What Happens When the AI Bubble Pops? - With James Li

0:00 And these are the people that are supposed to be the most powerful people in our in our world. And yeah, the problem is too is that they don't really have much allegiance to America, do they? No, it doesn't seem like So, what what do you think, you know, is going to happen here? You know, what does the collapse look like? It does seem like there will be some kind of realignment collapse, whatever you want to call it. But what I would say it's from the ashes? I think that we can always look to the past, right? And so, probably the closest in our lifetime analogy that we could look to would be the internet. And we know that the internet bubble had

0:45 happened and it wiped out like 95% of all the companies. But the companies that stayed like Amazon that weathered that storm uh became one of the biggest companies on the planet, right? And so, uh here we'll see something like that where it's all going to go. They're either going to all merge together. Everything that they do is predicated on, well, if we don't do it then China's going to do it first. But China's nationalized it and they've put the resources in. And instead what we have, instead of five companies united in one pursuit, instead they're five companies that are up until like 2 weeks ago were in-fighting and all just trying to get better than each other. Instead of coordinating their

1:30 efforts, they had a lot of redundancies from a business perspective. So, um I think that China becomes the boogeyman. And I think in that pursuit, they're uh they're going to nationalize AI. Um but still make us pay for it even though we're going to pay for it regardless. I also think that quite soon everybody is going to have their own AI agent. I think that the future of the internet, we've talked about this offline. I think the future of the internet is agent first and that's going to fundamentally change how we interact with the internet, how commerce is done over the internet. I think it's it's going to completely change everything and I think that America is going to own its AI and

2:18 it's going to be one unified thing and I could see that by like spring. I could see like the unification or nationalization of AI in totality by like spring, summer because I don't think they I don't think they have enough runway. I think the problem is is for a lot of these people is that there's not enough there's not enough time. Like why why is Peter Thiel going so hard on investing in all of this AI stuff, all the surveillance tech stuff? Well, it's because he's like 60. The dude's like literally got like 20 years before like if he's brain doesn't upload in the next like 20, 25 years, he misses the thing that he's been building

3:04 his whole life for. You know, so I think that there's some of that kind of stuff going on as well. But I don't know if it if it goes well. I think it goes well if we have if we open source the internet, if we kind of kill SAS companies and all that, like I think that is going to be a tough thing, but that's ultimately what's best for all of us and then we could have our own sovereign AI usage where we're not giving everything, every bit of our lives to some corporation that's probably run by the American government from in some way. Um but yeah, I don't know I don't know where the I don't know where it goes,

3:50 but what I do what I do feel pretty confident in saying is that I don't think AI is going to kill us all. I don't think that. We could just yeah, turn it off basically. Cuz they say all these chatbots are villains. I just read a thing. I don't know if it's true or not. They had to cancel their most recent version update of what is it chat GPT and and Anthropic because the agents were too villainous or something like that. I don't know what that means. You're more technologically savvy. I don't really know exactly what that means. Like the agent is doing crazy Um but you do have painted sort of a rosy picture. It's like oh everybody has their own agent nationalizing there's there's some benefit to that I think. I don't know

4:36 if if it will turn out that way. I just think there's going to be too much ego or the incentives won't align in a way where these AI companies are going to do what's good on behalf of people or even you know national security or or whatever. I I just don't have the faith that that Oh, I have no faith in that. I have no faith in that. I just think that AI isn't this rogue thing with goals that's going to like permeate through everything. The humans are going to create the problems and are creating the problems with AI. AI isn't creating it could it ever be a situation like the movie Space Odyssey where the AI computer Hal because of the way he's programmed he's like well, you humans are things up. I now will take

5:22 action to basically terminate people who are against my you know mission um whatever he was programmed to do. Yeah, I think again that would mean that you would have to put some sort of human emotion behind a machine and I just don't know if we can ever get there, right? I don't think we'll ever get there at all. I I So, then you couldn't program an AI to be like, all right, let's say your goal is to maximize profit for this company, right? You have this AI that's going to find all kinds of ways to make things more efficient. That's the ultimate goal. And let's say a human

6:07 would come in and say, well, no, we actually, you know, in some cases we don't want to maximize in in in this way. Let's say there's some human costs to it, and the AI will be like, sorry, that's not what I was programmed to do. Like, we have to do [clears throat] this. You think that's something possible? Uh maybe. I think that Cuz it's not emotion-based. It's just mission-driven. Yeah, so I think that I think that that's actually more likely. So, like, when we talk about like, you know, AI doing random things, it's usually because you set it on to a task, but you didn't define exactly how you wanted that task to be achieved. And so so like, for like example, be like,

6:54 AI, eliminate all war. And so, how could it do that? Well, if there's no more humans left, well, then there's no more war, right? Like, but if you say, hey, eliminate war by but like you have to have like you have to weight these things as like what what needs to be done, things that you can't you and you set the the guardrails. I think if you do that, even with an advanced AI or a super intelligent AI, which we haven't gotten to yet, Trump, um I think that that may be um that may be uh more likely. But I think that we're going to get better at prompting and guardrailing how like like what AI is

7:40 capable of. But also, it's just like part of the issue is that people go like, "Oh, I'm like not good with computers." And then they ask ChatGPT to do something, and then it does it, and then it goes, "Wow, like I can't do that thing, and it did that thing with such ease that therefore it must be smart or this or that." And I just don't think that that's the case. If you really get into AI like I do, um you realize how stupid it is on like very simple things, too. It can do very complex things, but it can't do very simple things. I just think we're not there yet, man. I just think we're not there yet. Well, it's advancing quickly. That's that's the that's the worry I think a lot of people have, right? It's like we

8:26 don't really know, and there's no Like I'm trying to think of a way where this would make sense, and how to regulate this, if we need to regulate this. For me, you know, it's like technology I think it's it's there should be some like safety factor involved of like can this actually like cause harm to people, right? That's how you you know, you regulate new technology. Like you're you're you're building let's say there's a new machinery like that was invented in the industrial revolution. Like [clears throat] then there needs to be some kind of like safety regulations that can this thing like blow up or whatever and and kill people. If it does that, then you know, you need to install fail-safes. The same with like building an airplane or something like that. Where this is a new technology, you want to make sure it's safe. Do you

9:11 have backup systems where if this, you know, primary thing fails, then you have like a secondary thing that'll kick in. Stuff like that is where it would make sense to me, but I just don't know if like the people in Congress like really understand this at all, like what it even does. And then, you know, so then you end up having people who make this technology self-regulate, and they say, you here are the guardrails." But then, we always see a problem with that same thing to use the airplane example, when Boeing starts self-certifying their airplanes, things can go wrong because they are incentivized by profit and speed and efficiency versus actually, you know, human life. Because if a plane does crash, they'll most likely, you know,

9:57 have the insurance kick in or they'll blame something else, and they won't actually have any accountability. Yeah. And that's the same with the AI companies. It seems like if something does go wrong, are they going to actually be accountable for it? And you kind of see that now, I think, with maybe like the the cases being brought up where people are using AI and then killing themselves. I don't know what the right solution for that is. Are they really at fault? Or was the person, you know, they're saying ChatGPT convinced them to kill themselves. It's like, is that ChatGPT's fault? I don't know. I don't I don't think it's ChatGPT's fault necessarily, though there should be because it is I mean, you can absolutely put guardrails on the models. I had interviewed a

10:44 person who actually trains models for one of these companies. Actually has worked with almost all of them at this point. And and she assured me that like, "No, you can definitely put guardrails up." I think that there's definitely that. It's just there's too much at play here, too much uncertainty. Because we can't even understand yet what AI is. So, therefore, how can we like control it? You know, I don't know how much control there can be. Um, but I think that the problem is with AI is with anything else. It's about the people who own it, regulate it, what the laws are around it. And that's what we got to That's what we got to focus on.

11:30 Is we got to focus on on on that. You know, I think I'm I'm against the open sourcing or or the taking down of open source, closing all of that. I'm super against that. The internet itself was built on open source. I mean, you know how many critical systems run on Linux? Linux has literally never been paid for ever in like 30 years it's been around or 35 years. So open source is the way. That's the open and free internet. That's that's like what we need and we should we should be down for that. We need to we need to embrace technology. As we as we do. That's how we advance.

12:15 And if we don't embrace it, you know, but like do we have something that people can embrace right now? Well, I think it's you know, it's People don't trust the billionaires or the people in power, the government to apply these technologies in a way where it will protect people. I think in an ideal world, and this is kind of what's going on that I've read in China, there's not a huge opposition from the population against AI. They're actually in general very pro AI because they trust the companies or the government for whatever reason going on over there, but they trust them that they're going to have the people's best interest at heart. So they're like excited. They will go and watch like robot fights, you know, as an

13:01 entertainment. And for us, I think when I see that, I see dystopia. I see like, oh wow, this is like, you know, basically robots replacing humans. But is there anything inherently wrong with robots fighting each other? Not really because it's kind of fun. So it's like, can we get to a place here where we trust the people and there or there's some kind of accountability mechanism? Because I think what goes on in a in a dictatorship, and I'm not necessarily pro-dictatorship, but this is just a reality of a dictatorship is that there's less place to to to to to apply blame to. So, if if AI goes, you know, crazy or something bad happens in China, the government has to take ownership of it because there's no other people controlling. Here, we have too many like

13:47 quote-unquote like sources of of power where people start blaming each other. Republicans blame Democrats, Democrats blame Republicans. Then they blame billionaires, whatever. So, it's like everybody's pointing their finger at each other, and nobody wants to take accountability or responsibility. Um so, I think that's like my main concern is that like can we get to a place where we actually trust the government? And this comes not even just AI, but in everything else in general, public health, whether or or or any kind of application of criminal justice. Nobody trusts anything. That's why I think, for example, the Charlie Kirk thing has taken off to a degree where they're like, "The FBI's corrupt. They're lying to us. Everything they give us is a lie." So, now you have people doing their own investigations,

14:33 which is fine. Other issues, you know, whether it be like COVID or you know, health-related stuff. Is the food being poisoned? That stuff it really should be handled, you know, in a in a functioning society by government. Uh just to make sure like, "Hey, we don't poison our food." That should you know, that's pretty Look at what's happening right now. simple or you know, it should be agreed upon by everybody, but now we're not sure. And so, I think that's that's going to be a problem that I I don't know how to solve. I don't know how to And it's not my responsibility to restore trust in institutions. It's the institution itself that has to reform and gain people's trust back. I think that they think the only chance that they have is AI being so omnipotent

15:21 and so ingrained within our system that we don't need to trust the government anymore. We're going to trust the AI. This is a lot of like, you know, in a in the AI, you know, world they go solve AI first and then AI will solve everything else. This is what Sam Altman has said, this is what Google DeepMind has said, this is literally their their thinking.