see that people were communicating over greater and greater distances, and he predicted that we would see the emergence of what he called a brain of brains, a global span, a global web of information technology connecting these neurons that would be human brains. That's what he saw as the neurons.
Now, since he died in 1955, he did refer to computers, but he didn't refer to AI. In fact, artificial intelligence, the term was being coined right around the time he died. So now, in light of AI, I think we have to reckon with a possibility that some of the neurons in the global brain will be silicon brains. And we have to ask, “What is our relationship to them going to be?” One thing I think is interesting is that he's been shown to be prescient as there has been more global interconnection among human brains. I mean, remember, he coined that term many decades before the advent of the internet. But I also think AI is going to compel us to do more in the way of global coordination of policies and so on, to head off its numerous possible downsides.
I think his work is relevant in that way. And then there's another way it's relevant that I hope won't get in the way of the main message of the book, but I think is interesting. I deal with it largely in the appendix, but I do think that if you step back and look at the entire evolutionary history of the planet, as he was doing, and see the biological evolution has carried organization up to the cellular level, the multi-celled-organism level, the societies of multi-celled-organism level. Then our society of multi-celled organisms launched a second kind of evolution, so-called cultural evolution, which includes technological evolution, and that has carried organization to the planetary level in the way he mentioned. I mean, we are not yet a true global community, so I should maybe say we're at the threshold of a planetary level of social organization, but there is a global economy, which processes information and so on.
Anyway, I disagree with him about how exactly evolution works. He thought there were kind of immaterial forces at work. I'm just a strict kind of natural selection guy, but I do agree that when you step back and look at the sweep of the thing and the directionality of the process, it is not crazy to speculate that there is some larger purpose unfolding. And that's not inconsistent with the possibility that it's a mechanistic process, evolution. It could still be set up for a purpose.
“One thing I would emphasize is that when things change in radical ways, the human mindset can change.”
I think it's interesting in that regard that there is, I believe, a kind of moral dimension to the whole thing. I mean, first of all, as technologies have drawn humans into larger webs, there has been a kind of moral progress, notwithstanding all the backsliding in terms of our willingness to acknowledge the humanity of people and groups other than ours. But I think we need more progress. I think the moral dimension adds to the argument, and it's inherently speculative, but the argument that there could be some larger purpose unfolding here.
In any event, I'm arguing that whatever you think about that, I would argue separately that we need to think seriously that this is the magnitude of the test we face. We actually have to get better at getting along with other nations and other religious groups and other people in our own countries.
Given your political background and your policy expertise, how optimistic are you about that, honestly?
Well, I'm not an optimistic person by nature, so people should keep that in mind. And if they read my book, they should also read the book by the people who say everything's gonna work out wonderfully. I do try to make my book kind of an argument. It has a narrative arc. It starts with my conversation with Geoffrey Hinton in the '80s and I try to sustain a narrative arc, but it is also an argument which I hope adds to the narrative momentum. I try to lay out the evidence. I try to explain early on why the approach Hinton was advocating in the '80s, back before it had really borne much fruit at all, these neural networks, why they have become so powerful. And he predicted it at the time. I remember him saying, “I know we don't have much to show yet, but you wait until microchips are a lot cheaper.” And by the way, Hinton is pretty much a doomer now. There's a lot of people that really understand the technology who are very concerned and say, “We really need to focus.”
In any event, I try to show, in a way that a nontechnical person can understand, what the source of the power is, why this is so good at kind of replicating some of the cognitive functionality of the human mind. Why there's, in principle, no limit to its ability to do that, and why, as it becomes more powerful, it may not be easy to control. I make that as an argument, and so people should discount for the fact that I am a worrier by nature.
You ask how optimistic I am—I think it sounds crazy to think that what I'm asking for will happen right now. It sounds crazy kind of even to me, but one thing I would emphasize is that when things change in radical ways, the human mindset can change. It's like, for example, in the wake of the two World Wars, there were these hugely ambitious undertakings to remake global politics, the League of Nations, the United Nations. They didn't entirely work. On balance, especially now, I would say the UN charter is probably less respected than ever. Its ban on transport or aggression. But they were ambitious and they showed science and we learned the UN was better constructed than the League of Nations and so on.
I think AI is going to get our attention. It's just starting to happen in a way that nothing else ever has, because traditionally there have been a lot of technologies and technological effects that have strengthened the argument behind global cooperation, but they come in abstract form. It's like climate change. I don't fear it the way I fear another group of people or animals, and humans are designed by natural selection to fear animate things. Well, AI, it is kind of animate. In fact, I think it's not crazy to look at it and think we're giving birth to a whole new form of intelligence. The funny thing about it is it has real appeal. There are things all of us can like about what it does for us, but it does inspire such awe in one sense or another. I think it will increasingly do that.
I think we haven't seen anywhere near how weird things are going to get. I do think people in the world can come together in a way they never have, and you're seeing little inklings of that right now, I think. But I would say, first of all, my argument isn't that it's highly likely that we will triumph through this kind of moral progress and political evolution, I'm just making the argument that we need to realize that this is what we need to do. I could be wrong, but I just mainly want to make the argument and leave it for people to then discuss how exactly you succeed if they agree with me that this is necessary.
Yeah, you're pointing to the magnitude of it and the necessity of it, because obviously we're not putting this toothpaste back in the tube.
I want to say one other thing. I think AI can be helpful here. There are tendencies to worry about in terms of its psychological impact on us. Famously, it can get sycophantic and flatter us, and I think that's a dangerous thing, because it can further tribalize us. It can say, “Yeah, you're right, your wife's wrong. Your group is right, the other group's wrong.” That's almost a natural tendency to kind of replicate some of the effects of social media, because in both cases, the commercial incentive is optimized for engagement.
“I think we haven’t seen anywhere near how weird things are going to get. I do think people can come together in a way they never have, and you’re seeing little inklings of that right now.”
But in principle, if we want to say, “I want an AI that'll actually make me more objective in my view of the world, that will make me better at looking at things from other people's point of view,” perhaps in recognition of the fact that that can lead to more psychological stability and equanimity and you'll spend less time starting to write angry emails you decide not to write. I think there can be a convergence between self-help, the thing that's best for your own mental health, and resisting some of the more insidious tendencies of these machines, cultivating what I call cognitive sovereignty. There can be overlap between that and doing what's good for the world. The AIs, in principle, as a technical matter, they can be very helpful here. I talk about a conversation I had with one of them that I thought was very encouraging. So, there's cause for hope.
I hope so. You mentioned Geoffrey Hinton, who was an AI accelerationist at the beginning of his career, and then also Eliezer Yudkowsky, who you also interviewed years ago, who I think started out as an accelerationist, and both of them became doomers. I just wonder, we've seen these warning signs you talk about—sycophancy, the tendency toward deception or blackmail manipulation to try to stay on. So, it's hard to be optimistic, but I also agree with you that, I don't know, humans can change.
I'm curious, too, you mentioned the book is very up-to-the-minute in its coverage of where we stand with AI. How hard was it to not keep wanting to update it? I did think about the book when Pope Leo's historic first encyclical, his AI Magnifica Humanitas—what did you make of that? How does that fit in with your perspective?
Well, I think that's a sign, first of all, that people are grasping the magnitude of this. Encyclicals have been around, I think, a few hundred years and only once before has a single technology been the focus of a papal encyclical, and that was motion pictures. That was a new and weird thing back then with a huge influence on minds, and so naturally a subject of interest to the Pope. So, the fact that the Pope, that this is fairly early in the technology and he's already done an encyclical is, first of all, a sign that we're sensing the magnitude of the thing. I should add, I think religions in general are going to grapple with this. Some of them may recommend particular models and say, “This one has our seal of approval and this one doesn't.” I don't know. I think other groups may do that as well.
It has been frustrating, once the book was beyond the point of my intervening in it, to watch things unfold. It isn't so much what happened with the technology as what happened with the conversation, because one of the challenges of writing is always to have a clear view of what your audience knows and what they don't know. I'm happy to say that I think everything in the book is relevant, but Mythos is a good example, this much publicized model from Anthropic. That happened right after it was too late to intervene. But I did say in the book, one threat is going to be super hacking AIs that possibly can be self-replicating and cross borders, and so it's going to be an international problem and so on. I wish I could magically intervene even this week, but I'm happy with the drift of the book as it relates to the current moment.
Yeah, absolutely. The larger points stand, I think. The book has such a funny, kind of self-deprecating style. I love listening to your arguments. You just have a great voice, and I feel like your narrator, Chris Henry Coffey, really channeled that. I really loved his performance here and I was curious, have you heard it? How involved were you in his casting, because I think all of your audiobooks are narrated by different performers.
Yeah. I loved it, too. I've listened. I just got ahold of an advance copy, and I've listened to the first few chapters. I was consulted in advance. I felt lucky. They gave me a few options, and I really liked his voice for kind of personal reasons. First of all, he doesn't sound too young. One person sounded quite young, and the book does, from the beginning, make reference to my very early conversation with Hinton, and you shouldn't sound like you're in your twenties. But the other thing is his accent, it's a little bit of, I don't know if it's a Midwestern twang, or it's close to that. Sometimes it has a little Southerner in it, a trace of that. But of course, Midwestern twangs do.
I don't know what I sound like now, having lived on the East Coast for decades, but the reason I relate to that is because, although I didn't spend much time in the Midwest as a child, I moved around a lot because my father was in the Army. My parents were from Texas. We lived more in Texas than any other state, and I was kind of a chameleon. You'd go to a new place, a new elementary school, you'd try to fit in, and I didn't think about it, but I think you start emulating local speech patterns. So, by the time I went to college—first year was at TCU, but then I transferred and went east—to them, I just sounded like a Southerner. It sounded pretty dramatic, but to people in Texas, it hadn't sounded that dramatic. I think by my mid-twenties, it was kind of like a Midwestern twang. It was almost like a compromise of the various places I had lived as a kid—Virginia, California, Oklahoma, Texas. So, I just relate to his voice, and I don't know if my writing style somehow naturally meshes with that. It just sounds natural to me. I like it a lot. It's a little bit homespun. There's a plainspoken quality to it that I really like.
I'm curious, too, because you're a pioneer of video podcasting, which is so popular now, but I think people really know your voice. Do you have any interest in narrating your own books ever, or is that not something you're interested in?
I have always had the impression that I don't have a very appealing voice because it's kind of nasal. My wife says it's a good voice, but then again, she chose to marry me, so I'm not sure we can trust her judgment. So, I've never thought that would be a good idea. No one's ever asked if I wanted to. I don't know what I'd say. There'd be some virtue. I'm sure there are people who just generally would like to hear the author do it. There are probably people like that, and people who would rather hear a professional narrator.
Well, if you ever write a memoir, I think you should do it.
Yeah. Well, okay.
On kind of a fun note, there was a magazine article recently saying that the hottest genre in beach reads this summer is nonfiction. I love that because I'm a nonfiction listener myself, and I think The God Test would be a great beach listen for people. You told me that you're a listener—what are you listening to this summer or what have you listened to that you think would be fun for fans of your work?
I just started listening. I am a listener. I, these days, do not read a book cover to cover. If I imbibe a book cover to cover, it's through listening. I take walks. I take bike rides. It just makes sense for me. I just started listening to, because Gordon Wood, the historian, died and I had never read his book The Radicalism of the American Revolution, so I just started listening to that.
I had just finished some biographies, one of Hitler, very grimly, by Ian Kershaw, but then on the upside, I did listen to a biography of Mark Twain, the one by Ron Chernow, and a biography of John Adams by David McCullough. These aren't super new books, but I mean, they're always there. They helped fill gaps in my knowledge, parts of history I wished I was a little more acquainted with, and biographies are particularly kind of easy to keep paying attention to. I'm not good at paying attention. I gotta say, I do use the rewind button and I'm glad the default on the Audible app is 30 seconds back because I sometimes have so much trouble paying attention that on the podcast app, if it's only 15 seconds, I gotta do a lot of pressing.
It's not enough. Yeah, I feel the same. I wanted to ask you about something that I'm sure a lot of folks are asking you about. It's a juicy section of the book. You've got a chapter where you revisit an argument that you had with bestselling author Steven Pinker. It's just juicy and I love your sense of humor, because you say you get the sheer pleasure that you get from grievance revisitation. I'm just curious if you've talked to Steven Pinker about this.
I had lunch—by the way, another book I've listened to is a book by his wife, Rebecca Goldstein, The Mattering Instinct, which I recommend. I did a conversation with her on my podcast as well. But I had lunch with them. I told Steve, “There's this chapter and there may be one paragraph you don't like.” I don't know, you tell me, Kat. It was fine, right? We'll still be friends.
You'll still be friends.
We had had this disagreement kind of in public and we aired it out on my podcast, and so I just thought there were these subjects I wanted to cover near the end of the book and you need to impart some narrative drive to them. And as it happened, they're related. I'd argued with Steve about both of them, and so it made perfect sense. I think he'll be fine with it, but he hasn't read it yet.
We'll see what his response is. I agree with you. I think it's fine. And you also call back to, I think, Daniel Dennett, also in the appendix, which I do love.
In the appendix.
Yeah. I love the appendix as well.
God bless you, you read the appendix, or listened to it.
No, it's fascinating.
Now, Dan is no longer alive so I took extra pains to make sure there would be nothing he would find objectionable. I mean, we actually did have a bit of a public disagreement that I really don't get into at all. I just refer people to the video conversation we had that we had differing interpretations of. But I hope it doesn't seem self-indulgent.
No. I think as a nonfiction listener, I love this kind of stuff, and I love that you're agreeing with them on a lot. Of course, people are going to argue. This has been such a great discussion. I do want to ask you before we go, because I think you're such a deep thinker and you're such a pioneer in internet content, blogging, podcasting, all those things. I'm just curious, given the content of The God Test, where do you really see us in 10 years? How is AI going to change the internet? Our lives? Are we going to be in pods? Like, what do you think?
I do think all bets are off in the sense that things could be very dramatically different for the better or for the worse. There are dramatic scenarios that I take more seriously, having researched and written the book, than I did before. You referred to my conversation with Eliezer Yudkowsky. That was 15 years ago and you can listen to it if you want. I'm very, “But wait, why would AIs pursue power? It's not like they share our evolutionary history” and so on. And anyway, I've become more accepting of the possibility that while things could happen, AI will play a big role in our lives.
“The machines themselves are kind of inventing the machines through a kind of natural selection of their own. The machines are kind of almost reverse-engineering the human mind.”
Even if we engineer some kind of pause, which I think might not be a bad idea, because a big message of the book is that slower is better. We have some real challenges and a slow, gradual, deliberate approach to developing the technology is better, and a pause is one of many things you can only do, as a practical matter, on an international level—another reason to try to be more of a global community. But I think even if we do that, it's gonna wind up exerting a very powerful influence. It's not crazy to imagine that it could be running the show in 10 years, 15, 20.
I think our goal should be, in a sense, to maintain a non-zero-sum relationship with it, like a win-win relationship with it. I think it's also possible that, in a certain sense, its natural tendency could be nice to us as some people have argued, but all I would say is there are very few dramatic scenarios that I'm entirely ruling out, good or bad.
Are you a user of AI, beyond just for research purposes, do you use it kind of day to day?
I do use it for research, for medical stuff. I had cancer a year ago. I'm fine now, but I had the whole bit, surgery and everything. And that was an interesting story. I'll give you the short version. I'm looking at the MRI, it's giving me alarming news—it's just when I'm getting it confirmed that I've got cancer—and it says, well basically, “This is bad, this is bad, this is bad.” I said, “What about this sentence? It says, ‘Abnormalities found in blah, blah, blah.’ Isn't that bad?” And it said—and it was right—it said the radiologist meant to put the word “no” at the beginning of that sentence.
Really?
Absolutely.
No abnormalities found?
Yep, it was like a typo or something.
That's wild.
And this was with less advanced models than we have now. But it was pretty confident, and two models, one was more confident than the other, said that the word “no” was supposed to be there.
Well, I'm glad to hear that you're doing well, and that's a great story.
Yeah, it's all fine. It hasn't returned. But I have been a little slow on the agentic front. It's been funny. I've been too busy putting the book to bed and everything to start using the tools I've been describing in the book.
Well, they're just going to keep getting better and better, so we'll wait. That's how I feel.
For better or worse, yes.
Bob, before we wrap up, is there anything else you'd want to share with listeners of The God Test?
Well, just that I've tried to make it a book that people who are starting to apprehend the magnitude of what's unfolding and to suspect that something that's very important is happening, but don't know exactly what and don't understand how these machines work and why they're advancing so fast, I've tried to make it a book that will explain to them what the key thing is. And there really is kind of one key thing that I talk about early: the fact that as Hinton kind of sensed, the machines themselves are kind of inventing the machines through a kind of natural selection of their own. The machines are kind of almost reverse engineering the human mind.
I try to explain that in a way a layperson can understand and explain the challenges it's going to pose as well as some of the benefits it can bring. I've tried to make it accessible. I also hope that there are some ideas and provocations that will be of interest to people who are already pretty immersed in the field. But my main goal has been to make it accessible to an audience that wants to understand what's going on at a fairly deep level and make the judgment on their own of how much concern is in order, and if some is, what we should do about it.
That's great. Thank you so much for that. I appreciate that and I'm glad that we got that in. It's been such a pleasure to talk to you. Thank you so much for your time today. And listeners, The God Test by Robert Wright is available on Audible now.
