Tech leaders are debating how risky AI is. What does Bill Gates think?

0:00It's Consider This, where every day we go deep on one big news story. Today, decisions about AI that could affect the life of every person on Earth are currently concentrated in the hands of a few fantastically rich tech executives. We will ask Bill Gates, the fantastically rich former tech executive, why he thinks that is a problem. Now, as you have heard a lot on this podcast in recent days, AI companies are warning about threats posed by further advances in AI and actively asking to be regulated. The issue we're talking about is there's a collective action problem here. That's Anthropic co-founder Jack Clark speaking with NPR Tuesday.

0:40The CEO of his company, Dario Amadei, has called for a slowdown on AI development. It's okay for individual companies to do this, and absolutely they will, But what do you do about the whole set of companies that are in deep competition with one another and are to some extent racing to build more and more capable systems? And we'll note here that Anthropic is a financial supporter of NPR. Senator Bernie Sanders is proposing a bill that would ban what he calls super intelligent AI and temporarily halt advanced AI development to give federal regulators time to develop safety standards. Whether you're conservative, whether you're progressive, we want this world to be healthy for our kids, our grandchildren and future generations.

1:23And we've got to err on the side of caution here. Which brings us to Bill Gates, who sees huge potential and huge peril lurking well short of an AI apocalypse scenario.

1:38Consider this. Bill Gates says left to the free market, AI will be designed by and for the richest people in the world. So what does he want to see instead?

1:51From NPR, I'm Scott Detrow. Imagine political debate that sounds like this. Okay, Mo's totally right. That's such a good point. Sarah laid it out very well when she laid out the two sides of the argument. It may seem like Mo and I agree too much. Yeah, that's about to change. No need to imagine this. It is real. Left, right, and center from KCRW, part of the NPR Podcast Network. Actual facts, all sides, and good vibes only. Out every Friday, wherever you listen to podcasts. Up first from NPR News is your way to slow down a fast-moving world and make sense of it, which is just what tech leaders say they need right now.

2:27One of the big stories this week is that tech leaders say they want to slow down AI development. Also, the latest fighting in the Middle East could push oil prices even higher. Start your day with what you need to know on Up First. Listen on the NPR app or wherever you get your podcasts. This is Tanya Mosley, co-host of Fresh Air. Each week we go beyond the headlines and dive deep into an issue like the Epstein files, the dismantling of voting rights, and the conflicts abroad. Listen to Fresh Air interviews to better understand our world. Find us on the NPR app or wherever you get your podcasts.

3:11It's Consider This from NPR. Bill Gates is an interesting person to talk to about AI regulation. He was CEO of Microsoft in the 80s and 90s when it was the juggernaut of the tech world. And the company pushed back hard on the federal government's attempts to check its dominance through antitrust cases. But today, Gates is focused on giving away his fortune through the Gates Foundation, including, he announced Tuesday,$1 billion to expand access to AI tools in order to tackle social inequality. Gates sees AI as a technology that could be an engine for equality or for injustice, depending on what happens over the next year and a half.

3:54And that is where we started our conversation. Since we've heard so many apocalyptic scenarios this week, I want to start with a concrete example from you of how you think AI tools could help lift people from poverty. Well, the Gates Foundation, the majority of its work is helping low-income countries in health, agriculture, and education. So AI, for example, if you take a farmer, it can take your photos of your crops, tell you what's going on with the disease. It can give you weather data that you never had before. It can tell you about your soil and what fertilizers you should use. And it'll understand the market prices and when you should store your crop and when you should sell it.

4:38And so it's been phenomenal to see that already hundreds of thousands of farmers in Asia and Africa are using the AI-based tools to make themselves way more productive than before. You talk about those in your writing that you've recently published. You also raise really serious, specific concerns. And some of the things that you are flagging are a little different than some of the things we've heard in recent days. You're not just talking about models going rogue, cyber attacks, things like that. You're talking about society-wide inequities. Can you walk us through some of these specific scenarios that you have become worried about?

5:21Well, I think everybody's taxonomy of the concerns is pretty much the same. There's the biosecurity, cybersecurity. There's the job market and how AI, including future humanoid robots, will affect the job market. There's the psychosocial effects of kids or even adults having a dependence on the AI for their relationships. And finally, there's the issue of control. Will the AIs either be under control of bad people or out of human control in a way that's detrimental to all of humanity? And so, believe it or not, we have to address all of those challenges that I call that minimizing the bad while we accelerate the very positive things.

6:20And so the billion-dollar announcement from the Gates Foundation, where we're going to make sure African languages are very highly accurate, which they're not yet today, that's about accelerating the good uses of AI into health, agriculture, and education. Given that, what was your response when President Trump said earlier this week that he did not see a role for the federal government regulating it? And he said, the only control are guardrails, and this is a quote that AI needs, is a strong and smart, high IQ president. I wish it was that simple. This is a technology that was not developed really with government R &D.

7:03And so unlike rockets and nuclear weapons, where the government has tons of people who understand how quickly it's moving, here, the federal government, as big and as smart as it is, is really playing catch up. And so they have to deepen their expertise in biosecurity and cybersecurity. My memo was a rallying cry that people in universities, people in nonprofits, and the government get more engaged in how we accelerate the good and minimize the bad. So I think as the U.S. government deepens its expertise, the message to the president will be these are, in fact, very, very serious problems that should not be taken lightly.

7:57And in the meantime, the main focus is on whether the the AI giants self-regulate. And that's something I was interested to hear your take on, because I think you probably understand the Sam Altman's and the Dario Amadei's more than anybody else in the world. if an outsider had warned you to slow down when you were building Microsoft in the 80s and 90s, would you have listened? Well, fortunately, you know, nobody ever said that Microsoft Word, you know, was a problem because somebody could type something rude or a ransom note, you know. Yes, it is very surprising that in my comments on AI, I am saying this is a case where going full speed ahead is not the complete answer.

8:39My whole life, I've always said, you know, more PCs, more internet. You know, I believe innovation can solve almost every problem. But this is such a dual-sided technology that here, some degree of slowing down may be important. And I think the industry is not the issue here. The issue is getting all of the rest of society who really controls these tradeoffs to get them up to speed. And so that without reducing the good stuff significantly, we can make sure to minimize those five all significant areas of potential harm. I did want to ask, as we talk about these potential solutions, I did felt like I need to ask you about your association in the past with Jeffrey Epstein, because the Gates Foundation has been such a major player on policy.

9:41I'm wondering if you are concerned, if you feel like its ability to impact these future conversations, to be in the room to make decisions, was harmed in any way by this association? Well, certainly I'm very regretful that I spent time with Epstein. I really appreciated the opportunity to go in front of the Congress and go through, you know, meeting by meeting during that three-year period that ended in 2014, what my interactions with him were. I made it clear what I did do. I did spend time with him, which was a mistake. I never had a social relationship, went to an island, made any payments.

10:24So people should look at that. There are many reasons. I am a billionaire who's benefited from technology. So hopefully, I'm just one voice of many. But I tried in my memo to say, yeah, you can say that I'm not the perfect messenger, although I'm not sure who is. And on AI, whether it's preventing the worst or putting the guardrails in place or making sure it's distributed in a broader way, how much time do you think we have to get these changes in place before it's too late? I think the sooner we say that models need to not be usable by bioterrorists, the better. The threshold of being capable of aiding bioattacks and cyberattacks, that is being crossed right now.

11:15And I know that the models that will come out later this year are far more powerful. And so there really isn't time to spare. That was Bill Gates, philanthropist and entrepreneur. Thank you so much for talking to us. Thank you. And we will note that the Gates Foundation is a financial supporter of NPR. This episode was produced by Connor Donovan and Erica Ryan. Our director is Jonas Adams. It was edited by Patrick Jaron Wadanan and Tinbeet Ermias. Our interim executive producer is Courtney Dorney.

11:58It's Consider This from NPR. I'm Scott Detrow. Want to hear this podcast without sponsor breaks? Amazon Prime members can listen to consider this sponsor-free through Amazon Music. Or you can also support NPR's vital journalism and get NPR Plus at plus.npr.org. That's plus.npr.org. Looking for love? One New York event promises to find you your perfect match using Palantir, though not affiliated with the actual software company. They can sue me. I could lose everything, but it would all be worth it if at least two people here have a spark. So, no pressure. The facts, myths, and dark jokes surrounding Palantir.

12:41Listen to Planet Money on the NPR app or wherever you get your podcasts.