The Dangerous Truth About AI Regulation | Bill Gurley

Through OpenAI and Anthropic, investors are rewarding them with the highest valuations we've ever seen in the history of the venture capital world. When people get rich quick, you invite in charlatans, interloper speculators. Like that's going to happen, and people have already gotten rich quick here. I'm not sure there's ever a finish line to this.

You know, Daniel Pink wrote a book about regret. And he said in there that there's studies across the globe, not just in the US, across different cultures, different countries. And as people get older, the number one thing that kind of weighs on them is what he calls boldness regrets or regrets of inaction. And so it turns out humans are really good at forgiving themselves for making mistakes after the fact. They like rationalize it, 'Oh, I learned something, whatever.' But the thing they never tried kind of is always out there weighing on their brain. And so it's a bit of a behavioral science problem where when people are young, it's easy to put off the thing that seems a little too bold. And then as you get older and you have more restrictions on your time and your money, then it's harder to even consider that you could do that. And so it never gets tried. I think it's worse today for the young people because there's so much, or at least pre-AI, there's been so much pressure to get kids into the right lane, into the right school, into the right career. I think it's well-intentioned, you know, like they're worried about the economic stability of the young adult, but we've lost the time that, when everybody back in my day, but people used to run around a lot more and explore and discover. And I fear that we've traded that for more of an industrial education complex that's putting people in places that were perceived to be safe but lack this essence of what I refer to as fascination, which I think is an unbelievable superpower in a career.

And they're remarkably overscheduled. Jonathan Haidt calls it the resume arms race. But it's so hard to get into college these days that starting in sixth grade, you know, parents are pushing kids into Mandarin lessons and cello lessons and lacrosse lessons and volunteering. And like they're very overscheduled. The young kids are way more overscheduled than we were when we were growing up. And in some sense, they get taught to grind, but that'll burn out, you know, that'll eventually burn out. And then the other thing, I don't have any analytical study on this, but I do think somewhere around 19 or 20, you get worldly enough to start thinking about possibilities. And it used to be that you weren't allowed to pick a major until the end of your sophomore year at college. So, you know, you're 20 years old and they wouldn't let you. They just said, 'No, you can't. You have to wait till that point in time.' And I think everyone knows why that happened. But today, at least in the US, many schools, you have to apply to a major in your original application. You're applying to a particular program within a school. And so, you're taking a 20-year-old decision to a 17-year-old decision. I mean, I don't think it's one thing. I think it's all those things.

Yeah, there are a lot of people out there that are like that. And I think it's probably growing, especially in the US. The polling of AI anxiety is really negative in the US right now. And yeah, I think there was a wave of congressmen, especially at the state level, who feel like they didn't move fast enough on social media and they're very afraid of a TikTok-like conversation with young people that kind of becomes overwhelming. So there's a lot of brake pumping happening in the US. I guess I'd tell people two things. One, if you're so reluctant that you're not curious about how you might be better at what you do every day, then I would go back to square one and read my book. I think you're in the wrong job because everyone I've met who's really fascinated with their career and super curious and excited would relish the idea of being able to go deeper, faster, and know more. So that'd be my first thing. And then look, with that great book 'The 7 Habits of Highly Effective People,' he talks about your circle of influence. Like you're not going to stop AI, and you didn't stop the personal computer or the internet or any of these things. So sitting around having anxiety about something that's out of your control is not a very high-agency decision.

Here, I mean this one's kind of fun because it relates to the book. When I was finishing the book, the editor asked me to write the conclusion. I hadn't done it yet and I spent all weekend and I turned it in and I had done a classic summarize the book thing and he hated it. It was his least favorite piece of writing and he was very frank with me. So, I went back to the pro version of OpenAI and I said, 'Tell me 10 amazing concluding chapters in a non-fiction book and why it worked.' And it gave me a list of 10 books. And eight of the 10, the author had decided to make kind of an orthogonal move in the concluding chapter. Rather than summarize, it had reflected back in a different way. And once I came upon that, I then asked AI, 'Well, tell me a bunch of books that have done that even more.' And a whole bunch of ideas popped in my head. And then I rewrote a different concluding chapter that wasn't a summary. And the editor loved it. And so that was like a learning process for me that would have taken, I mean, if I had read, I'd have to go read 400 non-fiction books to get to that place. But it's just an example.

Well, I think it's interesting the question as you talked about it because there's this amazing book by Carlota Perez where she makes the point that technology waves, successful technology waves and bubbles always happen at the same time. And when it's phrased the way you phrased it or someone phrased it to you, there's an implication, 'Oh, people are spending too much money, therefore this must not be real.' And the fact that they're willing to bet that much money is precisely because it's real. Could they go over the top? They almost always do. And you know, you go back to the gold rush, like when people get rich quick, you invite in charlatans, interloper speculators. Like that's going to happen. And people have already gotten rich quick here. And so you're going to see a fury of investment and speculation, not all of which will be successful, but it doesn't mean, you know, and there's these great articles about the internet boom and bust where everyone made fun of all these companies that went out of business in the bust and then some new version of that company emerged, you know, 10 years later and was wildly successful. So the ideas were directionally correct. We just got over our skis a bit and I think that happens every time and will happen here. But I don't think the implication that the question had loaded at the beginning that and therefore this wasn't real. I think it's the opposite. It happened because it's real. Does that make sense?

In the US, two players have emerged that are investing tons of money in the model business, which are OpenAI and Anthropic. Investors are rewarding them with the highest valuations we've ever seen in the history of the venture capital world.

Like they're both close to a trillion now in valuation as a private company. Unheard of. Like unheard of before and they're raising over 100 billion as a private company. No, that's like by a factor of 20 or 25, you know, versus anything we've seen before. But those two have kind of emerged as the leaders. Google and Meta to a certain extent, and xAI are laying chase but it looks like a two-horse race and it's all proprietary which means it's just classic business where you keep everything internal even though they started a different way. Almost all the players in China are open-source which means they share the information about how their technology works with the world and that technology wave is something that started about 25 years ago. Linux probably being the most famous example which is the most pervasive operating system in the world today and open-source and there have been 50 other open-source software projects that have been wildly successful and there's reasons different people do that but the Chinese government starting with the, I think, 14th five-year plan suggested it was a good idea to embrace open-source so they encouraged the companies to do this. I personally think that given enough time because they do have less, they don't have access to these best Nvidia GPUs but given enough time a whole bunch of players using open-source models I think will catch and pass a world where you just have two proprietary models just based on what I've seen over the years because they can all learn from one another. But we'll see. I mean, I'm perplexed at how much attention is spent in the media and in Washington with this notion that, 'Oh, America has a three-month lead and therefore we need to restrict this and invest here because we got to win.' And like if you're three months ahead, like it doesn't seem to me like you have that big of a lead. And then they start talking about, 'Well, it's going to become super intelligent and program itself and it'll escape exponentially.' And I'm like, okay, that's a bit of a stretch but anyway, it is two very different approaches and we'll see how it plays out especially in the rest of the world. I think America's accustomed to winning in Europe and South America and non-China Asia with their products. I think they're accustomed to that and that may not play out that way here.

Yeah. Let me answer the question by telling you what other people say. They think that these systems will eventually become super intelligent, start solving our biggest health care problems and then potentially be used as part of government and part of the military. And it's with that mindset that they say we have to win the AI race. Having spent a ton of time in China, I know how effective and intelligent their entrepreneurs are and creative. And in many fields they're ahead of us in EVs and solar and batteries. And so the idea that we're going to develop this technology and have some kind of permanent massive advantage over them that then has some implication in how the world is run or managed, I don't buy it. But I'm not being called upon to offer that voice in Washington or whatever. But there are a number of people that do buy it. And the China hawk message resonates with some of the American populace on both the right and the left. It's not, you know, it's a kind of a bipartisan tool that's used or fight that people want to get behind. So, it gets attention.

I don't think there is. There's a wonderful book, I might have it laying around, called 'Finite and Infinite Games.' And it suggests that most of our game strategy we learn from playing games that have a beginning and an end, a game of tennis or soccer or whatever. And many things are infinite games, which is, you know, the point you just made like it never ends. So, it's like to say 'I'm going to win the AI race,' it's like, 'Oh, we're going to win the nuclear race.' Well, when is the official going to blow the whistle and declare the victor? Like, it doesn't happen. And the idea that you're going to keep a nation like China from using AI and developing AI, I think it's foolhardy to think that that's even possible. And we might do better for all of our citizens by focusing on what can we do better ourselves than trying to say how can we prevent them from doing better. It's kind of an odd mindset from my point of view.

One easy answer is, you know, look, that's the history of Silicon Valley and how these things work. Kind of how did Yahoo and Microsoft let Google emerge? You know, these things, it's kind of constant disruption and innovation. There's another element that is new in the American venture capital ecosystem where there is so much money available for the winners because some of these investors have institutionalized and gotten really big and they've come to believe in power laws and network effects. We could explain what those are if you think your audience needs to know, but it means that in technology often times, you know, once someone gets a lead, it becomes self-compounding and they get bigger and bigger. So now you have investors that are confident investing ahead. And what this means is these companies are raising and burning more money than an incumbent would be comfortable doing. And that's part of how they got the lead. I mean, let's talk about Google, for example. You know, for many people, the starting place where they type in a question used to be Google and now it's, you know, OpenAI or Claude. And you could say Google should have done everything in their power to prevent that from happening. But they had one of the most amazing cash flow printing machines in the history of the world running. And to make sure that that didn't happen, they would have had to put that at risk. And that's a hard thing for a management team to get behind. And they didn't know what the consequences would be. If they could run it twice, which is a poker phrase, like if they could run it twice, they might do it differently at this point, but we kind of are where we are.

Oh yeah, I did read that.

All those things are true. Yeah. When you grow headcount in an organization from one to 100 to 1,000 to 10,000 in rapid time. It's almost impossible not to have chaos. Like you just think about what that would mean. Like there's new people every day. You don't know who they are. People are getting assigned management roles that have never managed before just because it's growing so fast. There's no way for that not to happen. And so, yes, and when the burn rates, you can't give somebody $50 billion in venture capital. They can't go use that effectively without taking on a massive burn rate. Which is another thing that can kind of lead to chaos because it's hard to know what your real unit economics are once you get comfortable burning a billion a year, 3 billion a year, 10 billion a year. Like what's the constraint that causes you to not do this or to do that? And like it gets... I think I've never heard anyone describe it the way you did, but now that you have, I think it's a good lens from which to consider.

Yeah. Well, if they ever run out of money, it'll be abnormal also. That is true.