The peril (and promise) of AI with Tristan Harris: Part 2

0:00Wondery Plus subscribers can listen to How I Built This Early Early and Ad Free right now. Join Wondery Plus in the Wondery app or on Apple Podcast. Listening on Audible helps your imagination soar. And no matter what you like, Audible's romance collection has something to make you swoon. Here's your invitation to have it all. Find a book boyfriend in the city and another on the hockey field. Or if nothing on this earth touches your heart, you can always find love in another realm. Hear modern rom-coms from authors like Lily Chu and Allie Hazelwood. The latest romantic series from Sarah J. Maas and Rebecca Yaros.

0:36and Regency favorites like Bridgerton and Outlander. And that's only the beginning. Audible has an incredible selection with over one million audiobooks, podcasts, and Audible originals all in one easy app. And you can enjoy Audible anytime while doing other things, household chores, exercising, on the road, commuting, you name it. Audible makes it easy to drop into your fantasies during your everyday routine without needing to set aside extra time. There's more to imagine when you listen. Your first great love story is free when you sign up for a free 30-day Audible trial. Visit audible.com slash built.

1:14I love traveling with my family. We did an awesome trip this summer. And one of the things that made the trip so special were the Airbnb experiences we did. Immersive tours, cooking classes, a chance to get coffee with a world-class barista. I had so much fun on those experiences that I decided to host my own Airbnb original experience in San Francisco, designed to help you think about how to unlock your next big move in your career or even in your life. To learn more about my Airbnb original experience, head to Airbnb.com slash Guy. AI companies have unique business models, each with distinct billing needs.

1:57Stripe is the go-to choice for AI leaders from early stage startups to scaled enterprises. Whether it's OpenAI offering tiered plans based on user needs or Cursor implementing flexible billing structures for code access, redefining the future requires sophisticated business models. With Stripe Billing, you can support any business model from credit-based to sales-negotiated contracts and easily align your monetization strategy with customer value. Join the ranks of 78 % of the Forbes AI 50 and millions of businesses worldwide that trust Stripe to help them build more profitable, scalable businesses.

2:41Discover more at Stripe.com.

2:48Hello and welcome to How I Built This Lab. I'm Guy Raz. So what if you could no longer trust the things you see and hear? I'm not talking about conspiracy theories. I'm talking about the breakdown of what we now consider to be hard facts, evidence. What if you couldn't trust the signature on a check, the documents or videos presented in court, the footage you see on the news, the calls you receive from your family? Because they could all be perfectly forged by artificial intelligence. The breakdown of trust in our society, that's just one of the risks that could be headed our way as AI gets smarter and smarter.

3:27And that's why my guest today, Tristan Harris, is sounding the alarm about the rapid development of AI. This episode is part two of my conversation with Tristan. He's the co-founder of the Center for Humane Technology. And if you haven't listened to part one yet, you'll want to go back and listen to that first. In that episode, Tristan talked about how so many of the technological tools we use every day, things like social media and search engines, were designed to grab as much of our attention as possible. And that's had some really damaging effects on our society. We also talked about the exponential development of AI and how it's advancing so quickly that even the people that work on it aren't aware of the full scope of its capabilities.

4:11Today, Tristan is back to talk more about how AI is changing our lives, what we need to worry about, and how we can protect ourselves from some of the scary stuff. But not everyone is worried about the dangers of AI, which presents its own challenge. Tristan, I'm here in the Bay Area. You are too. And I go to, from time to time, I'll go to events, meetups, just to observe what people are talking about around generative AI. And there's a lot of excitement about what's happening in San Francisco and talk about, you know, this is the next big thing. People are saying it's like what it felt like to be here in 2003, 2004 with Web 2.0.

4:55So there's a lot of excitement around it and not that much skepticism. And so I wonder, in a world where profit is incentivized, obviously we live in a capitalist system, what would stop somebody from pursuing this at lightning speed? I mean, if they're incentivized by financial rewards to pursue it. Well, if they're fully incentivized to go as fast as possible, and there's no counter incentive that says you're liable, let's say for the harms that might show up, then of course, you're going to go as fast as possible. And so that's, that's why in our work, like people think that we're criticizing Sam Altman, or OpenAI or one company, or we're criticizing AI overall.

5:40No, neither of those things are true. What we're criticizing are perverse incentives that lead to bad outcomes because we are true futurists who want the good future. And we see that to get that future, we have to change the incentives that we're currently operating with. And a good example of this is liability. So what lesson did we learn from social media? For those who don't know or remember this, in 1996, there was this thing called the Communications Decency Act, in which there's a section famously called Section 230, that basically gave all internet companies an immunity shield, that you would not be liable for anything that your online bulletin board, where someone posts hate speech or something like that, or tells people to commit suicide or smear someone, that you would not be liable for any of those harms.

6:28And that made sense when the internet was just a bunch of bulletin boards, and it was not powered by AI. But we used that immunity shield and applied it to social media companies when they came along later. And so when they go and intentionally addict children, use social comparison, use variable schedule rewards, use social proof and social validation and direct messaging and all that to try to jack up their products, we allowed social media companies to not be liable for any of the downstream harms that we're now living with. And a correction we could make to AI companies is that instead of being incentivized to race as fast as possible, what do we want to incentivize?

7:08We want to have them move at the pace that we can get this right? Well, get this right would mean, what if everybody was liable for the downstream harms that could occur, and they all moved at a slower pace, not a generically slower pace, but at a pace in which we're doing the relevant safety work? And how would we rebalance that equation so that everyone is doing a race to safety versus a race to power and capabilities? How will people start to see this impact their lives? I was on Instagram this week and I got delivered a surfing video and it was a video of Kelly Slater surfing in beautiful, pristine waters.

7:44And he was just weaving in and out of other surfers, maybe 40 surfers. It was an amazing video. And I looked at the comments and they were uniformly, you know, you just went down and they're like, wow, this guy's the goat. This is amazing. Wow. And then finally there was one comment and it was like, guys, this is AI generated. And I looked at this video and I'm pretty sure it was AI generated, you know, that's already happening. Some of it is very good. And it's not even a fraction as good as it's going to be if, as you say, this, this is exponential curve, as good as it'll be in a year, five years, 10 years from now.

8:20What are we talking about to lay it out for me, Tristan? I mean, are we talking about a world where nothing is real and everything is? I mean, it's like Aldous Huxley, again, on steroids, we just will not be able to even know if a call from our spouse is real. Yeah. So AI is not going to get worse at emulating someone's voice, someone's handwriting, someone's likeness. That's what generative AI does, is it gives you those capabilities. Anything that can be emulated will be emulated. And that's why it's generative AI. It's generating text, generating images, generating 3D models from scratch, generating architectural designs, generating movie scripts, generating amicus briefs, generating fake articles about people.

9:03Anything that can be simulated will be in New Hampshire. Someone had deepfaked Joe Biden and automated some robocalls with Joe Biden's voice telling people not to vote in New Hampshire. And I actually listened to it. To be honest, that one, I would have thought that that sounded like Joe Biden. And the point is that that's the worst that it will ever be. So if you're not impressed today with where it is, just look at the growth rate of how much better and how quickly it's getting better. And what can you do in the face of that? Well, you know, by default, yes, if we live in the world that we live in today, we won't know what's true.

9:36But I was just talking to the digital minister of Taiwan, Audrey Tang, and she was talking about the need for authenticated privileged messages. So like, anytime the government sends a message, it now comes through one number. If you get a text from the government from that number, you know, it's the government. If you don't get a text from that number, it's not the government. You know, Apple and Google could start working on an interoperable standard saying that we're going to verify and make sure that when there's a phone call, there's a real handshake, that there's a new secure encrypted handshake.

10:05It's just like we moved from, we went from the default on the internet being HTTP to HTTPS secure. So we went from kind of an unsecured, open, unencrypted internet to more secure and encrypted connections. I think that in the age of generative AI, we're going to move to these more privileged and secure environments. I just wonder how it's like, it's like putting a finger in the dike. Yeah. And there's more and more water building and it's just about to, that dam is going to burst. And I just think, well, already people have used chat GPT-4 to figure out how to break into passwords. I mean, even something as simple as every now and again, I'm sure this happens to you.

10:47I'm sure this happens to a lot of people listening. You get a text and it looks like from your bank and it says fraud detected on your account. And many people who aren't as used to getting these things might click on it. That's simple. But you know what I mean? I mean, it's just a matter of time. It's going to get smarter. It's able to just break through all of these systems that are designed to protect us. Yeah. Well, and this, by the way, I think is how we frame the way that we're worried about the risk, which is that we're just simply releasing more capabilities into society faster than society has the immune systems to absorb and adapt to all the new changes that accompany all of that AI getting released.

11:30You know, when the first time someone released that open source code that said, the AI that said, with three seconds of your voice, I can speak to your bank. Was every bank in the world prepared for that and planning for that like years into that? No, they don't know what new AI capabilities are going to be released. And that's just one tiny one. There's literally hundreds of them per week. It's hard to track. In fact, in our AI Dilemma talk, we quote the co-founder of Anthropic, Jack Clark, who said that unless you're scanning Twitter every single day for all these updates, you are missing updates that are critical for national security and sort of what it means to have a safe world.

12:06And so that's where you would say, okay, so why don't we just stop all this? Why don't we just not race? Why don't we just stop releasing all AI? Well, then people would respond to that saying, but if China doesn't stop, then the US is just going to fall behind. But I want to push back against this, which is not just that I think we should just stop in the US. We have to get smarter about what does it mean to beat China? Because if they race so fast that they release stuff that then undermines their own society, that's not in their interest either. We have to be smarter than that. The US has to lead and say, we need to set the terms of the race.

12:38And it's actually a race to the responsible and conscious deployment of technology that in its effect strengthens your society relative to other ones. That's the true competition. We're going to take a quick break, but when we come back, more from Tristan on the measures we could take to responsibly deploy AI and the role Tristan played in a recent White House executive order on AI. Stay with us. I'm Guy Raz, and you're listening to How I Built This Lab.

13:10If you've been listening to the show these last couple of episodes, you've probably heard me mention the Airbnb original experience that I'm hosting. And I'm so excited to have a chance to connect with listeners in a completely new way. It's called the Reinvention Lab, and it's designed to help you think about how to unlock your next big move in your career or even in your life. I'll help you discover your own story in ways I do on this show with my guests. And attendees will get a chance to take a deeper dive with me on so many lessons I've learned from this show. Lessons that have transformed how I work and think about the future.

13:49All proceeds from the event will go to support the Ronald McDonald houses that helps families stay near their children who are being treated at nearby hospitals. It's going to be really fun and I can't wait to meet you. So come join me in San Francisco and take your idea to the next level. To grab your spot, visit airbnb.com slash guy. It's going to be great. Ever had one of those afternoons where your brain just quits on you? You're sluggish, hangry, maybe even a little foggy. What if it's your glucose? See, glucose is an energy currency for your mind and body. When it's stable, you're on point.

14:27When it crashes, so can you. That's why Lingo is so interesting. Lingo is a glucose wearable designed to help you connect the dots between your glucose and what you eat, how you move, and how you feel. It shows your glucose data in real time. Instead of guessing, you see the impact of your choices. Maybe that healthy snack is actually sending your glucose on a roller coaster. Or that afternoon walk is the perfect stabilizer. It's about unlocking your consistent best all day long by truly understanding your body's unique responses. Get to know your glucose and learn about how to build healthy habits that work for you with Lingo, designed for you by Abbott.

15:07Through November 30th, use code GUY10 on HelloLingo.com to get 10 % off a Lingo plan purchase, one use per customer. This offer cannot be combined with other offers, U.S., Puerto Rico, and U.K. only. The LingoGlucosystem is for users 18 years and older, not on insulin. It is not intended for diagnosis of diseases, including diabetes. Individual responses may vary. It seems like every day we hear about how AI is coming for our jobs and people are getting anxious and fearful. But a recent survey from Miro reveals that 76 % of us believe that AI can actually benefit their role, while 54 % struggle to know when to use it.

15:49Well, enter Miro's innovation workspace. It's an intelligent platform that brings people and AI together in a shared space to get great work done. Miro has been empowering teams to transform bold ideas into the next big thing for over a decade. We've recently started a new project using Miro, and it's a really awesome creative partner. It helps us develop better processes for organizing. It even generates new ideas. You don't have to be an AI master or toggle to yet another tool. The work you're already doing on Miro's canvas is the prompt. Help your teams get great done with Miro. Check out Miro.com to find out how.

16:31That's M-I-R-O dot com.

16:42Welcome back to How I Built This Lab. I'm Guy Raz, and my guest is Tristan Harris, co-founder of the Center for Humane Technology. And Tristan, as you've heard, has been sounding the alarm about the rapid development of AI. And he says that advancements in the technology could unravel the very fabric of our society. Societies, human societies, depend on, more or less depend on a sense of trust, that there's common information. And even if there are, you know, differences of viewpoints and so on. I mean, you can trigger riots, conflicts, violent demonstrations with misinformation that is so credible that seems so real.

17:25I mean, videos. You know, I keep thinking about this movie, The Running Man, that came out in like the 80s with Arnold Schwarzenegger. Have you seen that movie? It's vaguely familiar. Remind me of the plot. So basically, I think, if I'm recalling correctly from my childhood, Arnold Schwarzenegger is a Bakersfield cop, and he's a good guy, but he's disliked by his superiors or his colleagues or whatever. And there's a scene where they're all in a helicopter, and there's an anti-government riot, and the helicopter fires on these demonstrators and massacres them. And they essentially frame Arnold Schwarzenegger, who tries to prevent the other pilots from doing this.

18:01They frame him as the guy who did it and they create a video where it looks like he is the butcher of Bakersfield. And so he's sent to prison. But you can imagine that future. I mean, it's so crazy that that film, you know, what happened in that film could easily happen. You can imagine the court of law, you know, documentary evidence being presented. Yeah. Video evidence being presented, signatures of our documents, photographs, recordings, all of these generated by AI that are so good, it's impossible to discern from real evidence. And again, I'm on a rant here, but it seems like this is going to completely upend how we think about communication, what we believe, what we present as fact and evidence, how we function as societies.

18:53Yeah, 100%. I mean, here's a metaphor. Imagine that the whole world is run on top of Windows 95. You know, it's running the world's computers, and everything in the world runs on Windows 95. Governments run on Windows 95. Banks run on Windows 95. Hospitals run on Windows 95. Legal documents, court cases, lawyers, it all runs on Windows 95. And then imagine one day, someone publishes this code to the whole internet. And it basically teaches you how to hack any Windows 95 computer in the world. So now Windows 95, which runs the world, is not secure anymore. It's insecure. So in this metaphor, the way that our whole society has been constructed is like sitting on top of this box called civilization 2000s, right?

19:42Like we've sort of living on a early 2000s world stack of the assumptions of paperwork and signatures are actual signatures and photographic evidence is actual photographic evidence and people's voices are real and can only represent their only voice. But suddenly we just undermined collectively with AI that set of assumptions. And so what do you do when this happens? Well, you don't try to pretend let's all keep running the world on Windows 95. This moment with AI is forcing a kind of rite of passage. Humanity has to kind of go through a bar mitzvah or bat mitzvah to upgrade the systems that we have been relying on to accommodate the new assumptions.

20:23And we've done those upgrades before. In democracies, you know, when you said the printing press came out, the printing press both killed the previous forms of government, of feudal governments, and it made way for democracies. First, through a really unstable period. And then ultimately, you could have public education, you could have the fourth estate and news articles, it forced this reorganization of what kind of governance that we need to live in. We are in this uncomfortable, but we have to do it adaptation period, where we are needing to upgrade the basic legal philosophical mechanisms, we have to come up with new meaning for what is evidence in a world where AI can generate that evidence.

21:02And there are ways of doing that. We could live in a world where the only places that any media you see on the internet will only be on the internet if it's watermarked because we know that it was real. So there are things like this that are the building blocks, the puzzle pieces of this upgrade, but there's about a million pieces that has to happen. And I know that can sound daunting to people, but I almost want us to be collectively saying, okay, we're going to hold hands together and we got to go through this transition. And yes, it's going to be a little bit rocky and we have to make these changes together.

21:32I know that a big part of what you do is just creating public awareness, but you also went to the White House to help put together an executive order around the stuff late in 2023. Tell me what that order actually, in practical terms, will do. Will it slow down this process? Will it actually create actionable protections for us? Or is it just, I don't know. I mean, again, it's an executive order. It's not a congressional law. What does it do? Well, so there's multiple parts to answering this question. So, I mean, this is, I think, like 111 pages. It was done in record time. In six months, it touches algorithmic bias, so AI that's used in current AI systems that are biased and how to deal with those issues.

22:18It deals with AI and biological weapons and needing to lock down the supply chains for where people can get dangerous materials and saying, you know, we need to handle that better. It deals with AI and the next GPT-5 and GPT-6 systems. It says that if you train a system that uses more than 10 to the 26, I believe, flops or floating point operations, too technical jargon, then you have to notify the government. That's basically like saying, if you're building a nuclear weapon that's really powerful, the government has to know. But to your real point, your real question you're asking is, what can that executive order do?

22:51Because it's not law, it's an executive order. It's not legally binding for making sure that all the companies have to do all these things. A lot of it is changing what's called federal terms and conditions. So to get federal funding, if you're a biology lab, you will not be able to get that funding if you don't do these new sort of protective measures for the dangerous biological materials. So what that's doing is using the leverage of the government and its funding power to start to incentivize different aspects of the supply chains of the world, educational environments, banks, et cetera, to do more of the things that are AI resilient.

23:23So think of it as a movement and a signal, like a big bat signal, blasted into the sky that says the US government is taking AI seriously. Now, it's not the security blanket that suddenly makes the world safe or suddenly open AI and anthropics, stop everything they're doing. And before they study or do more research, they look at the executive order. They're still racing to build AI as fast as possible. And we mentioned the nuclear metaphor. How did we get to nuclear proliferation safety? How did we get to nuclear non-proliferation controls? There was also back then a lot of track, what are called track two dialogue.

23:54So informal conversations between American nuclear scientists and back then Soviet nuclear scientists about basically making sure that we had safer controls on nuclear weapons. They couldn't accidentally go off. I'm happy to report that informally, there are some of those dialogues that are happening between Chinese AI scientists about the risks and US AI scientists about the risks. As I say all this, is this adequate to where we need to go? No, it is not. It is a small drop in the pond compared to what needs to happen. What we really need now is for people to demand from their lawmakers that we take these issues seriously.

24:31And I think things like liability as a regulatory framework are powerful because people understand it, right? You as an AI company shouldn't be worried about being liable for the harms if there are not going to be any harms or risks. So if you don't think there are risks, then go ahead and release it. But if you do think there are going to be risks and you're liable for them, what that does is it has everybody move at a slower pace at the pace that we can get this right. We're going to take a quick break. But when we come back, what Tristan thinks it'll take for the world to unite against the dangers of AI and how he stays motivated in the face of such an enormous challenge.