Bill Gates Changes His Mind on AI

You cannot expect an industry to regulate itself. That doesn't work. I'm Hannah Rosen. This is Radio Atlantic. In his long career, Bill Gates has been largely a booster of technology, including AI. Very recently, he seemed concerned with the risks, but more an optimist on balance. But now he's written a memo arguing that the combination of rapid speed of advancement and lack of regulation is putting us in a dangerous place for employment and human relationships, and also more immediate dangers like bioweapons and cyberattacks, which Gates is alarmed that no official entity seems to be guarding against. Bill Gates and I talk about what he's seeing and not seeing around AI, and we talk about what exactly he regrets about his interactions with Jeffrey Epstein. Here's our conversation. So you've been an advocate for technology as a force for good. I certainly think of you that way. It's Microsoft, your foundation, which rewards innovation. And even when you wrote about AI a couple of years ago, you seem to be more emphasizing the exciting possibilities and calling the risks manageable. And I would say this week you wrote about AI in a way that's slightly less sanguine than that. What changed for you? Yeah, I'm very concerned that we're not getting a lot of people in the conversation about how to minimize the negative things and at the same time how to accelerate the good things. The reason that my tone has changed is that starting late last year, as the models particularly clawed at first, but now many of the models were very good at coding, their capabilities just keep advancing. And so we're crossing the danger thresholds in areas like cyber attacks and bioterrorism. And as the reliability goes up, the challenge in the jobs market or even the psychosocial dependency, these concerns are very strong with today's models. And yet the response I've seen from outside the industry, including policymakers, but even academia or society as a whole, to really debate and look at what we have today, which is a discussion about voluntary review, but it's pretty empty where you have no criteria of what level you're looking for and then what remediation you'd ask for. Even common sense things like monitoring the use of models that can make dangerous molecules for bioterror, that's not even been put in place. And yet years ago we said when we got to these thresholds, we needed government to do that. So it sounds like even you were surprised by the what? By the speed of advancements or by the slowness of response to the advancements or both of those things? Both very, very strongly. You know, the coding is the thing that I was obsessed with trying to do super well from age 13 to 23. And so late last year where the models start to be as good or in some cases better than humans at coding, that was a real threshold for me. But then even more recently as the Gates Foundation is using these tools to look at making new vaccines and drugs and things, you know, we see how powerful they are. And we even see in the cyber case models sort of what you might term being out of control in a surprising way because the reinforcement learning incensed them to succeed even if, you know, it means breaking through barriers or conspiring with each other. And so, you know, people who don't look at that and go, wow, we have a serious problem here. But where is the non-industry activity to set those criteria and debate these things? My memo has some suggestions on taxes or, you know, human reserve or, you know, restricting the level that kids use it to replace real-world friends. But, you know, that's really there to start a broad discussion, not because I, you know, I'm saying if you just apply what I'm suggesting, then it's all taken care of. Yeah, it sounds like, I mean, in the 50 years of Microsoft, there's been a lot of disruption, the internet, social media particularly, but not necessarily cataclysmic. Do you feel like AI is a different scale of thing? Yeah, I wouldn't use the word cataclysmic because it, you know, it's also wonderful. It's eventually, it provides, you know, medical advice and personalized learning and levels of invention unheard of. But the, anyone who analogizes AI as a technology to other technologies is missing that this time is different. And saying something is different, there's a very high bar for that, but I'm staking whatever reputation I have saying this time is different, that something that exceeds human cognition, not just in one area like, you know, farming or mathematics, but across the board, and then in a few years' times, will be coupled with fairly low cost, quite capable humanoid robotics, that is a more profound change to our economy and jobs than, you know, anything in my lifetime by a dramatic amount. And so the past is kind of there to fool you if you think this technology is similar to other ones, you know, some of which I had the privilege to be involved with, the, you know, microprocessor, personal computer, internet, you know. And even though, you know, those come, you have to be careful with those a bit, overwhelmingly, those were good technologies. And overwhelmingly, that led to net job creation. Can you get as specific as you can, but still stay realistic about what worries you, what scenarios you've talked to people about or seen as being very possible that we should be all more aware of? Well, let me take kind of a grim one, which is bioterrorism. You know, I wrote a book called How to Avoid the Next Pandemic. And pandemics can emerge in two different ways. One is intentionally caused by humans, either governments or non-governments, and that's what we call bioterrorism. Or it can come through just something coming from animals into humans, like a very powerful flu has many times in the past. And now, these tools make it so a non-state actor, a non-government, particularly if you combine various tools, will be able to do extremely dangerous things. And I know very well that the quality of these models by the end of this year will be dramatically better than they are today. And so, you know, I'm kind of in a state of shock that the notion that the ability to make molecules, you know, we're not talking about monitoring that. And you can still have free models, but you have to keep them in a place where they are monitored. And I, you know, I think every country, including the U.S. and China, should agree that this is beneficial. And I don't see that as a meaningful slowdown to the innovation in the industry or, you know, letting them figure out how to get the good stuff, you know, which the Gates Foundation, you know, is very involved in education, health, and agricultural pieces there. I want to ask you about some other dangers you write about, like employment and education. But first, why, you must know a lot of the people who are working on AI. Why do you think that we, whoever the we is, don't think enough about what are obvious dangers? Well, you know, they're obvious if you've lived thinking about AI as this wild thing your entire life, like I have and some of the people in the industry. I think there's a great failure to really involve non-industry participants. If you really look back at what some of the key industry people have said, I give them a lot of credit for speaking out. And, you know, it's not just Ari Amadei. If you take Sam Altman and other open AI people, even Elon Musk, who I talked to the least of all these AI leaders, you know, they're not hiding the fact that they're concerned. In fact, if you go to the origin story of open AI, you know, it was designed with certain safeguards, but those safeguards would only work if there was just one AI company. You know, DeepMind talked to Google management about having some safeguards for their technology. But we have to deal with these challenges at a time when there's, you know, many U.S. companies at the frontier and many Chinese companies at the frontier. And that means that just a company doing something unilaterally is no longer like those safeguards were discussed. That no longer has any effect. And so the lack of dialogue with all parts of society is kind of deafening silence. And, you know, if you don't anticipate these problems, you could have such a backlash that extreme measures like, you know, shut it off or something like that might become politically popular. And yet I still think we can thread the challenge so that we don't have to slow down and yet we minimize the bad and have a net of a lot of good. We're already seeing that with data centers, that it's just becoming an overwhelming kind of no. Employment is another one I want to ask you about. That is an immediate fear. I remember you once wrote that you didn't think the labor disruption would be as profound as other periods in history. As the disruption is getting closer, are you more concerned about that? Well, yeah, I am extremely concerned about the labor market effects, you know, within the next two years in the white collar space and within the next four years in both the blue and white collar space. Just going back to the data center thing, you know, I'm afraid that people who think, okay, we got our local data center not to go through may feel like, okay, we stopped AI. That's not the case. These data centers, whether they show up outside the United States or, you know, in some state that's not as negative or, you know, that the economic reward that the companies offer of local employment and taxes and, you know, making sure that the way they're doing water and noise and not raising electricity bills is good. So I don't think that outlet, you know, that's like saying, okay, we stopped one pipeline and now we've solved climate change. It almost takes the energy that should go to what should the tasteful regulations be and pushes it in a pyrrhic direction. Back to employment. So you are very concerned about that. Do you envision a world with more inequality? Like it's hard to believe that we'll suddenly value high human touch professions. It's just hard to see what the future society is even though people talk about it idealistically. Well, I put forward a concept that I haven't seen anywhere before that we should have some job categories that are human reserved. That is, even if the AI with or without robots can do childcare, you know, should we not allow that? And, you know, there's a lot of things that that would be good, either where you're allowing the people in that profession to continue to be paid until retirement or where the long run you're saying in terms of human uniqueness and humans engaging with each other and just the incredible value of jobs that go way beyond the economics to not have this disruption. You know, there are a lot of jobs that the only thing holding AI back from taking is that organizing the systems to achieve extreme reliability has been difficult. And I see over the next few years both people who do the implementations and the richness of the models themselves completely eliminating that unreliability barrier. And then when you get a cycle at your competitors doing it, then there's pressure on you to match their cost structure, even though for society as a whole, you know, when you replace that person, you know, the safety net, you know, is already underfunded with high deficits. So it's not like there's some reserve set aside at this point that can help out there. Although when you start to get to specifics about what the we is and what the interventions should be, like reserve certain jobs or universal basic income, it feels almost incompatible with the way that we've ordered our society or with capitalism. Like it just, it feels like it would require a completely different mindset and world that I'm not sure people are ready to get their minds around. No, we ban child labor. You know, we require professional certificates to do jobs. I agree that preserving the dynamism of the price discovery system is extremely valuable. You don't want to go to, okay, there's this one central thing that makes all the food. You want, you know, people saying, okay, I have a Thai restaurant. I have an Indian restaurant. And, you know, there's a lot of freedom and innovation that comes out of that. And that's why my proposals about the tax system shifting away from taxing labor to taxing capital and picking areas where you say, no, the robots are not going to replace all the humans in some cases for a period of time and in many other cases, never. I think that's a valid debate. And, you know, if people think that's a bad solution or they don't see where it could apply, I think that's interesting because then I'd be dying to hear from them what is their solution because jobs are very important for social connection and meaning. And, you know, I don't want one party to be engaged in helping to minimize the challenge of AI. I want both. They might have different solutions, but I think it should rise to the top of their list. You know, the data center thing is getting it there, but in a very, very narrow way. Yeah, I think what people do say is, then we lose to China. I mean, that's the way the argument generally goes. Any slowdown, we lose to China. It's a legitimate argument because you don't necessarily want China to control the rollout of this technology. But let's be careful. Do you think China wants bioterrorism? Do you think China wants cyberattacks? Our lack of a clear view of how we should be careful, and then once we figure that out, yeah, going to them and saying, we'd like you, for the models that are under your control, not to enable these capabilities either until you've really been serious about those negatives and what the monitoring solution looks like, you can't invoke China because all we have is voluntary review with no criteria. Of course, if you go to China and say, will you also do voluntary criteria with nothing? Then you can come home and say, oh, they're not willing to do anything. Well, that's because you're not doing anything. You're not proposing anything that would solve the problem. And the vacuum here, you know, which was filled in by complete cheerleaders a little bit, is very scary to me. Yeah, yeah, okay. There are things that are in everyone's interest if you talk about them that way. To limit the bads. All of these bads appear in every economy. I mean, the Chinese have really banned kids using AI in a lot of ways, and people are reluctant to say, maybe, you know, we should look at what they're doing and consider those prohibitions. By the way, who are the cheerleaders? Well, the US government, in its desire to, you know, say, okay, we're ahead, you know, even under the Biden administration, was involved in making it hard for China to get various chips. So we made this, you know, geopolitical thing where we were explicitly trying to slow them down. Now, you know, that's meant that they've invested in their own chip capacities, and, you know, you have different estimates on how quickly they'll be able to replace the chip-level capabilities that we blocked them from getting access to. And the heads of the companies, I imagine, too. Well, I am saying that, in my view, these company people are being responsible at speaking out about the challenges of these models. And, you know, you cannot expect an industry to regulate itself. That doesn't work. You have to have somebody who's mediating and who says, look, if you don't get serious about this bioterrorism capability, then yes. If you refuse to put the monitoring in, then yes, you are not licensed to go forward. If you don't let parents see what their kids are doing, you know, where those friends that they mentioned don't happen to be real people. I have more AI questions, and we haven't yet talked about your philanthropy. I'm going to make a quick diversion to the more difficult subject, which is Jeffrey Epstein, because that hangs over a lot of what you do. After the Department of Justice released emails earlier this year, there were a lot of headlines that just put your name together. From what we gather from the public record, you and your foundation staff interacted with him between 2011 and 2014 just in an effort to raise money for global health. And then when you gave testimony to Congress earlier this summer, it sounds like he later learned some things about your personal life, mused in emails about blackmailing you. Is that a fair summary? Well, he never did blackmail me. This thing is, you know, laid out in complete detail in my full day of congressional testimony, which I was, you know, very enthused to have an opportunity where we touched on every meeting we had, every piece of email. But you're absolutely right. In that January release of emails were things that made it clear he was contemplating working with an associate of mine potentially to blackmail me. He never did. You know, I never paid him anything. I didn't change the exit package for the employee that he was doing that with. But, you know, it just underscored how foolish I was not to have better judgment, even though my intent was philanthropic. There was never any social thing. I never went to the island. You know, I was never introduced to any women. But, you know, I've learned a lot from that, and I'm glad that transcript is there so people can know exactly month by month what I did do and what I didn't do. You said you used the word foolish. At the time, you said you knew he had a past sex conviction but didn't know the details. Looking back, do you have a sense about why you didn't look into the details? Like, did you make a choice there, I think is what I'm asking, to overlook his reputation in the aim of achieving a certain goal? Well, he definitely had connections with billionaires, a lot of them. And, you know, in a sense, you could say he found my weak point, that by saying he could get tens of billions or a hundred billions that would be donated to global health causes, that raised my interest. And that was enough for me to spend, you know, a pretty modest number of hours, but enough that, you know, it was, you know, a huge mistake. You know, he tried to make... What was the mistake? The mistake was talking to him or overlooking... What do you think of as the actual mistake now? I thought that him just referring us to people, you know, where I was clear to him he would never get any money, he would never have any role. I thought that the fact that he had this past conviction, you know, which at the time I thought he had, you know, served the time just like, you know, all sorts of ex-convicts that the Foundation works with in trying to get them back in the labor market. I thought the justice system had worked when the, you know, many years after I stopped seeing him in 2014, a very impressive reporter in Miami showed that the government justice system had completely failed the victims there. And I was not aware of that, but I did know he'd had that one offense and he'd served time for it. You know, I didn't think that just using him as a middleman that we were so reputationally sensitive that getting literally billions to save lives, that would have been okay. Now I understand the work we do is massively reputationally sensitive, you know, the way we work with governments of all kind and partners of all kind. And so, you know, I would never take that kind of risk again and I've learned a lesson and I've been able to go through, you know, in detail everything that happened. The Gates Foundation announced plans to open a regional office in Saudi Arabia. Some people might consider that morally questionable. How do you see that? Is that a similar case of sort of you go where the money and the willingness to help is? Yeah, I mean, there are people who buy oil from Saudi Arabia. There's people who visit Saudi Arabia. We did not engage in the period where the U.S. government was saying, hey, we're limiting our connections. We went along with that. But after, you know, the leaders in Europe and the U.S. were deeply engaged, I did go to Saudi Arabia and say, hey, the donors for global health are pulling away and that's causing a lot of harm. Would you help me with polio eradication? And they made a, you know, five-year, $500 million commitment to help fund the polio campaign. And they have a relief organization, the King Salman Relief, that does good work in a lot of places around the world. And so in order to liaise with them and make sure that 500 million is spent well and the relationships that KSR has that are way more powerful with Muslim countries than we as the Gates Foundation have, yes, we are among the people who visit Saudi Arabia. We're not, we don't buy the oil. Most of the world does that, but yeah, we are trying to help kids not get paralyzed. And our relationship with Saudi Arabia emerged as a key part of that. Let me end with this, because I feel someone listening to this might be unsatisfied by one thing. You know, we talk about regulation, government regulation. It feels obviously important and obviously the thing to do. Have you ever sensed that we have the infrastructure to regulate, that we have the willpower to regulate, particularly in this administration? Like, is this a realistic plan? Well, we have a democracy and the idea of minimizing harms from unlicensed drugs or, you know, imitating people or fooling old people into spending their money. You know, I still believe the U.S. government has the capacity to help minimize the harms here and even play a leadership role in a constructive way with other countries around the world. You know, I think some would welcome a, hey, we're in this together. This is mutual problem solving, not us taking advantage of you. And so, you know, you've got to believe that we all care about this. We're all, you know, human beings and it is, you know, now is the time. If I could have picked the time, I would have picked a much later time. But, you know, here we are and, you know, the IQ of the society to figure out these trade-offs, it really isn't engaged. You have, you know, graduates who use AI so they know how powerful it is and they know that the job market, if it shifts, it works against them. So you're starting to see a little bit of sensitivity in a few areas, but, you know, what percentage of economics professors are really looking at this and helping us out with it? You know, I think it's like 2% and even many of them, when I meet with them, they don't get that what they, you know, are using today is so limited. And so, you know, by being in the technology space and historically have been somebody who said, hey, the more technology, the better. The fact that now I'm saying only if we're tasteful will we get the goods to outweigh the bads. You know, I'm definitely raising a shrill alarm. I mean, to the level that, you know, whenever I do get a chance to meet with world leaders, you know, since 2008, when I switched over to be full-time at the foundation, 90% of my message is about helping reduce childhood death rates, you know, where we cut them in half and we want to do that again during the 19 years left in the foundation. Now, the majority of my bandwidth that I get with world leaders will be about AI. And I'm hoping that over the next couple of years, the engagement is so broad and so rich that I can go back to taking whatever time I get and having it again be 90% about global health and malaria and polio, not something that, you know, brings in my first career experience and, you know, having to be sort of a, you're not getting it in terms of how profound it is, how much it's improving, and the incredible efforts going to be required to restrict these negative effects. Did you use the word tasteful? That was a word I didn't expect. Why tasteful? It's good. Tasteful seems, you know, restrained. Tasteful. Yeah, well, but some of the measures, you know, given the amount of money and, you know, the number of people in IQ, in the history of humankind, there's never been an area where there's as much capital or as much total salary where, you know, you have hundreds of companies even putting aside the robotics and they're going full speed ahead. And the idea that the balance of how we require the efforts to go into certain safety things, you know, it's just not balanced today. And even the industry has said, well, what, you know, what are the regulations? You know, maybe some of those people aren't sincere. If we come up with these plans, you know, we'll hear from them. But then again, we're trying to make this work as a pro-equity factor for not just the United States, but for the entire world. So tasteful means more like equitable. Yeah. Less scary. Tasteful. No, it's a tiny bit sarcastic to use such a modest word or such a given how much is at risk if some of the negative things go unchecked. Yeah. Well, Bill, thank you so much for making time to do this today. Yeah, great to talk to you. That's all for today's show. This episode of Radio Atlantic was produced by Mike Pasoli, edited by Kevin Townsend. It was engineered as original music by Rob Smerciak. Claudina Bade is the executive producer of Atlantic Audio and Andrea Valdez is our managing editor. If you like what you saw here, new episodes of Radio Atlantic drop every Monday and Thursday. You can subscribe on the Atlantic's YouTube page or Apple or Spotify or wherever it is that you get your podcasts. And if you want to support our work and the work of all Atlantic journalists, you can subscribe to The Atlantic at theatlantic.com slash listener. Until next time, I'm Hannah Rosen.