EP 8: The Accelerationists

27:28Reid, I know that you were one of the earliest investors in OpenAI. Yep. And that you were a board member there. You worked closely with their leadership, especially with Sam Altman. And recently, in some leaked emails, the public got the chance to see how some of the senior people at the company were talking about AGI all the way back in 2017, 2018. And I think a lot of us found it pretty dramatic. There was language in there about how if this technology were to fall into the wrong hands, it could lead to an AI dictator, about how superintelligence was going to reshape the world, reshape the balance of power.

28:11Essentially, about how the thing that the company was making, the company understood it as having this unimaginable amount of potential, but also an unimaginable amount of power. And I just want to know from you, is that an accurate reflection of how you and other people in Silicon Valley talk about AGI, think about AGI, talk about the future? And is that weird? Well, the first short answer is yes. I think there's a small – it's not Silicon Valley as a whole, but I think there's a group of Silicon Valley technologists who think this largely and broadly. And a few places, DeepMind, OpenAI, and others aggregate those people.

29:03So the short answer is yes, we were talking in that arena. And it kind of surprises people because you say, well, right now we have intelligence that's kind of the equivalent of a bumblebee. And it's like, and now you're talking about superintelligence. Like, how do you get there? And the answer is, well, because we understand things like possible compounding curve. We understand things about, like, well, what happens when you move from, you know, one GPU to 500 ,000 GPUs if you can use them in an appropriate way? And to most other people, it seems like, okay, that's nutty. And by the way, just to go back in history, similar discussions around the Manhattan Project.

29:39And I don't mean to indicate that there's a parallel in terms of danger or anything else, but these technologies have these massive impacts. And, like, that was another one. And, frankly, if you go back to the printing press, a lot of the discussion around the printing press was very similar to the kind of discussion we have around AI amongst the elites. So the, oh, my gosh, this is a moment in history is not like, oh, it's only these kind of crazy technologists. But it is a rare and small number of people kind of looking at the future, talking about possibilities and probabilities that engage in these conversations.

30:14And of course, I have enough self-reflection to know this is amazingly rare, right? Yeah. And it's something to be handled with intelligence and absolute focus in terms of how you navigate. And I think that was true for, you know, what is now not only the open AI folks, but also the anthropic folks and the Microsoft folks and the deep mind folks, you know, all kind of looking at this. All right. So help me square that with something else I've heard you say. I believe you said this to Tyler Cowen on his podcast, that people should be less worried about the robots coming for our jobs and more worried that AI advancements aren't coming fast enough.

30:57Which, by the way, great copy. But obviously, people are worried. They are worried that whole industries in our economy might go belly up. They are worried that AGI might lead to an economic collapse. Some people go much further, including some people in the AI field, and say that they're worried that AGI poses the greatest existential risk that humanity has ever faced. You've got the so-called AI doomers who think that we need to stop making AGI. You've got this other group that I call the AI scouts who think that this threat is so big that we all need to come together to ensure that the AGI that we're creating doesn't destroy humanity.

31:45So help me understand, what is it you mean when you say that we need to be more worried about it not coming fast enough? Well, let me break it into three issues. I'll start with the biggest risk, which is so-called existential risk, humanity-level risk, et cetera. Then go to society and jobs and then go to looking at how to make human progress. Now, on existential risk, the reason why I think the analytic frame is wrong is they look at AI only as a potential add to a negative risk. And you go, okay, well, it is an add to a negative risk. So then you go, oh, my God, you just stopped. It's like, well, actually, in fact, existential risk for humanity is a portfolio.

32:28We don't just have AI. We have nuclear war. We have pandemic. We have asteroids. We have climate change. We have a whole stack of things that could actually, in fact, have this existential risk. And so when you think about a particular intervention, you don't just think, does this one thing add a negative? It is what does it do to the whole portfolio? Even when you look at superintelligence, I go, okay, yes, there is an added negative risk. But I can also see ways that a superintelligence, they could add very positively to all of these other risks. Like, yes, there is a scenario in which somehow natural law of physics or somehow us fucking it up, we fuck it up, right?

33:09But there's also a lot of scenarios where it improves the longevity and the survival characteristics of humanity by dealing with these other issues. So it's kind of like I look at the portfolio of existential risk, and when I do my analyses, I look at this and I go, okay, I think this is going to be on balance even as it's going, has a much higher likelihood or probability of decreasing the overall existential risk portfolio. So it's going to decrease the overall existential risk even if it itself, to some degree or another, will likely pose an existential risk. That adds in a new existential risk, right?

33:49That's how I look at the existential risk thing. That'll be the first part. I want to do all three answers. Second part is I refer to AI as a cognitive industrial revolution. The transformation is equivalent or more than the industrial revolution. The industrial revolution, you might say, gave us physical superpowers and gave us manufacturing and transport and a bunch of other stuff. And this is mental superpowers. And the mental superpowers may even be more important and more transformational in various ways. And so the upside is as important. But by the way, the transition is very difficult.

34:21I don't think there's any way that we as human beings and societies go through these transitions without a lot of difficulty and turmoil and chaos. But the upsides are possibilities. Like, for example, I co-founded Manus AI with Siddhartha Mukherjee for trying to do drug cures for cancer and other kinds of things. Like the upsides are just spectacular. Now, this gets to the last question, which is kind of part of the reason I broke into third is part of what I think you have to be is kind of like as a student of human beings, the progress of technology, of societies, of industries, how do we operate as human beings?

35:00And the answer is we divide into various groups. But what that means is once you're kind of – the intellectual fascination usually tends to be like, oh, no, we get in a room, we all talk, and we all agree, right? And I have never seen human beings do that about anything significant. And by the way, you'd think we could do it for climate change where you could see more simple physics. And we don't do it there either. So the theory that we're simply going to go sit in a room and talk and all agree, for example, of an existential risk or of a job transformation risk. And then since we all agree, we're all going to massively slow down and stop doing things.

35:39That's never happened in human history. for things that are even much more visible. So that as a pipe dream for the conversation doesn't actually, in fact, yield the results you hope for. So you're saying that you can wish that human nature was otherwise, but the truth is that we're just going to have to admit that we are competitive and we're disagreeable and this is how we work. And therefore, we're not going to get together and solve all of the problems that AGI might pose. Is that about it? Yeah. And you can't just pre-think it by sitting there and going, I imagine how all this is going to work because you actually don't know what the landscape is.

36:20Like, for example, if you said, before I'm going to deploy a car, I'm going to think of every possible thing to go wrong and prevent everything. And you're going to end up with a thing that goes at five miles an hour, has six feet of armored plating, et cetera, et cetera, and it's just not going to work. And that will never work. And then someone else is going to go build other things that are going to be there. So you have to be iterating, kind of iterative deployment and be thinking about it and navigating as you're kind of going. I'd love to ask you what you make of the camp that I call the AI scouts, the people who agree with you that the AGI future would be awesome, would be ideal.

36:58the people who also want all of the great stuff that you want to come our way, but who say that to get that without destroying humanity, that we're first going to have to really solve a lot of problems. We're going to need to invest in AI safety research and alignment research and control research because this is a really serious problem. Why aren't you with them? Why not take the acceleration part seriously but also the getting prepared part? Yeah, look, I'm happy there's people that are doing AI alignment research, but I don't want there to be a sort of cottage industry. You know, if you say every corporation needs an AI safety department and you're in AI safety, yes, you're kind of arguing for your own paycheck.

37:43And it's not clear to me the utility of these folks yet. And it's not clear to me that the research has gone far enough. And whether it is of extremely high utility, I just don't agree that we should plow all our capital towards this research. And again, you're trying to steer a moving target. And so, again, it's just like I just don't think it's an effort that's going to yield outstanding results in terms of our ability to control AI. I think that the real driving forces are sort of the market forces and geopolitical forces. So just so I'm clear about what you're saying, you don't think their project is going to work.

38:19that trying to do something like ensure we have total control over AI is what? Useless? Is a waste of time? It's a subtle argument. I think you will have some control knobs, but just not as many as people want, essentially. And that's just an argument about complex systems, and it's very difficult to have control over really self-adaptive complex systems. It's just like the economy, right? We don't control the economy. What we control is, for example, the interest rate, the discount factor. over time. And with that knob, you can actually do some stuff. And we might have that for rewarding certain behaviors with AIs.

38:58And that might do something. And that's great that we do research there. But I just don't think we'll have control forever over these systems. And I do think that markets and capitalism are far superior technology for aligning incentives than sort of top-down prescriptions. Of course, you need a bit of both. You need some legislation. You need some market incentives. But because it's such a moving target, I think right now, I don't think most of the capital should go towards AI safety research. Some of it should go to AI safety, but I think there's already a good amount going into it. To go back again to your time at OpenAI, did you ever feel, as I've read Peter Thiel said, that the company had an effective altruist problem, that especially before Dario Amadei left, there were some people there who were too afraid of the risks of AI, that there was too much fear, and that that fear at times would get in the way of the company's broader mission.

39:59Yes, in a couple of different ways. I do think it was moderately too much fear by some individuals. And by the way, you still have that element. Like you have Jeffrey Hinton, winner of the Nobel Prize, going around saying there's a 60 % chance that AI will extinguish humanity, which I think is, frankly, completely incorrect. And that, by the way, induces panic and concern and causes you, as always with panic, to navigate in crazy ways. And I do think that part of what's happened is people have gotten more of a sense of, oh, actually, in fact, we have more controls over this. We can steer in some ways.

40:36And there's a lot of really good upsides. So I think it's still there as a we must get this right and is not an effort to kind of say, hey, everyone, I'm so important because I'm working on something on the humanity level. And more, I'm really trying to navigate this the right way. Well, because you just mentioned Jeffrey Hinton, I've had the chance to speak with both him and Yoshua Bengio. And they were not subtle with their opinions or their concerns. Ben-Gio went as far as to say that he has regrets about the work that he dedicated his whole life to and that he's worried about the future of his children and his grandchildren.

41:13It was a very intense thing for him to say, a very dramatic statement. And I want to ask you, when you hear scientists like them saying stuff like that, ringing this alarm and trying to get the public to come up with ways that we can ensure our survival, what do you make of that? So look, I always respect super smart people with criticisms. When the Manhattan Project was created, some of the people who created it thought that when you exploded an atom bomb, it would break the Earth's crust and turn us into a molten globe. So scientists are not always correct about things. And I think that part of what Joshua and Jeffrey think is that this should be done through university science.

41:58and part of, atmospherically, what makes them nervous is being done through corporations. And so it's like, well, they can't be thinking about safety the right way, they can't be serious, they have a profit motive, etc., etc. But by the way, the only way this kind of scale compute gets created is by companies, not by kind of universities. Now, that being said, I think that their theory of how human beings and human ecosystems work is incorrect. And I think it's the people get together in groups, they build stuff, they compete with other groups. There's even competition, of course, within academia.

42:33So I would go back to the question of it is a very good thing to be doing safety and alignment stuff. The work is to add in to the current progress on these things, how to make that happen within the framework of where we're going. that the theory of going around and going, we should just tell everyone to stop, is a theory that doesn't work. And matter of fact, it is, in a sense, sadly destructive because if anything, it just creates conflict around it in the groups that care about it. Like you could almost run this kind of two-by-two matrix about what is like people care about humanity, people don't care about humanity, people will listen to your message and slow down, people won't listen to your message and won't slow down.

43:22by going around and going, you should all slow down. Only the people who are aligned with your point of view are likely to slow down, and the other people are not likely to slow down. And therefore, you're having a dystopic interaction. Self-defeating, yes. Yes, right. And so the point is to say, actually, in fact, the good dialogue, the good discussion is, how do we steer this better? Like, what is the ways of doing that? And just contributing to it. And it can't just be, well, shift control to me. so that I do it. It's like, that's not a compelling answer, right, for any individual. It's a, here's how I'm going to try to contribute.

44:01And that's what I've been trying to do. And that's what I've been trying to get the critics to participate in doing as well. Not to say, hey, ignore all these crazy people. Reformulate your thinking into the things that can help us navigate the boat as we're going down the fast-moving river. Do you think that it would be good for our society if we were turning on our TVs or if we were going to live events where you had representatives from the accelerationist camp, the scout camp, and the AI Doomer camp engaging in public debates about this? Fundamentally, yes. Although, you know, by the way, fear is always the easy one, like in any new technology, fear will be more rhetorically compelling.

44:50Part of the reason why Hollywood creates tons and tons of movies about evil AI and almost no movie about good AI. And so that debate will have that characteristic. And so people just said, hey, it's just an objective thing. They'll walk away fearful. But I do think the public engagement with it is useful. And part of the reason why I'm here doing this interview with you. I don't think there are many people who are listening to this who would disagree with you about the pessimism and the distrust that is growing in our society. We're living at this time when people are regularly telling pollsters that they are worried the future is not going to be better than the past.

45:36Telling pollsters that their trust in every institution, you know, my vocation in journalism, government, medicine, like it's all in the tank, right? Yep. Do you see a version of accelerationism as being something that might inspire and unite people? The idea that we can build this technology, we can forge a better future, we can overcome all of the doom and the gloom of our current age. Is that how you conceive of this project? I would say there's an opportunity to rebuild all institutions, right? For the AI age, I think we need to rebuild education. We need to rebuild healthcare. We need to rethink how we do housing.

46:14You know, I think disruption represents opportunity. And that is the engine of social mobility. And again, it's like, if you are pessimistic about the future, you will feel like this disruption is happening to you or not for you. But the whole point is that we are trying our hardest to make it permissionless. You don't need anyone's permission to participate in the acceleration and the growth. And so you could just go out there and build something and capture some of the upside and, you know, secure a future for you and your family. And, you know, that's the point. It's the only real equalizing force that there is out there.

46:50And it's a rising tide that floats all boat rather than leveling everyone to the same level by, you know, having a drought and all boats hitting the rock bottom. You're saying that, yes, there's a lot of change that's coming, but we don't have to be afraid of it. In fact, if you refuse to be afraid of it and you look ahead with optimism instead of with pessimism, you actually are a part of making a better future more likely. Yes. I think people should be optimistic. We live in a time of extreme opportunity right now. If you believe you can change things, you will actually change things. I'm an example of this.

47:31If you explained to me how much impact I would have had on U.S. policy three years ago when I started Anonymous Account to just share some of my thoughts, I wouldn't believe you. But at the time, I thought it was important and I had to put these thoughts out there. And you just have to start. And I think a lot of people have self-limiting beliefs. They limit their agency. But now is the time of the ability to execute on almost any idea now with this amazing technology. so what we are in the need of is people of more agency and pessimism erodes your agency and wears you down so you need to shed that and embrace optimism and a belief that you can make the future better because if you believe that then you will it's that simple

48:27So I want to go back to this idea of a transformational period that you think is coming our way because you compared this to the Industrial Revolution. And it is hard to summarize just how impactful that transition was, how tough that transition was, where, yes, we got the train and the car and electricity and the radio and eventually the airplane and all these amazing technologies. But alongside them, we also went through these really tough social changes. You know, the mass population movement into cities, followed by, you know, crime and diseases spreading and then pollution getting out of hand.

49:09We had labor abuses and then labor unions and labor riots and then counteractions from the government to tamp down the riots. And it was very chaotic. Eventually, you would get the rise of socialism and communism and pretty much all the stuff that we fought world wars about in the 20th century comes from this transition. Are you envisioning that kind of a tough transformation coming our way? Is it that dramatic? I hope not. It's certainly not impossible. And that's part of the reason why I'm focusing on them. And part of the reason why I'm saying, look, we have tool sets that are better, including AI itself, for this transition.

49:53Like, I think we've learned lots of different things in these transitions. One is, you know, can we deploy the tool sets to do that? Another one is, you know, what Roosevelt was doing with the New Deal. It's like, okay, what are the things that we can do in order to try to make these transitions less painful, shorter, only smaller percentages of the population any particular year, et cetera, as we're doing it? And that's part of where the dialogue and the focus should be around this.

50:28as profound as technology has been AI will be more impactful as we gather this afternoon we're still in the earliest days of one of the most important technological revolutions in the history of the world this AI revolution is not made up it's not overhyped the last invention was created as a series to investigate the background and emerging debate around this AI moment. But this is a story that in many ways has only just begun. We need to advance legislation that promotes long-term AI growth and innovation. Who is determining what's happening? I've got a handful of people who are really determining the future of the world.

51:15That's scary stuff. And as the AI debate moves further into the halls of traditional power, and as the effects of AI begin to shape the world around us. The arrival of this new intelligence will profoundly change our country and the world in ways we cannot fully understand. We expect that surprises are coming. We know that what seems normal right now may not be normal for very long. And so going forward, we're going to be speaking with the people who are grappling with exactly how to understand, how to harness, how to regulate, how to adapt to this emerging technology. They hail it as the greatest breakthrough for humanity since the Industrial Revolution, maybe the printing press.

51:59From lawmakers. Of course, it will also make them additional ungodly sums of money, but that does not seem to explain the near religious fervor of AI's loudest advocates. To skeptics. Pretty much this entire generative AI boom has been based on lies. It's been based on suggesting what this might do versus what it actually does. There is no it. All there is is a belief system that we're making God in a box. And so while this episode marks the end of our initial limited series run, we would love for you to, as we say, stay with us as our reporting continues. And we just want to say thank you to all of the people who have been willing to speak with us, both on and off the record, and of course, to you for listening.

52:43and we would love to hear from you. What questions do you have about AI? What keeps you up at night? What are you excited about? What would you like us to investigate? Who would you like us to interview? You can reach us by sending an email to hello at longview.report.

53:15The Last Invention is made by Longview. It's produced by Andy Mills, me, Matthew Boll, Gregory Warner, Andrew Parsons, Megan Phelps Roper, Seth Temple Andrews, and Ethan Manello. Special thanks to Carmen Hilbert and Jonathan Farber. Music was composed by Scott Devendorf, Ben Lands, Kobe Beanert, and me. Scott and Ben have put out a soundtrack. You can find a link to that album in our show notes. The artwork and design for the show is by Jacob Boll. We're proud to announce that episode one of the series was just named by Apple Podcasts as one of the best episodes of 2025. Thank you to Apple and thanks to you for listening.

53:55We hope you'll continue to share the show with your friends and community as an introduction to what might be one of the most consequential conversations shaping our future. For everyone out there already supporting the work we do, we are deeply grateful. It is because of you that making a series like this is possible. If you would like to become a supporter and learn more about our team, visit us at longviewinvestigations.com or click on the link in our show notes to become a subscriber. We will be back soon. Goodbye for now.

54:38on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required, compatibility and availability varies 18+. You can't reason with the sun. Trust us. We've tried. This summer, it's time to put that angry ball of fire on mute. Columbia's OmniShade technology is engineered to protect you from the sun's harsh rays. that can burn and damage your skin. The sun is relentless, but so is our gear.

55:13Level up your summer at Columbia.com to spend more time outside and less time slathering on aloe lotion. You're welcome. Columbia. Engineered for whatever.