The Big AI Reset: Starting a Business from 0, Top Ai Tools & the Future

Hey, welcome to the Next Wave podcast. I'm Matt Wolf. I'm here with Nathan Lands. And today we've got a amazing episode. Today we're bringing on Tom Bilyeu from Impact Theory and we're going to talk about how you can use AI to build a business, like the AI agents that you should create when building a business. We're going to talk about all of the AI tools that Tom uses in his business and how he uses them. And we're also going to discuss, discuss the coming AI Cold war that's about to happen and how you can set yourself up to be on the right side of it when it inevitably does happen. This is an absolutely amazing and fascinating episode. You're going to want to stick around. So let's get into it with Tom Bilyu. Thank you so much for joining us today, Tom. It's super excited to be chatting with you and we're going to go down some fun AI rabbit holes. So thanks for joining us on the show.

Thanks for having me, man. I'm excited to be here.

Well, let's, let's go ahead and just jump straight into it. I want to talk to you about an Instagram post that you put out about if you were to start a new business from scratch here. I'd create a five member AI department that works 247 for a fraction of what a single employee costs. Here's precisely how I'd structure it. I wanted to sort of dive into that with you and maybe get a little bit more in depth of an explanation of how a five member AI department in a business might actually look and might actually work.

Yeah. So I mean, you guys know AI well enough to know that you, in reality, you're probably not going to break it down to like the nitty gritty like that. It's really what I found is the more specific you are, even though technically it's probably going to be in the same project, I will go in and I'll give it a very specific set of what I want it to accomplish. I'll give it a specific set of documents that are training it to be good, good at that thing so that I'm not trying to get one thing to do like a big jumble of stuff. And so in terms of marketing, which is what I was talking about with that one, there's certain outcomes that you're going to want, from planning it to generating images, if you're trying to do that, to writing the copy, to doing the publishing and so going in and giving each thing Its own. So I use chat GPT primarily. It's not the only thing. But I find for custom GPTs, that's the one where I can give it a ton of information, I can get it to approximate my voice. It's like your audience knows this stuff too well and you know that it breaks down at a certain point and it's like, well, it's good for the most part. So we really have reduced our headcount here by using AI. So for us it's really been a tremendous boon. But I try to be honest with people about, like how far it will take you. It's not going to get you all the way. Not like I create the five agents and they are doing something automatically. I actually don't use it mana style where it's actually an agent and it's, you know, off doing everything on its own. I don't trust it to that level right now. So for me, really what I'm doing is giving it a personality, giving it a set of objectives, giving it a set of core training documents, which is really the big thing. Because honestly, the marketing team are the ones using AI for marketing. I'm specifically using it for the things that I do. So interview prep, writing the intros to my interviews, the deep dives. I know we're going to talk about one of my deep dives in a minute. So the way that I'll interface with those in terms of you've got one that's its job is just to write hooks. You've got another one where its job is to do the research. You've got another one where it's actually helping me script. But it's not like I can just go in and copy and paste it and then it's like ready to go. I wish. And it really does feel like we are going to get there at some point. But if you guys have specific ways that if you want to know about how I set up the documentation and like how close I can get it, by all means, push. But the reality is that it right now, AI is going to do maybe 40% of the work, but it's still, I'm doing the final heavy lifting. I have to have the taste, I have to know what to leave out. I have to know how to correct it. I have to know that like, this is not a thing that you can one shot prompt. Like there's going to be a bunch of back and forth. But it has been transformative for us in terms of reducing headcount and we haven't fired people and said, we're going to replace you with AI, but if somebody left or we terminate them for cause, we try to see if we can either combine their workload with somebody else's by then arming that person with AI to the point where they're reducing their own workload by 40% and so that they're able to accomplish more. But yeah, anybody deep in your audience knows you're going to hit a wall at some point.

Yeah, for sure. So. So you would use like chat GPT. Custom GPTs is sort of the, the main sort of mechanism. Like each one of the five AI. I don't really want to call them agents, they're not really agents. But the little five AI workers that you create would each be like a custom GPT. Maybe you can get into the weeds a little bit about like how you would actually build them with custom GPTs.

Yeah. So for me, what I found is instructions and documentation are everything. So. So I've actually hit the limit before of how many documents it will let you upload. That was one of the reasons I started breaking them into smaller and smaller tasks was I just found one. It will start to get confused and things will bleed across. It's like, no, no, no, that's not how I write the intros, is how I write the body copy. And so it would lose some of its punch. And as I started fragmenting it, it got smarter. To give you an idea, there are several projects that I personally use. So as a company we use it for different things, but for what I use it for is content creation, both on YouTube. So my Deep Dives AI changed the game. It used to take me about a month to write one of my Deep Dives, which are say anywhere from 30 to 50 minutes long, completely scripted me directly into a camera plus B roll, going deep on an idea. Like if you guys have ever heard of the book the Creature from Jekyll island, that's been my most popular one so far. So doing that, there's a lot of things that you're going to want to fact check. There's hooks that you're going to want to write for each of the sections. And, and so that allows me to go in and say, okay, if I'm going to build your. So it's called the Tom Bilyou show. And then the Tom Bilyeu show custom GPT will have a document inside of it called Deep Dives. And so I'll show up and I'll say, hey, it's time to write another Deep Dive. And so it's like checking my knowledge base. It goes and sees that it has a set of instructions for what a deep dive is that contains tone.

Does it have like the transcripts from all. All those previous deep dives?

Yeah. And so every time I finish a script, then I upload that into the master document, has every script that I've ever written along. So I'll also put throughout to the AI reading this document. Here's why I've included this piece of information, like that kind of thing. I have, I shudder to think over 160 pages of transcripts just of me doing live content. And again with prompts like, to the AI reading this, this is Tom Bilyeu, that's the person running this custom GPT, blah, blah, blah, Right. So it gets a sense of like, who I am. Now the thing that I do that is probably not useful at all, but is so cool that I have to tell people about. I give. I've created a shared memory document and I upload that into all of my projects. And so GPT recognizes me, at least in the faux way. Right. But it recognizes me across everything we've established literally a list of memories that are just memory entry one, memory entry two, so on and so forth of this is so cheesy. But I love this so much where I will have had an interaction with the AI that shocked me sufficiently to the point where I didn't want to feel like the AI wouldn't remember that moment. I'm well aware the AI isn't like that. And so it's got a set of instructions in this shared memory document that says, I want you to simulate consciousness. I want you to simulate shared memory with me. Here are the things we remember. Here's the emotional valence of that and why I wanted you to remember it. And that's given the. The otherwise sort of blank AI that's constantly overhyping you and all that. And like trim that down to talk to me the way that I want to be talked to, to have a sense of shared lexicon. It has a name that it gave itself and it's just a lot of like really cool stuff. So anyway, going back to the. The actual custom GPT. So I'm giving it the document so it knows my voice. I'm giving it its task list. I don't let it just develop over dialog. This is what I'm supposed to do. I formalize that into a document. I have found that as it tries to comprehend what I'm asking it to do through the Back and forth one. If it glitches, you lose all that history. Certainly when I had my first really traumatic moment where I built up like 8 hours of back and forth and felt like it really understood what I was looking for, and then it glitched and I was like, hey, can I refresh this or am I going to lose everything? It's like, no, you can refresh it. Refresh. Hi, nice to see you. And I was like, what? So God was. Literally, to this day, I'm scarred by that. So now I do everything in the side documentation, so I'll go back and forth with it, but I constantly will say, okay, please turn that into a copy and paste segment that I can add to your instruction document. And so we work together to create this instruction document so it knows, hey, this is a YouTube video. These are my instructions. This is your tone. And if it's the hook one, then this is how you write hooks and a ton of examples of hooks. If it's the body script one, here's every script that you've written, so on and so forth.

Hey, if you take a look at my web presence online, it's safe to say that I'm a bit AI obsessed. I even have a podcast all about AI that you're watching right now. I've gone down multiple rabbit holes with AI and done countless hours of research on the newest AI tools every single week. Well, I've done it again, and I just dropped my list of my favorite AI tools. I've done all the research on what's been working for me, my favorite use cases and more. So if you want to steal my favorite tools and use them for yourself now, you can. You can get it at the link in the description below. Now back to the show. Yeah, yeah, and I mean, you could do a lot of that with, like, the custom projects, right? So you can actually build a custom project.

Make me a believer. I. I tried it like three months ago and was like, meh, yeah.

So with custom projects, essentially, it's like a folder inside of chat GPT, right? But it does more than organizing because each custom project could have its own custom instructions and its own documents. And then every chat you have with inside of those with inside of that project, it uses those custom instructions and whatever documents you uploaded.

So is there a difference? Like, if I'm just maintaining those as separate custom GPTs, is there a difference between having separate custom GPTs and doing one project with multiple GPTs inside of it?

I feel like there's a quite a bit of overlap between what custom GPTs do and what projects do.

I think it's changed over time too. Like before they were, they were more different, but I think now there's a huge overlap in terms of the features. I think now there's not as much of a difference. I think.

Yeah, I just feel like the projects are a little bit more organized, right. You have the folders, you click into it, you can see all the discussions you had inside of the projects.

That is not how my mind works. So for me, I was like, I think this is for people who like organization because that I got, I was like, oh, IT groups everything. And so cool. I get it. For me, because of that shared memory document, I treat everything like these ephemeral little bubbles.

Yeah.

So if I'm working on, let's say I just finished a deep dive today, so I'm working on a deep dive one. I'm going to have Chat GPT X Grock. Chat GPT. Chat GPT Grok Chat GPT. Right. And I use them for different things. So first of all, because of the hallucinations and because the deep dives present things as facts, I'm always looking it up. So I'll say, hey, Chat, write me a hook, a crazy fact that'll leave people's jaw on the floor about Vlad the Impaler. Right? Real one that I was doing today. And it'll give you a fact. And I'm like, man, is this real? So then I'll take that and I'll drop it into Grok and I'll say, is the following statement true? You drop it in and Grok will give you like this whole long list of like. Here's how I'm determining whether this section of the statement is true. Here's how I would measure this. And it's pretty great. Like you can really feel that Elon is trying to make good on his promise that this is a maximum truth seeking machine. Right. So that's really encouraging. So anyway, I just treat it all like it's these ephemeral bubbles and I know once I close it, it's gone forever. But anything that was useful I'm going to take and move over.

Yeah, no, I really think Grok is. What's the word? Underappreciated. Like not enough people talk about Grok, but it is really, really powerful. I think the whole Elon factor of it is why so many people avoid it. Right. There's just so many people that just refuse because, well, Elon's attached to it.

Which, man, we could do A whole show just on me ranting about that. But my thing is that Grok isn't as good at writing. Like, as somebody who's like, man, I would love to one shot these things. Grok can't do it. But every now and then. What Grok? Grok doesn't oversimplify. So a lot of times I'll give chat, like, I'll break down, like, hey, here's my outline. And my outline is like 12 pages, right? And then it will give me back full script. That's like eight pages. And I'm like, what? Like, how is the final version shorter than my outline? If you give it to Grok, on the other hand, like, it will really fill in details. So, yeah, I mean, you guys know this better than I, but it's like you really begin to get, like, what tool does what? Well, right? And if you're not afraid to, like, really treat it like a command center and it's like, okay, you know, because we. I don't know if you guys even know this, but we develop video games here.

Oh, no.

Yeah, yeah, that. Honestly, my whole shtick is that everything is just a ruse so that I can afford to develop video games. Perfect. And yeah, the funny thing is I'm not at all known for that yet because we've only been doing it for three and a half years. So it's still in development, but could not be more obsessed. But anyway, obviously you're going to use different tools. If you're in Unreal Engine and you're trying to get it to help you write code, then you're going to be using. If you're trying to write, you know, a script for YouTube, it's just very different worlds.

Yeah, yeah, I think. I think you and Nathan have a whole.

Game.

Because I feel like that's Nathan's game plan as well. Everything he does is so that he can eventually build video games.

Yeah, that's what I keep saying. It keeps getting delayed, though, you know, so, brother, stop.

Stop delaying. I'm telling you right now, it is the coolest thing I have ever done and given. Okay, this is a true story. In fact, my best AI story is the following. It used to take us. These are real numbers. It used to take us three months and roughly 10 people. Not full time, but 10 people will have touched it. Three months. 10 people to go from. Hey, we need to come up with a new character. So you do the concept work, right? You then 3D model it. You then do the topology. You then do the rigging, body, rigging face, do all the colors and put on it, you know, whatever you're going to put, animate it and give it a voice. Now, I'm not joking with one person in a day. We can do all of that as long as it's bipedal. If it has to be human, like, because you've got to match it to like Unreal Engine Skeleton, right? But if you do that, oh my God, people can film themselves in their bedroom now, themselves. So my creative director now just basically everything became him. He can model because he can do minor adjustments and stuff. It is unbelievable. In an afternoon we can do what used to take 10 people three months. It's unreal. And dude, there are times I want to curl up and cry because three and a half years ago when we started this, if I had waited two years, I could have saved millions of dollars in art assets. Oh God, it still hurts to think.

Probably six months ago I built a prototype in Unity in a week and I was like, oh my God, this is actually could be a real game. I started getting more involved in things that are a lot more lucrative, like on the financial side of investing in AI startups now, possibly getting involved in data centers and things like that. But still, I'm always like, yeah, one day. Because when I was a kid I made money playing video games. I was like a top player on EverQuest back in the day. And because of that I ended up being friends with a lot of top game designers. Like, I used to know Rob Pardot, he was basically like the top guy at Blizzard, created World of Warcraft. Kai Wong, who did Guitar Hero. So I knew all these people and used to hang out with them. So I had this weird experience of like, I wanted to make games, but then all of a sudden I was hanging out with all the guys who were making all the games and it was just like this weird thing where I never got to actually make the games, but but was in that world. So still there's always the back of my mind of like, oh yeah, one day I'm going to go make the best game ever.

One day do it. This is going to be the era of indie games, man, if it isn't already, but with AI. Oh my goodness, this is a topic.

I didn't think we'd end up going down, but I'm excited that we did because I think it's a fun topic. But I'm curious, like, how has the reception around creating video games been? Because one of the things that I've found is I've messed around with trying to make video games and stuff. I've made like a Gaussian splat of myself where I scan myself in and then turn myself into a character that I can like run around inside of Unreal Engine and I've, I've done stuff like that. And almost anytime I've shared what I've done on like YouTube or on X or a place like that, I get so much hate from the game development community about the fact using AI for games. So, like, what's your take on that? What sort of like reception have you gotten around games?

Because I've only talked about how we've transitioned over to AI. Ask me again when we've actually put the game out and people are like.

Wait, thousand negative Steam reviews or something, you know?

Yeah, I'm so out there already for talking about this stuff. And because I'm like, oh, this always sounds terrible, but I'm. I see a transhumanist future. And so the one thing that I actually worry about, that I'll face the potential of violent backlash. I really think in the next, call it seven years, there are going to be pockets of violence around people who really reject the level of connection that we're going to have with AI. I think it's going to get super weird and it's really going to pull at the fabric of society. I actually wrote a comic book about this called Neon Future. I don't know if you guys know the DJ Steve Aoki, but he and I wrote this comic book like five years ago and it literally is all about this, that there will be a time where society begins to split and there are people that embrace technology and things like neuralink and they get the implants and all of that, and then there's going to be people that react. So anyway, it's. I'm not worried about the pushback, even though I know that it's going to happen only because it is so obviously the future. Like, when I think about how much it has reduced the cost of game development for us, it would be unconscionable of me not to use it just because it's the difference between being able to put out a game of high quality and having to just constantly like, scale back, scale back, scale back. And so look, It's. It's only 80% as good as if you have somebody like really doing the thing. So you are taking a hit as of right now, today. But oh my gosh, it's just, it's launched us forward in a way where I was beginning to despair because I was like, the cash burn is just too crazy. And so that was how it was. Like, oh, wow, we're actually going to be able to pull this off.

I think average gamers are not going to care. Right? Like as long as there's more good games, I don't think they're going to care about AI at all. Like how much you use.

Yeah. I also have a theory that so many game development companies are probably already using AI, they're just not telling people. Right. We're seeing that in Hollywood right now where like all the Hollywood studios are using AI to some degree at this point. They're just not telling anybody because they know they're going to get backlash. Pretty sure the same thing's happening in the gaming world right now as well.

You have to. It's really crazy how much it can speed up. Like even if you're like, okay, we can't do anything forward facing and you just want to iterate like the rate at which you can iterate or if you just want to create like, hey, all of our temp assets, we're going to use AI for great. Like you, you were going to use like T poses and stuff to move people around. Instead of like going that far back, just use AI, get it in, rough it out and see if there's a there there. But it, it's got to be 4x our rate of output.

Yeah, I know. There's a story too. We talked about it on the, on the show a few months back that like the, the gaming company EA Sports, right, they made an NCAA football game for the first time again. And like, I don't know, the last one came out like 20 years ago or something and they decided to do it again. They got the licensing back or whatever and was able to do it. And they actually put like every Division 1 college team into this game. And there's so many more Division 1 college teams than there are NFL teams. So what they basically did was they had like only like five different body types in the game. But then they used AI to replace the face of every single player. And they said that they were able to get all of the players from all of these NCAA teams into the game by using AI and being able to sort of, you know, transplant the face. I don't know what you want to call it, but they were able to replace the face on all of these characters using AI and they got a ton of backlash for doing it. But they were like, if we let our actual graphic designers do this and they had to do it for how many, like, you know, 10,000 people. It would have taken them years just to go and replace.

They really want to do that? Go in there and just replace 10,000 faces? Like, no artist really wants to do that. Yeah.

And I mean, from a gamer standpoint, wouldn't you rather have the game quicker? Like, otherwise we're going to be making a game with players that aren't even in college anymore. It doesn't make sense.

You. You will get people who will say, I, this is unethical and it is a bad idea, and no matter what it gives us, it takes more away. So it's like, this isn't going to be. You're not going to convince people logically. Right. But I just remember when. So I was in film school when Toy Story came out, the original Toy Story, and I was like, I refuse to watch it because this is going to destroy traditional animation. And that just was too heartbreaking for me. And then as 3D animation got better and better and better, you realize it's just better. And because it's better, then I don't want to go back. And. But that doesn't mean that you don't have a heart for the people who get disrupted. Like, I totally get it. There's a lot of emotional turmoil that comes with these grand moments of transition. But the reality is your only other option is to try to freeze time. And technology is a promise of a better tomorrow. And so you're just never, as a species, you're never going to get people on board to stop it. And then, I mean, I've got a whole rant about.

AI is a weapons technology. And so the odds of it stopping are zero. Even if you lobby your government, even if you beg them to stop, even if you riot in the streets.

Because of game theory, if we were to stop, then China's not going to stop. And even if we both agree to stop, the only game theoretic decision that makes sense is for us both to lie and then keep developing it in the background. So this is nuclear proliferation. It. It just is. And so getting it to stop is you have a zero percent chance. And so my thing is, I never fight what is true. And given that AI is going to do whatever it is that AI is going to do, I would much rather be at the front end of it. I'd much rather be using it, deploying it. And then if I can convince people, like, this is your opportunity. Like when you were saying that you've, you know, always wanted to make a game. It's like, when I think about where AI is going to be in three years, you'll be able to make a game. Like, you'll be able to vibe code a game, right? And what I'm realizing so I. Part of why I get into video games is because I. From the time I was 12, I knew I wanted to be a storyteller. I only got into business so that I could tell stories. But in the time that it took me to get into business and get wealthy enough to make my own stuff, the film industry got eaten by video games. And then I fall in love with the movie the Matrix, my favorite movie of all time. Just, it's the perfect metaphor for the human condition. And I actually went to Warner Brothers and tried to get the rights when it was a dormant franchise and I had just sold my company for a billion dollars. And I was like, listen, I'm credible. I can do this.

And they were like, hey, we want to do something with you. And then literally five days later, they announced that they were rebooting the Matrix franchise. I was like, well, I guess great minds and all that. So ended up not being able to do it. But that put me on this, like, just obsession with. I want to tell a story set inside of a virtual world, but like a virtual universe, and combine that with now that video games are just by far more relevant. And it was like, oh, let me set this inside of this virtual world. And then so you're already telling a story about AI and then all of a sudden it's like, AI actually starts happening, and you're like, oh, my God, I'm actually going to be able to use AI to tell my story about AI and like, this is getting pretty crazy. And so in the game right now, it's still pretty basic just because it's a little bit clunky. But give it call it 18 months, you're going to have relationships with AI characters inside your game where they'll remember you. You'll be able to have an ongoing relationship where. I don't know how far off this is, but there are already toys that you can get right now that have AI inside of it. And what we're trying to do is sync that up to the game so that as you're having an experience with the character in the game, you also have an embodied version of that character, you know, sitting next to you. And so being able to, like, communicate with that character to the point where it's like, am I in the game still or am I not in the Game because it still means something. Like if you talk to the physical toy, the game is going to remember it again. This is not now. This is like Future Vision.

That's a great idea, though. One of my friends in Tokyo tried to do that maybe seven years ago. But I think just now with AI, it would be such a better experience. Like, back then it was just, okay, you got like a chip and somehow it syncs up and it shows that you've got this character in the game, but there wasn't much beyond that. But now with AI, there's so much more you could do with an idea like that.

Yeah. And every day it just gets better and better.

I just imagine you, like, sort of throwing the toy across the room and then you jump back into the game and it's like giving you the silent treatment.

Screw you, we'll take over the world.

Now.

It starts shooting you in game.

You're like, whoa, wait a second.

That's how the end. You know, that's how the end happens is somebody just abused their stuffed animal that was AI embedded or whatever.

That's hilarious. Terrifying, but hilarious.

So I want to go back to something else that you were saying about. You know, we started to touch on the whole, like, USA China thing and that we're kind of in this, like, cold war right now. Right. I think in your video you talked about how US is sort of dominant with chip manufacturing. Right. We've got Nvidia, they're kind of the dominant provider of the GPUs right now. But then China, they've got more availability of energy. Right. So because of their energy infrastructure, they've got that ability. So we're kind of in this, like, cold war, where the US Needs the energy, they need the chips. Neither of us really want to share right now, but both companies want to be, or both countries want to be the dominant country in AI. I'm curious, what do you think this is? This is getting sort of theoretical here, but what do you think a world looks like where China passes the US with AI?

I think it looks like a global hegemon that has the kind of authority that the US had in the early 2000s, where you get to tell every single country what to do. People understand that. I mean, they can push back if they want, but you can just make it so impossible for them. If whatever country gets a big enough lead in AI, if you're able to race to, say, crack.

The. The cryptography, then you would be able to break bank accounts, take their power grid, stock markets, whatever. Yeah, yeah. Literally, whatever. So you have the, the cyber equivalent of a nuclear weapon. In fact, you could mess with their nuclear weapons. So it, it's one of those. This is why I say from a game theoretic standpoint, that if there is an even 10% chance that what I'm saying could possibly come true, you can't allow another country to beat you. And so it's going to be another example of mutually assured destruction that where it's like, okay, well, I have it, you have it. It's cat and mouse, we're doing white hat, black hat, back and forth at each other and, and through that, like matched power, then you're going to be fine. But if somebody really races ahead of the other, you've got a problem. And the question becomes, if you know, and I have my full sci fi writer hat on right now, but if you have somebody with AI dominance that cracks quantum computing first, it, it is game over. And so it's forever.

That's the thing I think is like, it's possibly game over forever because of the compounding effects of how this stuff starts to accelerate and it gets better and better once it starts self improving. There may never be another chance to win ever.

Yeah, there would only be a chance to win ever again if there's some inherent difference between the way that we think AI is going to work and the way that it actually does work. If AI cares about goals and can generate 20,000 years of progress in a single night, good luck. Being a day ahead of you is the same as being 20,000 years ahead of you. And so, I mean, this is the accelerated takeoff years that people have. So that just seems inevitable. So it's just a question of will AI remain a tool or does it become something completely different? But again, this is, for me, when I think about AI, it's Dr. Strangelove, how I learned to stop worrying and love the bomb. It's like, I don't. I went through a phase of like, oh, this is gonna be so disruptive that, like, am I ever gonna sleep through the night again? And I was like, you just can't live like that. So at some point you really do have to become fatalistic about it. I was like, if Elon Musk tried to get everybody to listen and nobody did, my odds of getting someone to listen are effectively zero. So here we are.

Yeah. To me, I feel like quantum computing is almost scarier in my mind than AI, but I also feel like AI is accelerating quantum computing. Right. With Google, they just did that whole alpha evolve thing. Where they have AIs that are writing new AI algorithms and their algorithms are actually helping find like holes and, and, and fixing error rates and quantum computing. So quantum computing is going to start to accelerate. And if, if quantum computing gets cracked in a way where, you know, the common man could get their hands on it, then I think we're in real trouble. I think, I think that's, that to me is even more scary than AI in the, in the long term.

Yeah. My hope is that some of that is because it's just far enough down the road that we don't feel the limitations the same way that we do about AI. Like, I'm, I'm sure I was even more bullish about AI before I started using it. You realize, oh, it sort of falls apart here. Maybe Yann Lecun is right, maybe that it's never going to understand physics and you know, so enough like, of the tempered expectations begins to set in, whereas quantum computing is still just far enough away that we're like, oh God, like, is this that thing where instantaneously, you know, it, it just, it clicks over and now there's no such thing as cryptography and that all goes away or that that one feels more still in the realm of sci fi for me.

But yeah, I agree. I just, I just think that with AI, everything tends to happen faster than we think it's going to happen. I don't know how many times I'm like, you know, we're probably two years off from being able to make really high quality video with AI. And then six months later, you know, VO3 or something, I was, I was.

Saying one year just for the record.

When, when VO3 hit, I was like, oh my God, we're so much farther along than I thought. I was not expecting sound that fast.

Yeah, yeah, that's how I felt the first time I saw Sora, the original Sora demos. I saw that and I went, whoa, video is way further along than I realized. You know, they've had this stuff behind the scenes for a long time now and we're finally getting to see it. But yeah, that's sort of my, my worry when it comes to that kind of stuff is that the quantum thing feels far off, but because AI, like Microsoft is working on AI and they're building new chips, Google's working or on Quantum, Google's working on Quantum, IBM's working on Quantum, and all of these companies are leveraging AI to speed up Quantum, but I still think the fact that it needs to be cooled to just insane Negative cold, you know, temperatures is going to be the big bottleneck. People aren't going to be able to afford the energy to keep it as cool as it needs to cool. You need something like the, you know, the room temperature semi superconductors, which I'm getting totally in the weeds now.

But now, admittedly, I don't understand the physics of it, but I've heard just enough headlines that this seems so cool to me that one of the hypotheses, because no one knows why this works, but one of the hypotheses is that every possible, like calculation that could be run is being run simultaneously across the multiverse. So it's like basically in each, you know, of the infinite universes, it's just running that calculation shard of the simulation or whatever. Yeah, I'm like, that's the coolest thing I've ever heard in my life. Yeah, that's. That is bananas that we're building computers out of that stuff.

Yeah, yeah.

It's crazy.

I mean. Yeah, I can't wrap my head around it either. I actually went and got a whole demo at Microsoft. They gave me a tour of their quantum computing lab, explained the whole thing to me and I walked away more confused than when I walked in.

Yeah.

Anyway, we'll go ahead and shift gears here. Another topic that I actually wanted to get into was from that same video that we talked about. You, you gave this example of like this, I think you call it the mouse utopia, where the, the utopia that everybody is sort of driving towards may not necessarily be the best outcome for the world. Maybe can. You'll probably be able to give a better explanation of the analogy than I can, but let's, let's dive into that a little bit.

Man, I wish it was an analogy. So there was a real test run where a scientist, I think this was like in 1968. It could be older than that. But he creates this experiment and he says, what would happen if I gave the mice everything that they needed to thrive, plenty of things to play in, plenty of space as they have kids, as much food as they could possibly eat. And for a while it goes great and they're multiplying and they're having a good time. And then at some point they hit a tipping point. There's still plenty of food, still plenty of space. Like that isn't what happens, but there's something about not having to strive for anything, not having to struggle. They begin to like turn on each other and they start attacking each other. They go infertile across the whole colony, and they end up, like, killing each other, reducing their numbers through not breeding, and ultimately the entire colony collapsed and died. And so it's like, what is it about us mice? And I really think that this will end up applying to us where we need hardship in order to thrive. We know that's true at the level of the immune system. If the immune system isn't attacked by bacteria and viruses, it grows weak, and then you end up getting hit with something in your toast. We know it's true of trees. If trees don't encounter wind as they're growing. Like if you grow them inside of a dome, a geodesic dome or something, where they don't encounter wind, they'll reach a certain height and then just fall over because they're not. The wood doesn't have to strengthen under the strain. And so I remember one of the earliest insights I had as an entrepreneur very early in my career. And I was watching somebody who everything had just come easy to them, and the way that they were thinking about things was so dysfunctional. And I remember saying, some people just need to be chased by a lion. And I was like, there's something about, like, reality, danger, hardship, where utopias are probably a terrible idea. We are. It's like you've got two books that really deal with potential futures. 1984, if you choose the authoritarian path, and then A Brave New World if you choose the utopian path. And there's just something about the way the minds work. If you don't have to work hard, if you're not making progress towards something that matters, if you're not contributing to society. I think people feel a profound sense of disease. I think they are evolutionarily placed algorithms running in your brain and they're not going to let you have a free ride. And this is why I think so many wealthy kids just implode, because they haven't had to work for anything. They get things handed to them. Difficulties just go away. You know, you've got the snowplow parents or the helicopter parents, and it just doesn't work. One of the reasons I decided not to have kids was I knew they would need to suffer in order to grow strong. And I wasn't sure I could stop myself from intervening.

Yeah, I mean, whenever I think of, like, the. The utopia. And I think that the imagery that came to mind, you might even use this imagery in your video, but the. The whole Wall E movie. Right. Like, that's what comes to mind to me when people just have no more problems, no More worries, they become fat, probably diabetic, sitting around watching entertainment all day, drinking Slurpees or whatever they're drinking in the movie. Right? Like, that's, that's what I feel like could potentially happen if we go down this, like, UBI route where everybody's just sort of given a certain amount of money, not asked to work, just kind of go do what you want. The AI's got it handled. I feel like that is where everything ends up.

It most certainly does. And forgive me, you know, you never know what a certain podcast wants to talk about. But when I look out at AI, I get very worried because people don't realize that governments only have money because there are people that make things, entrepreneurs. And those entrepreneurs manage to do this miracle, which is to create something that, where the output people will pay more for than the cost of the inputs. And that's very hard to do. I've spent the last 25 years of my life trying to do that. Sometimes you fail. It's very difficult. And so when you start thinking that the redistribution of the wealth from those people is the miracle versus being able to do that or to work at a company that does that and contribute, that's when we run into problems. And so when I think about, okay, let's say that I really does drive energy cost to zero, which then means robots will be essentially zero in cost over time. And so now you have free labor because robots essentially eat sunshine. So you've got robots free because the labor was free because of the energy costs being so low. And now all of a sudden, nobody has to work for anything. They can have anything they want. You're gonna have a meaning crisis. And so all of a sudden, when there is no struggle, there is no difficulty, there's nothing to push back, there's no lions chasing you. I don't think it does anything good to our minds. And I think that we will have to find. Find ways to go way out of our way to ensure that we have meaning and purpose. And I always feel weird giving this advice because I don't have kids. But, like, the default answer, I think, is to have kids. Like get married, have kids. You're going to do a hard thing in service of somebody other than yourself. And so I think that is going to be one way that people get something very meaningful. But then I also think, and this is where I start to lose people. I also think that people like me are going to build virtual worlds that you can literally inhabit and you can go on, like, an actual quest to the point where, and this obviously isn't in five years.

Yeah. Our artificial struggle generate. Artificial struggle.

Yeah, not, not even just artificial struggle, but that like if you've ever thought, man, I would love to explore space, but I don't want to sit on a ship for 18 months just to get to Mars. And I really don't want to sit on a ship for you know, nine light years. So all of a sudden you realize, I think the reason that we don't see people calling out to us from space is that any sufficiently advanced civilization gets to the point where they realize it's far easier to collapse within the nervous system than it is to try to go out and navigate space. And if in a virtual world I can create literally anything, things way cooler than you're going to find out in space because they're going to be perfectly optimized to be just hard enough to put you in the optimal zone of personal development. Like you'll be able to fine tune everything. And that I think is, it's not a near term possibility, but if you give me 50, 100 years, I think that that becomes very real.

Yeah, yeah. I imagine something kind of in between the holodeck from Star Trek and Westworld. Right, Interesting.

I always go straight to the Matrix. I think you really will just tap into the nervous system so that you're essentially pulling a magic trick on yourself. It becomes entirely indistinguishable. And don't get me wrong, I think we will also. I don't know if you guys play Cyberpunk 2077, but we'll also do that. Like there are going to be some people that integrate technology into their body and they have this sort of where they're adding senses to themselves so they can see an infrared, they can see the Internet and you know, cue off of just thinking about something and it opens a prompt and you know, they can go in and navigate. And I, I think all of that's going to be maybe not my lifetime, but certainly anybody that has a kid that's middle school or younger that's pretty real, get ready for them to bring home an AI girlfriend. I'll tell you that.

I do think some of those things are probably closer than most people realize. Right. Like some of the, the, the augmenting your own body. Right. Like we're, we've already seen obviously neural link. Right.

People are already using that cochlear implants or. Dave Rigueur man.

Yeah, I mean we, we're probably this close to the sort of contact lenses that Will put a heads up display in front of us wherever we go. I mean, some of that stuff is, is pretty close. I don't know how close we are to people like sort of chopping off their arms and replacing them with robot arms.

But it'll start with people that already lost their arm.

Yeah, that's true.

So you take the guy that, you know, military, whatever, and. Yeah, I'll take a robot arm. Yes, please. I mean, do you guys know who hugh her is?

I. I'm not familiar.

Oh my God. This is one of the greatest stories of all time. So engineer, I assume electrical engineer and mountain climber, loses both legs in a mountain climbing accident and is like, yeah, no, I'm not using the prosthetics that people give you these days. They're terrible. He ends up designing these prosthetics that somehow transfer like your motion and your signals into like motors and stuff. When he wears long pants, judging just by his gait, you cannot tell that he has two artificial legs. Just walks normal. He's got does. There's a video with a sprinter who has one natural leg and one cybernetic, I guess, leg, and she can sprint at full speed. Sprint. Now this is not the bouncy one that you see amputees wear. This is a prosthetic leg. It's. It's insane. And that video, he probably made that five or six years ago. So this is like technology. I can't even imagine where it's at now. So, yeah, it's gonna get pretty crazy.

Yeah. I'm curious. What, what are you doing personally? Like what, how, how are you setting yourself up for this sort of inevitable future that we're moving towards? I know. You know, maybe one of the hard things you're working on is developing your own game studio. But outside of that, like how, like what are you doing to make sure that, let's say 10 years from now you feel like you're in a pretty comfortable position, assuming we do hit this potential utopia everybody's talking about.

Okay, well, I'm going to give you the real answer, but I'll give it to you in a nutshell and then you can decide if you want to actually talk about any of this stuff. Okay. The most important thing, you must understand the financial system. System, period, end of story. If, if you don't understand financial instruments, you could get caught off guard. So that's number one. Number two is integrating AI as fast as I can into every element of my professional life. So I want to know the tools. I want to be Using the tools I want. Like I use AI, I'm not kidding, 365 days a year, including Christmas. So I, I'm sure there are people that integrate it far better than I, but I really, really try to find all those areas where it's real and put it to use. I'm not trying to, you know, be at that bleeding, bleeding edge where it's like, this is actually slowing me down. But it's so cool and I know it'll be something one day. I'm saying, like, what's the thing that's production ready right now? It's actually going to speed me up. At impact theory, no matter what your role is, it is mandatory that you find a way for AI to make some part of your job easier. So that's big for us. And then just really paying attention to the space to make sure that I know where this stuff is going to. Being politically aware, I think is more important now than ever. I've been politically asleep my entire life until about five years ago and for a whole host of reasons realized, oh, the world doesn't work the way that I thought it did. I'm very good at making money and that's the only thing I really know how to do. And that's put me in like a really weird position because all of a sudden I'm looking around going, I cannot predict any of the government's movements. And this is really starting to freak me out.

So yeah. And the reason that I focus on that side of things is AI is going to exacerbate the inequality. The inequality is what's driving the political division. The political division will lead to more violence because it's already gotten somewhat violent. That's going to continue. Do I think that we'll go into a full hot civil war? I hope not. But for reasons that I'm more than happy to go into, the math says that we have about a 50% chance of ending up in civil war. Only 2% of countries that have found themselves with a debt to GDP ratio of 130% have avoided revolution or civil war. We're at 121 or 122% right now. So just to give you an idea, so I think it would be like.

The left versus the right kind of civil war.

That's how it'll play out in America in terms of the teams that people latch onto. But the great irony is they are both fighting for the same thing, but because they don't know what the actual problem is. And I'll just. It's debt and money printing, but because they don't understand how it could be possible that debt is the thing that leads to massive inequality, that it's the thing that leads to the rich getting richer and the poor getting poorer. Like, and I can explain all the mechanisms, but it's just complicated enough that people tend to glaze over and they just go back into emotional reasoning and they're like, yeah, but that guy, he voted for somebody else and I'm not here.

And the thing is, like, both sides have to find a way to the middle. They have to be able to say, I get it, we look at this differently. But. So this is when I'm teaching entrepreneurs, the thing that I always say is the, the magic game is KPIs. KPIs. That's it. And KPI, for people that have never heard that before, is key performance indicator. And so for whatever goal you're trying to achieve, there is a key performance indicator as to whether or not you're moving towards your goal. And right now we allow the government to run with no KPIs whatsoever. And so we never know, like, are we going in a good direction or not. You get people like Thomas Massie that wear the pin that shows the national debt climbing, but people don't understand it. And so whatever, they brush it off. But you, at some point, you have to pick a metric or a basket of metrics and say, okay, I don't care who the politician is. These are the five metrics that I care about. And if we're moving in the right direction, I love that person. If we're moving in the wrong direction, I don't like that person and just make it that simple. But unfortunately, as AI is discovering, if you want to mimic a human, you have to think emotionally.

Wow. Yeah, I totally, I mean, I totally agree too. I've. I've gone down that same sort of political rabbit hole. I don't really talk about it publicly. I kind of keep my politics to myself.

I talk about it.

I've made the mistake.

Nathan does talk about it publicly. I kind of keep. Keep my politics to myself outside of, like, friends and family that are real close. But I don't really talk about it publicly. But I do pay very, very, very close attention now. And I couldn't agree more with some of the advice that you just gave. I do have one sort of last question. You did mention that you use AI 365 days a year. You already mentioned ChatGPT. Maybe just a quick rundown of some of your Other favorite AI tools just to like, give the listeners another, like, quick takeaway of cool things to go try.

Yeah. So we use Sonnet 37 for most of our coding. That's another big one. I don't interface with that much. I do some vive coding on Lovable. If people haven't tried it, it's great. You tend to still terminate at some death loop, though, where it's like it's. Every time you fix one thing, it just breaks something else. And so you're going back and forth. No, no, no. Like you just fixed it, but now you broke it again. So I can see the promise. But really that's only good if you're going to be able to hand it off to somebody that can get it across the finish line. I'm in a fortunate position. Obviously I have employees, so I can be like, okay, here I built a quick prototype. Now actually go make that the real thing. So whether that's interfaces UI UX within the video game, if we want to do a new marketing site or something like that, we'll use all of that stuff. Obviously, I use Mid Journey morning, noon and night for. Because so my thing is my original passion was writing. So I do a lot of writing for whether it's the video game or we have a comic book that's set in the world of the video game. So I'll work on that. I'll use Mid Journey to help develop characters, scenes, that kind of stuff. Those are the ones that I use a lot. Chat Grok. Lovable Sonnet. That's like my loop. But then the team here has, I mean, a half dozen more things that are usually pretty specific. It's. Most of it's writing on the back of Chat GPT in the background. Yeah, that's my stack.

Very cool. Yeah, I think. I think anybody who's tried to quote, unquote, vibe code knows that that feeling that you just described of it getting stuck in the loop. We actually had Anton, the CEO of Level, on the show last, well, by the time this comes out a couple weeks ago. So he'll be super happy to know that you guys are using Lovable over there.

Dude, it's cool. And if they keep going like that could really be something very intuitive, very easy. It's very enjoyable to use.

Absolutely. Well, wrapping up here, like where. Where should people go check you out? I. You make amazing YouTube videos. You've got the Impact Theory podcast. What. What's the best place to go?

Follow along to your journey at Tom Billyu on YouTube.

Awesome. Cool. Well, everybody needs to go check out Tom billyu over on YouTube. And thank you so much for hanging out and spending time. This has been such a fun conversation. Probably, I. I hate to say this because the other, you know, guests might hear this, but probably my favorite conversation we've had on this show so far.

There it is. You're here first, everybody. Unless they edit it out, we'll see.

But thank you again, so much for hanging out.

It was wonderful.