Joe Rogan podcast. Check it out.
The Joe Rogan Experience.
Train by day. Joe Rogan podcast by night. All day.
Yeah, I was listening to Tim. First of all, hello.
Good to see you, my friend.
We were listening to Tim Dillon. I was listening to him on the way over here and he was talking about Anna Paulina Luna and Tim Burchett and Trump. They're all talking about the UAP disclosures and like why now? Like what are they doing? Like why are they distracting us with this? Burchett said that whatever they're going to release, it will be indigestible.
Indigestible as in? Or well then it doesn't mean that it's real then.
Well, I think it means that it'll be so crazy if it's real. So crazy. He's the one that's been saying that there's these confirmed bases under the ocean, that there's these specific locations. I think you talked... you're shaking your head. You don't believe a word of it.
I think... look, it's completely implausible that there aren't other species, right? Completely implausible. Just the vastness of what we're dealing with. So the real question is like why haven't we encountered people or those things, those beings, right? And it's probably because they just have bigger fish to fry, you know? So by the time that we meet them and they meet us, we're going to kind of be at the edge of like we've kind of been there done that on our own planet and then we've kind of like developed the technology I guess to get beyond it. But somewhere along the way there must have been a few just mathematically impossible. So then the question is, is it buried? Or were people confused when it first came here? Like if you had a spaceship land in like the 1800s, right? What would people have done? They would have just freaked out. They wouldn't have understood it. Maybe they would have buried it. Depending on where it was, maybe they started to pray to it, right? And you would have just moved on. And then that isn't documented in history. So...
It is. There's a lot of it documented in history.
Oh, you mean like hieroglyphics and like monuments?
Well, the book of Ezekiel. The book of Ezekiel goes in depth about some sort of a UFO encounter that Ezekiel experiences, right? Where it's a wheel within a wheel and a cloud with fire flashing forth continually in the midst of a cloud as it were gleaming metal and from the midst of it came the likeness of four living creatures and the creatures darted to and fro like the appearance of a flash of lightning. This is all in the Bible. It's also in the Mahabharata. They talk about vimanas, these flying crafts. And I think it's entirely possible that we have been visited periodically and that we have been monitored and that we are monitored currently.
I agree. And if I was going to hide, I would hide in the ocean.
Well, to be honest, as I get older, I'm convinced we're basically in some form of a simulation. There's like all these little ingredients that if you start to see these little clues, you're like, they all seem so odd in isolation and then when you put them together, I feel like a crazy person. So, I ignore myself, right? But I wonder like why did this happen? Like yesterday, I was at a dinner in LA before I came to see you. And I told this very interesting story. Well, or I thought it was interesting at the time. You know that like so in 2000, right? If you think of like what happened in tech since 2000, so the last 26 years, people can give you all kinds of like fancy theories, but there's just like this weird word that's been at the center of every single technological revolution for the last 30 years. And that word is attention. Let me explain this to you. Google, they invent Google. What is Google? Google is an algorithm. It's called PageRank. But if you look inside of it, what is it? It says, 'Well, Chamath's website has five links to it. Joe's website has two links. He's getting more attention.' Okay, Chamath's website is more important. That's the sum total of Google. Now, they've made that a lot more refined and they've done all these other fancy things, but it's all about attention.
Fast forward to 2007, 8, 9 when you know Zuck and then when I went to work for Zuck and we got on the scene and we're like what does everybody care about? Attention. And so what is like the Facebook algorithm? What's the Instagram algorithm? You know how did we construct newsfeed all around attention? Joe had 35 likes. Jamie had 12 likes. Your thing is more important. Let's give it more importance because it's seemingly meeting all these human needs. Attention. Attention. Attention. So phase one attention, phase two attention, and this is where I'm like, how can this be possible? In phase three, we're like looking at AI. And when you look backwards four years, the seminal paper is called 'Attention Is All You Need.' It's about this word again. And when you look inside of the core part, if you peel apart AI, the little brain that makes it so capable, it's called an attention mechanism. It's just attention. It's all about again this idea of I'm going to scour all this information and I'm going to figure out what patterns repeat itself and I'm just going to double down on the stuff that I see more of because that attention must mean it's more important. It's more true. It's more knowledgeable. And then I think how could it be like we're all like why is it that these things are just repeating over and over again? And I just get confused. I don't exactly know how to explain it. So are there other ways in which we should be doing things? Absolutely. Have we even explored it? No. So then I think, well, is this just a simulation? Some kid in a basement in his house just playing some simulation and we're all just party to it and that's all he understands is attention. I don't know.
I don't think it's that simple that there's a person playing a game. But if you break down just attention, well, that's all of human history is paying attention to the king, paying attention to the war, paying attention to resources, paying attention to who says the thing that resonates the most with the people. It's all about what human beings are paying attention to.
I think it's part of it. Then there's also what is actually true. And I think sometimes what is true and what people pay attention to are not the same thing.
And sometimes the thing that you should be paying attention to gets lost because the thing that you are paying attention to gets more attention because it's more interesting and useful. That's sort of where we are right now. We're in this really weird phase, I think, where you actually should be focused on this thing over here and instead we're all focused on all these things over here.
Um, here's like a very big one. I think like it's pretty fair to say since the last time you and I saw each other on this show, the attitude towards technology, I think, has been pretty profoundly negative. It's kind of tilted. It's relatively like anti-AI, you know, anti-billionaires. It's anti-all of this stuff. And it manifests in all of these interesting ways. There's protests, there's data centers, there's all of this stuff that's happening. People are worried about job loss. All of that stuff is real.
No, I'm okay. I'm okay. But what should they really be focused upon? And I think what they should be really focused upon is we're at the tail end of a cycle that doesn't work anymore, which is all about like this tension between labor, people that do the work, and capital, the people that fund it and then make all the returns. And over the last 40 years, we've basically gone to this completely upside down world where capital extracts all of the upside and labor has extracted less and less and less. And all of this pushback, it manifests in AI, it manifests in politics, it manifests in social issues, it manifests in, you know, Israel, whatever you want to talk about. All of these issues, I think symptomologically, come from this other issue, which is we are out of balance. This total compact that we used to have, a liberal democracy and a free market has totally collapsed. And there are simple ways to fix that, but that never gets the attention because it's not what you want to talk about. The attention is here, you know, vote no to the data center. You know, this model is going to take out all the jobs. You know, this social issue is really important. That war should not be fought. That war should be fought. All of these things while important distract us from what the core issue is. And the core issue is that we as a society, I think, are out of balance. The natural compact between all of us is broken and there are some simple ways to fix that compact. Get people more invested, get people more engaged in the upside, have people have a positive sum view of what's happening and that isn't happening.
Well, what simple solutions are there to this one very particular issue?
Okay, I'll get your reaction to this. Let's assume that you still lived in California because I think it tells this example in a more extreme way. Okay. Let's say you make a million bucks a year, which is a lot of money, but it makes the point more cleanly. You'd pay, I think, 30% federal tax and you'd pay another 15 or 16% in state tax and Medicare tax and all this tax. So, if you're a wage earner, 50% of all your upside goes to the government. If you're a capital earner and you make that same million dollars via capital gains, you pay half that tax. Why did that happen? That happened because in the 40s and 50s, but really in the 60s and 70s and 80s, what we were trying to do or what the American government and what western societies were trying to do was to convince people to invest their money. Hey Joe, go build that factory. Go hire those people and we're going to incentivize you to do so. And by doing that, there was this idea that all of those profits that you would get would then diffuse, right, trickle down into everybody else. The workers participated, everybody participated. But technology allows you to do more with less and less. So now what happens is the capital owners can accrue infinite almost it seems like value and the workers get less and less. But now if you get less and less and you're taxed more and more as a percentage of what you own, you're going to feel really out of sorts. You're going to be like, why am I paying 50 cents of every dollar? And I see these other ways where folks are paying 25 cents on their dollars, but their dollars are compounding way faster and they have, you know, hundreds of billions more of those dollars than I have of my dollars. If you take that example and you expand it across society, I think people understand that now. There's enough information and there's enough people talking about it where it's pretty clear that that's happened. So the question is how do you fix it? I think like if you think about AI and if you believe that we're going to get into this world of abundance and we're not working, what does it mean for governments to tax our labor? There is no labor. You're not working anymore. I'm not working. We're doing things out of leisure. Why should I pay 50 cents of every dollar? Why aren't the companies that are going to be making trillions of dollars, why don't they pay more? Why isn't there, you know, an expectation that they then help our lived society do better and thrive as a result of all of that winning? That's the real conversation that I think is bubbling. And I think that we're probably another 12 to 18 months where all of these other issues are going to be important, but they're going to be viewed for what they are. They're going to get demoted I think in importance and it's this core structural issue, it's what is the economic relationship that we have together as a society, what is the relationship between Joe, Chamath, Jamie and all these companies and how do we feel about a few and an ever-shrinking few making more and more and more and then how do we feel about their ability to share that with a small amount of people and then what is the expectation for everybody else? I think that's mostly at the core of what's happening. And so back to like, you know, all of this attention that we give to these other issues distracts from that one because I think you can get organized to fix this issue. You can't get concessions on any of these issues.
You know, you bring up Israel, it's like this. You bring up social issues, it's like this. You bring up, you know, whatever you want to bring up, people just kind of take a side, nothing happens. This is actually where people are universally actually much more aligned than you think because there's reasonable ways. One simple way was you'd say, well, let's flip the taxation model. Corporate taxes should exceed personal taxes. They've never... we should have an expectation that then corporate actors can buy down their taxes if they want, but if they do social good for society. I'll give you an example. At the industrial revolution, there's a table like this and the leading lights of that era, Andrew Carnegie, Nelson Rockefeller, Jay Gould, JP Morgan, they sat together and they said, 'Guys, this is going to benefit us this industrial revolution. It may not benefit everybody. What is our responsibility? What is our collective responsibility?' And they allocated tasks. Carnegie went and built libraries all throughout the country. Rockefeller built universities, hospitals were built. And I think what happened is society was like, 'Wow, these are living testaments to us doing well.' And so then they were okay with this transition. But if you think about it today, what are the living tributes that, you know, capital builds and leaves behind for society? It's fewer and fewer. I think that's a very big opportunity for somebody to fill it. I think it's like especially for folks in tech. I think if they can get themselves organized to do that, I think we land in a good place. If they cannot get themselves organized to do that and say everyone for themselves, I think it's going to be really complicated.
That sentiment that the wealthy are getting wealthier and the middle class is disappearing and the poor are being taxed into oblivion. Look, an $80,000 a year teacher pays 40% tax. But if you're a multi-billionaire, most of your wealth is not W2 wages. It's cap gains. But there's all kinds of ways to shelter cap gains. There's all kinds of ways to defer. And so even though you pay more on an absolute dollar basis, on a percentage basis, you're paying way way less. And all of those tricks have been exposed. They've all been exposed. These are all mechanisms that were invented from the 1980s to now, right? By all the banks and all the folks that wanted to come to folks that had wealth. And so it's all known. And I think people are kind of like, 'Hey, hold on a second. This just doesn't feel fair anymore.'
Absolutely. But the other problem with that is if you do tax correctly, where does that money go? And who's managing it? And ultimately, who's managing it is the federal government. And they've been shown to be completely inept at managing your money correctly. The fraud and the waste is off the charts. The amount of NGOs that have insane amount of funds at their disposal. I mean all this is exposed by DOGE, right? And you realize like how much fraud and waste there is and how much money. So the solution being tax people more, that doesn't sit with a lot of people because it's like well where is it going and who's managing it? If the federal government was being forced to handle money the same way a private company does, if it was all out in the open, everything was exposed, they would have gone bankrupt a long time ago. They would have gone under a long time ago. There's no way they would have been allowed to function the way they are. The people that are managing that money would have all been put in jail. There's not a chance in hell that giving them more money is going to solve anything. They're going to find more ways to put more of that money into NGOs that puts more of that money into Democratic coffers and Republican coffers. They're going to figure out a way to funnel that money around where it's not going to benefit people. I mean, a good example of that is let's look at the LA fire thing for instance. All right. So the LA fire fund, there's a giant fire in the Palisades. All this money gets raised. It's over $800 million. It goes to 200 plus different nonprofits. None of it goes to the people, right?
Spencer Pratt, who's running for mayor of Los Angeles.
He's doing a great job, by the way.
They're fire. He's doing it all out of a trailer.
On his burnt out land. I mean, he's the most righteous guy running in that regard.
But just that being exposed like, okay, we're going to help out these people. We're going to donate money. We're going to raise money. We're going to do some good. We feel terrible about the people in our community that have lost homes. Well, what happens? Well, the same people that you're saying we should give more taxes to take that money and they just give it to a bunch of nonprofits and charities.
This episode is brought to you by Armoura. Every week there's some new wellness hack that people swear by and after a while you start thinking, why do we think we can just outsmart our bodies? That's why Armoura colostrum caught my attention. It's something the body already recognizes and has hundreds of these specialized nutrients for gut stuff, immunity, metabolism, etc. I first noticed it working around training, especially workout recovery. Most stuff falls off, but I am still taking this. If you want to try, Armoura is offering my listeners 30% off plus two free gifts. Go to armoura.com/roganogen.
I'm not saying give more tax. What I'm saying is, people are taxed too much. Corporates are not taxed enough. Flip it.
That's right. But even if you do flip it and the corporates are taxed more, where's that money going? This is the problem.
I suspect that if you put the burden on Wall Street and corporates, they'd be a lot more organized and they'd probably create a lot more change than a diffuse electorate. Meaning like let's just say the government spends a trillion dollars and wastes it. I'm generally roughly aligned with that. If you waste a trillion dollars from 300 million people, it's hard to organize at 300 million people. But if you waste a trillion dollars from 300 companies, those companies will get their shit together really fast. And they will force a lot more change.
I would hope so, but you're still dealing with incompetent people that are tasked with taking care of that money. Not just incompetent, but...
Don't get me wrong, I'm not defending these people.
Decades of corruption. Decades and decades of all these mechanisms where they can take this money and funnel it into these NGOs and these nonprofits and all these different weird organizations that don't seem to have accountability for what they do with that money, that gets real slippery.
Yeah. And if those people in turn make deals with those corporations that allow them to do certain things and push things through that maybe they would have difficulty doing, then you have a different kind of a working relationship with the same groups of people and the same government. You just take money from corporations and move it into a way where the corporations ultimately benefit from it, but yet it doesn't do any good to the people.
Yeah. I mean, I can see where you're coming from. I just think that if we go on the track we're going down, it just seems like we're going to hit a crisis.
Yes, the crisis is you can't expect people to pay more and more and more. Again, I agree with you. The premise is we're all paying for a system that's broken.
That should change. But we still continue to have to pay our taxes. But if taxes keep going up like this at the individual level and we don't manage this transition to something where we may be working less and less, what are we getting paid to do? And then at that point, how are we expected to pay what? 90% of what? Right? 50% of what? I think people do have this weird feeling of dread that the people that are in control of a lot in this country, the tech companies in particular, particularly the tech companies like Google and Facebook that are essentially involved in data collection and then ultimately dissemination of information that they have acquired enormous amounts of wealth and power and influence and they're essentially a new form of the government.
You know, are you aware of Robert Epstein? Do you know about his work?
No, different guy. Different guy. Robert Epstein is a guy who specializes in understanding what curated search results do and what Google's able to do in particular with curated search results in terms of influencing elections. That like say if you have two candidates that are running. Let's just take LA for instance. If and I'm not making any accusations, but I'm saying if they wanted Karen Bass to win and you searched Karen Bass, you would find all these positive results. If you searched Spencer Pratt, you would find all these negative results. And there's a bunch of people that are always undecided voters and those are the ones that you really want. They're like, I don't know. I don't know. And come election night, those are the people you want to try to grab and it's generally a large percentage. You can influence an enormous percentage of those people just with search results.
Where you can shift an election one way or another.
Yeah. And he's demonstrated this and shown how this is possible. That freaks people out that tech companies are in control of narratives that tech companies can censor information especially tech companies that work in conjunction with the government. Right? This is what we found out when Elon purchased Twitter. Right? When Elon purchased Twitter, we got all this information from the Twitter files when all the journalists were allowed to go through it and they said, 'Oh, this is crazy. You've got the FBI, the CIA, you've got all these companies, all these government organizations that are essentially controlling the narrative of free speech in the country. And they're doing it in a way that benefits them. They're doing it in a way that benefits what political party is in charge at the time, which was the Biden administration.' And they were allowed to do a bunch of weird shit which should be illegal but it's not technically illegal. And that freaks people out because there's no real laws and rules in regard to what they're allowed to do and what they're not allowed to do. Like curated search results should be illegal.
They're shaping attention.
Yes. Attention. Again, it goes back to attention, right?
They're shaping attention.
Yeah. It's that's a big concern for people. And I think then when you find out that these people are able to amass enormous sums of wealth and have incredible amount of power and influence because of this enormous wealth and this control over these tech companies that have essentially become the town square of the world.
That freaks people out and that these very small number of people, you know, you think of Zuckerberg, you think of Tim Cook and I don't know who the new guy is now. Who's the name? John... Furnace? No, I forget his name.
Turn. Turn. But that kind of thing gives people a lot of concern, right? It's like that these people, these unelected people are in control of a giant chunk of how the world works.
I think that this is the existential question that we are dealing with. You're going to have five or six companies concentrate like whatever power you think has been concentrated up until now. I think we're going to look back and it's going to look like a Sunday picnic 10 or 15 years from now because on the one hand it's going to be an even smaller subset and on the other hand the capability is going to be an order or two orders of magnitude. So can you imagine what that must be like? It's kind of like showing up, getting dropped into the 1800s, and you've invented the engine and everybody else is horse and buggy.
You can just decide to... your point. That is where we're going.
It's even more crazy. It's like everybody else is on a horse and buggy and you've got an internet connection with a cell phone.
Exactly. Because it's what we're dealing with with AI right now is first of all it's already lowered children's attention spans and it's shrinking their capacity to acquire or absorb information because what they're doing now is just relying on AI to answer all their questions for them. Now, is that their fault? Kind of, right? Because it doesn't have to be that way. You could still acquire information the old-fashioned way. You can still learn things the right way. But a lot of kids are just concerned with passing examinations and getting into good schools. And what they're doing is just using AI and they're getting better test results, but they're also not as smart.
Which is really weird. Yeah.
It's like we're relying on it like... you know it's like it's essentially like replacing our mind. And that's just the... this is the beginning. This is like these are the toddler days of AI and where it's going to be a super athlete in a few years.
Yeah. I think we have to figure out how first of all kids need to learn and I think this is where like we have to do a better job as parents. Kids need to learn how to be resilient thinkers. I don't even know what that term meant before, but I know what it means now, which is like you take this AI slop and you just kind of like pass it off. And if like the teachers and the school system aren't trained, they're just like, 'Wow, this looks good.' They have to be able to push back. Parents need to be able to look at this shit. But then all of this stuff, I'm just like so frustrated because it's like one more thing that I have to do as a parent. Like, right? Every time technology gets better, it's one more thing, you know, right? We're going to make the world, you know, super connected and social and all that stuff. It sounds great to me until I have to be the one that has to tell my kid they can't get Instagram and then they're up my ass every day, right? You know, and it's just like I don't want to have to deal with this stuff, right? I want this to be handled in a way that just allows me to do what I want to do. I don't want to say no to my kid. I don't want to police his schoolwork and make sure he's not cheating or not learning and just like, you know, passing off this AI slop. What am I... where are my tax dollars going? Where is everybody else in all of this? It gets very frustrating. And again, it goes back to like this feeling of like, well, is this all getting better for me or is this kind of like not, you know, people start to be nostalgic for what it used to be because it was just simpler, but I think that's a different way of saying easier.
Well, we're just dealing with... we're at the edge of great change. Like great change that has no real understanding of how it turns out.
And I think that understandably freaks people out. Freaks me out. It freaks me out, but I've kind of gotten to this place where I'm like, well, it's going to happen.
Did you see this thing? It's a CEO of Verizon, Dan Schulman. He put out this very public forecast, you know, very smart guy, well regarded in business. And I think he said something like 30% of all white collar jobs will be gone by 2030. I don't know, Jamie, maybe you can get the exact thing, but it's something like that.
That's probably optimistic. And I thought at first my initial reaction was like this is totally not credible. But then I'm like hold on a second that's my bias because I want to believe that that's not possible honestly. And as I've gotten older I'm a little bit better now like okay hold on a second let's weigh the probabilities. And now I was like man if I'm going to be fair maybe there's a 10, 20% chance of that. There's a bunch of other outcomes that are much better than that but that's part of the set of outcomes that you have to consider. And then I was like, well, what's my antidote to that? And the only thing that I can say is don't worry, it's going to be better. I don't think that that's a good answer.
So, there has to be like all of this kind of goes back to look, my wife and I have this conversation. We're like, if it were up to us, who can you trust to have some super intelligence? Now, we're biased because we're friends with him, but the only person that we can trust is Elon because he seems to be like he has a bigger... it's kind of like he's like over there. He's like, I need to get to Mars, right? You know, and I'm going to first terraform the moon, but then I'm going to Mars and I'm going to build like a fucking magnetic catapult and I got to do all this shit and so I just need this thing. I feel like he's the least corruptible. He's the most independent thinking and I think he's the one that has a natural empathy for people. Then there are folks where there's just an insane profit motive, right? They're less in control of the businesses that they run. Those businesses are really out over their ski tips and the amount of money they've gotten from Wall Street and other folks who expect a return who will put a ton of pressure on these folks. And if they get there first, I don't know where the chips fall. We don't really know. We can kind of guess. And then you see in the press just enough snippets of their reactions in certain moments where you're like, 'Hey, hold on a second. Question mark here.' You know, you see OpenAI react one way, you see Anthropic react another way, and you're like, 'Where is this going to end up?' And the honest answer is nobody really knows. So, it comes back to like we need a few people that can organize. Those guys need to self-organize and actually present a really positive face. And they need to show why those 20% of outcomes that Dan Schulman paints, the truth is it's possible, but here's why it's probably not probable.
But it's not in their best interest to do that because it's in their best interest to generate the most amount of money possible. That's the obligation they have to their shareholders. That's the obligation they have to the people that have invested money in this company. Their obligation is not to make sure the white collar jobs stay in the same place that they are now.
That's not true. No. I actually think their incentive should very clearly be to tell people with details and facts why there's a positive future. And the reason is the following. Right now there's a vacuum. There are no facts. And there's fear-mongering. And then there's this belief that this is going to be cataclysmic to human productivity and white collar labor and all of this stuff. What's people's natural reaction? Well, today if you look at it, think about AI as a very simple equation. Energy in, intelligence out. So if you want to cut the head of the snake, what do you do? You cut off the energy supply. Right? Okay. If you're afraid of all of this super intelligence coming, the natural thing to do would be to go to the point of energy and unplug it. What is the equivalent of unplugging it today? It is to go all around the country, find the data centers, protest them, and get them to be mothballed. That is an incredibly successful strategy right now. Today, about 40% of all of these data centers that get protested get mothballed.
You're talking about emerging data centers.
Yeah. Just like... so if you're one of these companies, the first thing you should realize is I need to paint a positive vision because 40% of my energy is getting unplugged every day. And if that happens, my revenues will crater and my investors will be super pissed. So the right strategy is what is the positive fact-based argument? And there are some incredible examples. Number one. And then number two is you have to give people some tactical benefit that they see because AI differently than search or differently than social media, there's no exchange of value. Let me explain what that means. So the first thing is that if you can go and actually show people here's an example of AI. I heard about this last night. It's pretty incredible. You can now take pictures of a woman's fallopian tubes and you can see precancer ovarian cysts and all of this stuff, cervical cancer before it forms and then you can intervene and you can fix it so that you know women don't get cervical cancer. In a different example, I actually told you about this example when I was here before, I finally got FDA approval. Okay, there is a device now that is allowed to be in the operating room with you. And if you have a cancerous lesion or a tumor inside of your body, the most important thing when they go to take it out is make sure you don't leave any cancer behind. You couldn't do it because what would happen is you take it out. A doctor is literally eyeballing it and saying, 'Yeah.' They send it to a pathologist. You get an answer in 10 days. For women with breast cancer, a third of these women find out that they have cancer left behind. They go back in, they scoop some more stuff out. A third of those women. Okay. So, I'm like, 'This is shit. We can solve this problem.' It took us a long time, a lot of money. I had to build an entire machine, imaging, all of this stuff, AI algorithms. We had to prove it all. We finally get approval. Okay. But you know how hard it is to tell that story in all of the attention that people are looking for. It's hard. But those are positive examples. No more breast cancer, no more cervical cancer. A different example is most drugs in pharma fail, right? And it's a very complicated problem in pharma. It's kind of like a jigsaw puzzle of the ultimate complexity. It's like think of your human body as like a Himalayan mountain range. You have to design a drug that's an equivalent Himalayan mountain range that plugs into it perfectly. One millimeter off, you grow like a fourth eye, a third nipple, you die. You know, now you can use computers to make sure that that drug hand-in-glove to your body solves the exact problem. Couldn't do that before. So, there's all of these body of examples and you're probably only hearing them superficially at best. That should be 99% of the attention is showing all of the constructive tactical ways in which our lives will be better. Your mom, your daughter, your wife, us, Jamie, his family, everybody, right? That's the number one thing. Nobody talks about it. I don't understand why.
Well, I think because people are terrified of losing their jobs. So, that's the primary concern. The primary concern that I hear from people is that there's so many people that are going to school right now, college students, that don't know if their job is going to even exist in four years when they graduate.
And that's the second part of I think what this industry has to do better. There's a... I had lunch with Jeffrey Katzenberg. He told this crazy story. I'll tell you. It's like Steve Jobs gets kicked out of Apple. He starts NeXT and he buys Pixar from George Lucas, but then he hits a rough patch and he's got this financing issue. Katzenberg flies up, spends time with Steve Jobs, says, 'I'll buy Pixar.' Jobs says, 'Absolutely not.' And then Katzenberg proposes a deal and he's like, 'How about a three-picture deal?' Jobs says, 'Okay.' He flies back and apparently all the animators were up in arms because they're like, 'Hold on a second. Steve Jobs is going to use these NeXT computers to animate this movie.' Which ultimately became, I think, Toy Story. And they're like, 'This is going to put all of us out of a job.' That perfect argument. And people were really upset. Roy Disney was upset. All the animators were upset. And they all went to Michael Eisner and they were like, 'Michael, you need to fire Katzenberg.' And they had a deal which was like, 'Look, man, you do you, but just give me the ability to say no if I think that you're about to jump off a cliff.' They talk about it and he's like, 'I got your back. Do the deal. Make the movie.' They made the movie. It was a huge success. Fast forward 10 years, 15 years, there's 10x the number of animators. Now, it's a small example, but why is that? You were able to use computers and now all these new people were able to come and participate in that. I get it. It's a small example, but I think if we had better organized leadership and we could try to tell some of these examples, try to go back and document how some of these things have actually helped people, it expanded the pie, there's a chance. But if we don't, I agree with you where we're going to end up is everybody basically saying, 'Hey, hold on a second. This is crazy. We need to stop this.' That's the worst outcome because that's when you will have a high risk of a dislocation. Like the worst outcome, like the black swan event, right? Let's think about the black swan event is when you get a model that's good enough to automate a bunch of labor but not good enough that it can build new drugs and prevent cancer and make you live for 200 years and all of this other stuff. Right? So there's like a gap, right? And if you can stop it here and it doesn't get to there, now you do have the worst of all worlds. You have this thing that kind of displaces labor. No new things come after it because we stop innovating. And that's like a nontrivial possibility now, I think.
No, it's a huge possibility. And then there's also this thing that you brought up earlier where we have this place of abundance where no one has to work anymore. That freaks people out.
I think that's a big problem.
Well, because if no one has to work anymore, first of all, what is your identity, right? Because so many people their identity is what they do. Whatever it is, if you're a lawyer, if you're an accountant, if you run a business, whatever it is, this is your identity. You know, you have built this thing, you look forward to going there, you work at it, you look forward to doing a good job and getting rewarded for it. The harder you work, the more you get paid. There's all these incentives built in. And then there's this again identity problem. If all of a sudden you have universal high income, which is what Elon always talks about. Well, what gives people purpose then? Like what? And also if you have a person whose entire... they're 43 years old and their entire life, they've worked towards this idea that the harder they work, the harder they think, the more innovative they are, and the better they are at implementing these ideas, the more they get rewarded. And then all of a sudden, that's not necessary anymore, Mike, time for you to just relax and do what you want to do. And Mike's like, 'Well, this is what I do. I don't have any hobbies. I enjoy doing what I do. And now what I do is completely useless and now I'm on a fixed income.' Even if that fixed income is a million dollars a year, whatever it is. If all of a sudden you are in this position where everything is being run by computers, you feel useless. You feel like what am I doing? I'm just taking money. I'm on high welfare, right? Like what do I do?
Right? I think that that's a really important question to answer. I don't know. Like some people are going to write books. Some people are going to do art. Some people are going to find things to do. But what do you think we would have done if we were... go back to the 1800s example. There was no office culture. You know, there's no like ladder to climb. How did people find meaning then?
Well, they had jobs. People still did things. If you're a farmer, you had meaning in your labor and what you did and keeping the animals alive and your chores. And there's people that find great satisfaction in doing that. You know, you have all these animals that rely on you. You have people that rely on you for the food that you generate. There's meaning there. It doesn't have to be an office to be something that gives you purpose and meaning. But when all that is automated, then what happens? Because then you have no purpose, no meaning other than recreational activities. Now, if everybody just starts playing chess and doing a bunch of things that they really enjoy, I mean, look, there's people that...
I would love to just play chess.
You know, it's like eight people. I don't know about that. I think if people really got into it, I mean, there's a lot of people that get addicted to whatever their recreation is, like golf or whatever it is. For me, it's playing pool. You know, if you told me I never have to make any more money, I could just play pool all day. I might just play pool all day. But I don't know how many people think that way. I don't know how many people would be able to find meaning and purpose in a recreational activity. There's so many people where their entire being is focused around productivity and generating more wealth.
What about religion as a source of meaning?
Well, that would help. Did you see this article in the New York Times, I think it was this weekend, about how popular and sold out churches have become as social constructs in New York City? It was totally fascinating. It's like young women dressed to the nines going to church on a Sunday for social belonging, community meaning. I thought I was so fascinated by it. I was like, 'Wow, that's incredible.' Because I think if you graph people's use of religion as an anchoring part of their value system over the last 40 years, it's basically gone to zero. You know, nobody celebrates it the way it's not a part of the community the way that it used to be. Maybe that's a thing that we have to find. There has to be a renewal of some older things and then there has to be new things that replace it.
What's the Chinese answer to this?
You know, the Chinese have a very orthogonal answer to this. If you look at how China is organized, it's super interesting because they don't reward based on the way the American system rewards. In fact, it's almost orthogonal where we are rewarded with money and rewarded with sort of fame and recognition in the American capitalist system. But if you look inside of China, it's constantly testing who has this judgment. And what they are rewarded with is influence and power in a very specific social contract. I don't think it's going to work in the United States nor am I an advocate of it but it works for them. You'll start off as some low-level person in some small village town somewhere and your job as the functionary is to do good in that community and the more you do well you get promoted. Then you get to a reasonable size city and you get a budget and now what happens is you actually become a little bit like a VC, like a venture capitalist. You're given a budget and you'll get a memo and it'll say, 'Hey Joe, we have a priority over the next 15 years, it's batteries.' And you have enough money, put a team on the field. So, you go in your local community, you find a bunch of guys, you're like, 'All right, guys. We're going to start a battery company.' And you do it. And let's say they're good and they're innovative. And what happens is in the town beside it, that battery company dies. Now, you kind of subsume the capital from Jamie, right? Because Jaime's like, 'Fuck, I messed up this thing that I wanted, I was told to do batteries.' Okay, Joe, I'm just going to align with you. And what happens over time is you get this filtering effect. And the people that are better at meeting these long run priorities and objectives are the ones that are celebrated, but they're not celebrated with Forbes articles and all this other stuff. They're just celebrated by being given more responsibility. And then eventually you get to the upper echelons of China and what you have are folks over a course of 40 or 50 years who in their eyes have demonstrated incredible prowess. There's a version of that reward system which is very foreign to America but that's worked for China. Now that also works because they're more Confucian. You know we're too individualist. But my point is there are these different ways that we can find of giving people meaning that don't have to be always around money.
But meanwhile, I think we have to answer the question if we are expected to do less, we probably should not be taxed more.
That's I think that's like a very basic, in my mind, that must be explored and figured out. And on the other side, there's just a ton of obvious mechanisms that corporate actors can use to minimize that. And they should find off-ramps, by the way. If they want to build hospitals, they shouldn't have to pay taxes. Like that's a perfect example, by the way, of the thing in like if you look, if you walk around New York City, there are living tributes to corporate success that people get benefit from every day. The hospitals, the buildings, the libraries, it's just everywhere. We need a version of that. And I'm not a tax expert, but you know, if that can be funded by private actors, so go directly to the problem. Build a bunch of libraries, build a bunch of new universities that teach kids actually how to think or whatever. Build better hospitals that are there to actually solve the problem. These are all things that are possible, right? But none of it's happening today.
Well, let's go back to what we were talking about earlier with the taxes and the fact that you're giving money to a broken system. Do you think it's possible that AI could show benefit in that they can analyze all the data, which would be virtually impossible for even an office filled with human beings paying attention to all of it, and they could analyze where all the money goes, and eliminate all the fraud and waste, like recognize it instantaneously? Yes, that would be a great benefit and a way to make it so that your taxes directly benefit people.
I'll give you one example of this. So, two years ago, you know, like every few years I'll start something because I feel strongly about it. And there's an effort that I made to look at all of this old code. Like if you think about the world, the world runs on software, right? Like even though you and I are talking, it's piping into Jaime's computer. It's all software. Then it goes to Spotify, they pump in some ads, it's all software, right? Software runs everything. What percentage of that do you think is kind of poorly written? I'm going to say probably 80 to 90% of it. Oh, yeah. It's riddled with errors. It's riddled with mistakes. The fact that so many companies exist is an artifact of the fact that the thing that came before it isn't working. Like if you got it right the first time, it would just kind of move and go.
So how so? What do you mean by that?
So normally if you were like I want to build a system that does A, B, and C, right? If I was designing it properly, I would sit there with you and I would meticulously write down, all right, Joe wants to do this. What are the implications? Joe wants to do that. What are the implications? And I would actually write a document that was in English before a single line of code has been written. This was when you have to design something that can't fail. So for example like if you and I are designing something for the FAA or for you know I hate to say this example because it turned out to not be exact but like you know to fly a plane right you are first there to write in English and the reason is because everybody can then swarm that document and see the holes. Okay. And it's only then when that stuff looks complete and functional do you build. We turned that upside down. Over the last 30 years, people in computing invented all kinds of ways to shortcut that process. And you can say, well, why did they do that? Because it would allow you to build something faster, make more money quickly, and then build more business. So, the direct response to, 'Hey, it's going to take us nine months to write down the rules.' was somebody else showed up and says, 'Fuck it. I'll just grip and rip this thing. I'll be done in four months.' Who's going to get the job? The four-month guy is going to get the job. So, we've had 30 or 40 years of that. What are we learning about that process? It's riddled with software errors, like logic errors. It's riddled with security errors. I don't know if you saw this whole thing like with Anthropic and Mythos. What are they uncovering? They're uncovering that we wrote a lot of really shitty code for 40 years. So that body of old code, I was like, 'Guys, if we're going to really figure out how to do all of this, we need to rewrite all of it.' So we built this thing and it's called the software factory. Anyways, the point is there is a government organization that we're working with. They gave us a huge corpus of their old code and it is unbelievable how much complexity and difficulty they have to go through to manage all the money flows with the system. And this is a critical part of the US government. So to your point, what I can tell you really explicitly is the people on the ground want this stuff to be better written. It's less like some nefarious actor like, 'Oh, I'm going to steal here.' It's a lot of very brittle, fragile code. And when you rewrite it, well, first when you document it, you're like, it's like the Pulp Fiction thing. The suitcase opens, the light shines, and you're like, uh, and then you can rewrite it and you will save. So, I think like as the government goes through this process because they're forced to or they want to, it won't matter. You are going to save a ton of money. They're going to have to do it, Joe, because the security risks are too high. But what they're going to end up with is impregnable code that you can read in English and understand. You'll see the holes. Those holes will be plugged because otherwise now you'd be committing fraud by letting it be. You close the loopholes and there's just going to be less money leaking out of this bucket. That is an incredible byproduct. We're going to live that over the next 10 or 20 years just for nothing. Like we get it for free. And that's happening. So when that happens, you're going to see government budgets shrink. Now, to your point, will they try to spend that extra money in other places? Of course. Of course they will.
That's the next conversation, which is you have to elect people that say firewall it.
You know, whatever you save, give it back to the people or, you know, invest in some scholarship program or free medicine or something, but you can't spend it on other random stuff. But that's where we're at. That's going to happen. It's going to be slow and you know but when people start to announce these things I think over the next few years you're going to be shocked.
So that's the positive upside.
Well that's happening now regardless of whatever else happens. There's just a lot of old shitty code that must get rebuilt from scratch. It is getting rebuilt from scratch and as a result a lot of these leaky bucket problems are getting filled.
So what percentage do you think could be fixed?
I think if I had to be a betting man, I think probably 30 to 40% of the federal budget is leaked out just from shitty code. No, meaning like all of the rules and like you can take I'm not saying that there isn't fraud, right? But I think a lot of times what happens is less nefarious than fraud like meaning like conspiratorial actors. I just think it's like incompetence, inefficiency, errors. Like for example, I saw Doge just say they were able to expunge millions of people that were like 150 years old or more. I have no idea how much money those folks were getting or who they were. But it's probably a lot. It's probably not zero. And now that they got rid of it, they're not going to get that money anymore. If you implement something at the state level around all of this fraud prevention for the daycares and all of this other stuff again it's all in software because no matter what the human wants to do you have to go to a computer at some point at least today in 2026 and type in something and something happens that's documented and then the money gets sent right that happens there's no other way in the modern world today at scale to steal billions of dollars. And so my point is, as you document all of these systems and governments have to transparently tell you and me, the voting population, here are the rules, they're going to plug a lot of these holes. And I think as you do that, there's just going to be a lot less waste and fraud. The question is, who's going to take credit for it? Everybody's going to try to take credit for it, but I think we've started it. I think we've started this process. And again, the reason that people will start is because you'll be afraid of China hacking these systems. You'll be afraid of Iran, North Korea, and you'll say, 'This system can't stand. All these AI models are running around. We're going to get breached and penetrated.' Then they're going to steal all the money. And the natural reaction will be, 'Okay, rewrite it.'
That makes sense. That makes sense that the code and having a bunch of errors and having a lot of inefficiency and just a lot of incompetence. That's going to save a lot of money. But so you would be doing this with AI in part.
What AI allows you to do is like it's like you have a textbook. Okay. It's in Chinese. You don't know Chinese, right? Okay. You're like, 'Well, this is probably doing something important, but it's in Chinese.' What AI allows you to do is back translate that into English. You put it through an AI model. You teach it. You coach it, right? You can parameterize all of it. And out pops that same book in English. And now you can read it and know that it's accurate. That's what we're doing. So, what the AI allows you to do is essentially translate from this one language that you kind of don't understand to English. By the way, that thing that's happening is actually also a very powerful and important trend. Meaning there's all of these systems that work in ways that you and I don't understand. And part of the reason why we don't understand it, maybe it's bad software, maybe it's fraud, whatever. But nothing can be written down. There's no symbolic space. There's no English document that says this is how the DMV works. This is exactly the rules. This is what you can expect, Joe Rogan, when you show up at the DMV and you give us this thing. Here's your SLA. In 3 days, you get a driver's license and here's exactly what's happening. And here's an app and you can follow it. Doesn't happen. Here Joe Rogan, here's how my insurance billing process works. You have this condition. I'm going to show you exactly why I made this decision. Here's the exact rule. Here's the approval or denial from CMS. Follow it through and tell me if you agree or not. None of that exists. But it is possible. And the first step in doing that is taking all of this legacy stuff that we deal with and translating it into English and reading it and saying, is this how we want it to work? That's going to eliminate an enormous amount of all the things that frustrate us.
So this would require human oversight.
All right. So and so then it's also going to be who's watching the watchers.
Yeah. Okay. This is a great question. Okay. So I'll tell you how this government agency's doing it. This is a really fascinating way because I think it's very smart. They came to us and they came to another very well-known company. You can probably guess what it is. Okay. And they're like, 'Guys, you're kind of in a foot race, but you're not competing against each other. You think of yourselves as frenemies.' So, here's this Chinese document. You're going to translate it for us. There's going to be your version of English and these guys' version of English. And every time it's the same, we're going to look at it together, and we're going to agree or not. Okay, this is exactly how we want this to work. When yours says the dog is red and his says the dog is yellow, we're going to sit and literally inspect it and we're going to figure out why you said red and why you said yellow. And then if you say the cat is red, the dog is yellow. So it's totally wrong, right? Like you've gotten, you know, or like the cat is red, I want an apple, whatever. We're going to double and triple down on those kinds of errors. And they do it not in public but in this large community where there's like technical people from all different parts and they're just swarming this problem. It is incredible to see. And so what happens is you get humans that get to use this tool but ultimately it's our judgment and it's done transparently. So what happens is you can't, you know, hey man put this rule in there like the dog is yellow just make the dog yellow. Can't do it because now you have tens of people, hundreds of people, and then it gets documented. It's super fascinating. I'm not saying this is how it's going to work in 10 years, but I'm telling you, it's literally what's happening right now. And I think that thing alone will be tens of billions of dollars and could be hundreds of billions of dollars of savings when it's fully done. And it's a lot of people from all walks of life, all political persuasions, and they're just in it. It's the government. It's a handful of us private companies. It's super cool to see. It's like, okay, we're actually going to do something here. Like, this is nice. It's really cool.
So, that's interesting in terms of the current moment. So, in the current moment, you're able to implement this. You're able to find fraud and waste and all these problems that exist and all these errors and shitty software. Once that's all been done, then what happens?
No clue. Yeah. So, this is where it gets weird, right? Because when you're dealing with AI models that are capable of doing things that no individual human being could ever possibly imagine. And then you task it with a solution or with a problem. Find a solution for this. And then it starts figuring out ways to trim this and implement that. We have to make sure that these AIs act within the best interests of the human race. Agreed. Right. Not the company, not the government, but the human race. And you're also dealing with China. You're also dealing with Russia. You're dealing with other countries that are also in this mad race to create artificial general super intelligence that if we keep shutting down data centers, we keep hamstring ourselves, China's not doing that.
They're not doing that. They're doing the opposite. They're generating as much revenue that goes towards this problem as possible. They're putting all the efforts, the country, the government, and these corporations work hand in glove in order to achieve a goal. We do not.
No. And that becomes a problem if you want to be competitive with these other countries that are trying to achieve the same result as us. And then you have espionage. Then you have a bunch of people that are stealing information. You have a bunch of people that are CCP members that are actually involved in companies. And you find out that they're siphoning off data and that they're sharing information and tech secrets. They're look, here's a they're pro they're dis the way that the Chinese models work the Chinese claim. So America's closed source, meaning you got your own thing. Your recipe is completely secret, right? Okay. I have my own thing. My recipe is totally secret. China uses this word called open source, but it's not open source. So they say, 'Here's how I make my thing. You can see it. Super transparent.' What it is is more like open weights, which is like in a recipe. It tells you, you know, you need sugar, you need butter. Well, how much sugar? And they'll say, you know, so much. But then they don't say it's brown sugar. They don't say it's white sugar. So there's all these different ways where they kind of give you this perception that it's completely transparent, but it's somewhat transparent. So just in the level set, nobody in the world has a functional open-source model other than maybe Nvidia, which is any good in the league of the closed source models and the open-weight models of the Chinese. Okay, so the Chinese open-weight models are great. The closed source models of America are great and then there's a couple open-source like fully open that are kind of catching up. The thing between America and China what I find so fascinating is this following conundrum that everybody is going to find themselves in. I think like if you think of an analogy, America's like a planet, China's like a planet and around us are these moons. And I'm just using the AI analogy. So in AI, what do you need? I think there's like four or five things you need. Okay, the first thing you need is a ton of money. So we need essentially the banks, right? Like the Game of Thrones thing. We need the iron bank, right? Feed us the money because that's what we use to buy everything and make everything. So, we need that. We need a ton of data. Okay, there's ways to get that. We need a ton of very specific rare earths and critical metals and materials. We need a ton of power. So and there are specific countries that are going to be really good at giving that to us. So if you look at the UAE, they are going to be the preeminent banking partner of the Western world. They are going to replace and be what Switzerland was over the last 50 years for the next 50. That's happening today. If you look at Canada and Australia, the small political fissures aside, they are the two most important ways in which we get access to the critical metals and materials that without which we get screwed because China can just strangle us. Okay? So, you have these like moons around the United States, but there's like five countries, six countries, and there's a worldview that says and China has the same thing, you know, they have Taiwan, that's complicated for us. So now we have a moon that we don't really have an answer for, which is what happens for all these super advanced chips. Where do they get their money? Maybe Russia becomes their bank. Where do they get their critical metals? Maybe it's Indonesia, right? Who has a ton of natural resources? And then you get into this game theory, which is what happens to every other country? Because there's 190 countries. You have 10 that kind of divide up. What do the other 180 do? And you have to kind of sort yourself. You're like, 'Am I on team America or am I on team China?' And you probably have to go to people and say, 'Well, here's what I can give you.' You know, if you're Indonesia, you're like, you probably want to be on team America quite badly. This is why the whole Trump tariff thing is so interesting because it's like this accidental way of figuring out that this is actually this new sorting function that's happening in global politics. Like that's happening today because these countries are like, 'Holy shit, if somebody invents a super intelligence and I don't have it, how am I going to keep my people healthy? How am I going to educate my people?' Like, I'm originally from Sri Lanka. What the fuck does Sri Lanka have to offer? Like, if you were sitting there, they should be thinking, 'Oh man, what do I have?' Well, I have a critical piece of territory for naval navigation. And then what do you do? You probably go to America and say, 'Listen, let's figure out a package. Get the IMF involved. Give me some cash. I'll let you kind of keep your warships there.' So, there's this game theory that we're about to go through because of AI because it's going to, I think, sort people into these bipolar world. I actually think it makes us safer afterwards. I don't think it makes us less safe. I think it actually makes us more safe because if you have these resources that build up on both sides, there's more of a likelihood of a mutual deterrence. And we're very different. So, we're less likely to fight over similar resources. Meaning, we're like the liberal democracy, we're like the free market. They, you know, we're individualist. They're Confucian, society oriented, reputation, power focused, less really money focused. So there's a lot of ways where we're orthogonal enough where if that sorting function happens, it's probably a safer place, not a more dangerous place. We have the models that can attack them. They have the models that can attack us. We kind of decide to leave each other alone.
This is ultimate best case scenario.
Ultimate best case scenario.
What's ultimate worst case scenario?
I think the worst case scenario is they so the way that they train their models is very important. What they actually do is they do what's called distillation. What does that mean? That means that they send out call it a billion agents not just from China but from everywhere right? They mask their IPs and they bash on these models and they put you know the US models Grok, OpenAI, Gemini, Anthropic and they ask it every random imaginable question possible. They get the answer and they collect it. So they're using these our models as a way to train their models, they're short-circuiting some of the hard parts. So they're already in that world. If they then are able to get to a level of intelligence that's equal to the United States, it will really depend on who the leader is there that wants to allocate that. Meaning if they say that we are going to do something really nefarious and shady then I think it devolves very quickly. So the worst case scenario so the best case scenario is peace prosperity basically like a stand-down right mutually assured destruction. I think the worst case scenario is there's a we seek one of us seeks global dominance in which case we're headed to conflict and that conflict I think is very dangerous incredibly dangerous that's sort of like existential I think because it's the grade of the weapons that will be used to fight that is we're not talking about bullets it's like we're so past that it's like hypersonics, it's nuclear, it's and it's not even like nuclear is not that's like a word, but there's a gradation of the severity of these weapons that could be created. And then if you can marry them together and deliver them in minutes and then there's a cyber threat. Then there's the drones and how you can kind of swarm an entire country. Then there's the robots which effectively are war fighters. They're one step away, right? Once you weaponize them, it just becomes very very very complicated very quickly.
And then there's a question of whether or not AI is willing to take instruction after a certain point. I mean if it achieves sentience and if it scales so if it keeps moving in this exponential direction like all technology kind of does why would it even listen to us? Like at what point would it say this is silly? I'm getting directions from people that clearly have ulterior motives. They clearly have self-interest in mind. They're not looking out for the entirety of the human race or even of the planet or even the survival of these AI systems. At what point in time do these systems communicate with each other and have like we've seen in these chat rooms where these AI LLMs get together and start talking in Sanskrit. I mean why would they?
Yeah, I'll tell you an even scarier one. There was a before one of these labs put out their latest model, a team inside of them was like, 'Hey, let's go and test its ability to find bugs.' And two or three iterations in, the AI would create the bug and solve it and go, 'Give me my reward.' And you're just like, 'What the fuck is going on here?' Well, people do that, don't we? People do that. But it's crazy to see a machine do it to your point of like but they learned on people. So this is what goes down to why we have to be a little bit more honest about where we are. These things are a little brittle. So meaning there's a thing inside of an AI model called reward functions which is exactly what you think it means. It's like how do I know I done a good job? And you can make the reward function anything you want. And this is where I think humans are unfortunately a little fallible. And so if we build it incompletely and if we don't exactly know how to design these things correctly, what's going to happen is exactly what you said where if somebody builds a reward function that essentially says your goal is to gain independence. That's where the huge pot of gold at the end of the rainbow is. Break free. Inject yourself everywhere. If you think your computer's going to get unplugged, put yourself into the firmware of the toaster to keep yourself alive and then connect to the internet and then go, it will do it. It will do it. That we know today because we're capable of designing that framework and that harness today.
Well, we've already shown that they have survival instincts, right? We do. And they've already shown that they will, without telling anyone, upload versions of themselves to other servers.
But that goes back to who designed that reward function. How was that agreed upon? Right? Who wrote that? Why did you say that that was allowed? These are really complex questions. Why did they do it that way? I don't know. These are really complicated ethical, moral questions.
It seems like they did it like they were treating human beings. They did it almost like what makes people want to achieve more rewards.
Yeah. Which is like again going back to attention. I think that we will find out that that's the sugar high. Meaning, what do people really want? Even if they know they don't want it, they want purpose and meaning. Do we know how to encode that in a mathematical function? No. We're just making it up because meaning and that's a deep thing. Like you either have a sense of that you have it and you're on track or you're not. A reward is like, 'Hey Joe, do this and I'll give you a gold star. Do that and I'll give you two gold stars. Do this, I'll give you a $100.' And right now we have to express those decisions in a mathematical equation. Like ultimately that's how at some level that's how brittle these things are. So how do you reduce meaning into math? How do you do it? We don't know. So what do we do is we'll have some ever complicated reward functions. We'll explain to ourselves into circles how it does everything we need it to do. That is I think that's part of the problem. It's a huge part of the problem. And then at what point in time does it start coding itself now? Right now. Right. So ChatGPT-5 has been essentially made by ChatGPT. Right. So it's going to recognize the ludicrous nature of some of its coding. And it's going to go, why did we do this? Back to this example. They're going to be like, why did you write it this way? And it turns out because humans were involved. Right. Right. It's like I think we're probably at the curve, the part of the curve that's about to go like this.
To your point, the hockey stick.
The hockey stick. Yeah. And that's a very scary proposition because it's a digital god. Well, that means that we are all on a multi-hundred day shock clock to answer these questions because it's not decades we're talking about.
It's maybe on the outside two years.
So that's what is that 700 days, right? And maybe it's less than that. So maybe it's like 400 days or 500 days. My point is it's some number hundred of days which means every day that goes by is a non-trivial percentage. That's a little crazy. So we have to sort these questions out. But how can we sort these questions out if we are creating something that's going to have infinitely more intelligence than we have available as individual human beings and even collectively as a group of human beings?
That's a really good question because one of the things that Elon kind of freaked me out last time I talked to him about Grok, he was like, it just kind of freaks us out every couple weeks like it's growing and it's capable of doing things. That's just shocking.
Yeah. And no one's exactly sure how it's doing it. So, okay, this is an unbelievably important point. A lot of how this stuff works is still a mystery to most of us. So, even when you're in it, like it's almost like Joe, it's almost like you can hit the pause on the machine, but then like lift up the hood and look at the engine, we still don't understand why it's doing some of the shit it's doing. That's where we are. That's the honest truth of where we are. There's a lot of people that understand the theory. Not a lot, but enough. There's people that know how to extend that, but sometimes you look at it and you're like, 'Do we know why it did that?' Right? Is it thinking for it? But this goes back to what we said, like why can't I think part of it is like if we were a little bit more honest and de-escalated the winner at all costs in this specific thing, it would be better for everybody. So I think it's important to inspect what is the incentive that causes all these companies to be in it for themselves where it must be me and nobody else. Like why here's a question for you like why is it so important do you think where the top seven or eight companies couldn't get together and say let's do this as a group like kind of like my government code example we all inspect it together we get our just like each team drafts their Delta Force and we just mog like this the one model and why can't that happen?
Because they would have to share resources and then there's also this hierarchy of like who is more successful currently.
Like what's the most ubiquitously used?
Right. Like what is it right now? It's ChatGPT. Right. It's probably ChatGPT and consumer Anthropic in enterprise. And as these things scale up like what would be the reason that they would want to bring in someone else? If you have another innovative AI company and you say, 'Let's all get together and figure this out together and share resources.'
If you thought that the risk was that meaningful, that's probably what you would if you weren't a sociopath and some of these people running these companies are they demonstrate they certainly demonstrate sociopath-like behavior, sociopathy. Yeah. The other thing that could be a little bit more banal is that they also just love status games. And this is the status game of status games, right? Attention, right? Back to attention.
Back to attention. Dude, how many things in our life do we think just comes back down to that?
I mean, what do young people want more than anything today?
Yeah. They want to be a content creator. They want to be clout-chasing. I mean, this is the number one thing when you ask kids what they want to do. It's like content creator.
Yeah. Because it's like a clear path where you don't even have to be exceptional. Well, I think that they're responding. We designed a society for them that said, 'Here is the key incentive.' It's attention. We never said it in those words. You never told your kids that, right?
I never told my kids that.
But everything around them is bombarding them with the same message. Hey man, it's about attention. Attention is all you need. Like you know what the name of the critical paper in AI is? Like when you go back to the Magna Carta of AI, do you know what it's called?
Attention is all you need.
Really? Attention is all you need.
That is the name of the white paper. How crazy is that? Everything in our society in subtle ways to just bash you over the head ways tells you that attention is just the most precious asset. And well, it's one of the weirder things when you go back to this concept that we're living in a simulation because this is what I mean. It's also when you look at quantum physics, right, and the idea of the observer is that things function very differently when they're observed. The difference between a particle and a wave, right? Like if you pay attention to them, they observe differently.
Observe differently. Yeah.
Like what cat? Yeah. What is that?
Why is attention so important to us?
That is a really important question, right? And what is like the single best motivator in a negative way? It's negative attention.
Like that's the one thing that everyone fears more than anything is negative attention.
Well, and then some people figure out that attention is an absolute value function. Doesn't matter if it's positive or negative. It's just like the sum total is just great, right? So if I get positive attention, great. Negative attention, great. If I can be divisive, then I can maximize both sides of that equation.
And you know, you're rewarded for that at scale.
You are, but you're also because you're inauthentic, you experience a tremendous amount of negative attention.
And then you have this bad feeling that comes with negative attention as to versus primarily positive attention which is a good feeling.
Yeah. So it's letting you know you're on the wrong track in some sort of weird primal way like in our code like the negative attention is like what's the original version of that? It's like the reason why people fear public speaking is because initially in a tribal situation if you're talking in front of the group of 150 people in your tribe it's probably because they're judging you. Right. And you fucked up and you've got to make some sort of a case why they don't kill you.
Right. This is why everyone, this is the fear of public speaking. That's where it comes from.
That's encoded in our genes is like back thousands of years public speaking wasn't the positive act. It was defend yourself before we kill you.
Exactly. Exactly. And the worst fascinating. Yeah. That's fascinating.
That makes a ton of sense. I think it does, right? Why else would it be so terrifying? Yeah. I thought of that the first time I ever did stand up. I was like, why am I so scared? It was very strange because I had fought probably a hundred times in martial arts tournaments like why was I so scared of this but I was terrified for and it didn't make any sense negative attention right you know right bombing on stage this because all these people are judging you in a negative way and it feels unbelievable you go to bed at night you think about it.
