What A World Governed By AI Could Actually Look Like

**Emad Mostaque** (0:00)
In the next 20 years, I think you leverage artificial intelligence owned by the people, and the robots owned by the people to eliminate hunger, disease, all these other things by collecting our collective knowledge, and having humans and AIs work together.

**LG Doucet** (0:13)
What's up, everybody? It's LG Doucet here, and welcome to Milk Road AI, the daily AI show that still sits down to write an original opening line every single episode, and hasn't yet found an AI that can write a better one. Today is June 29th, 2026 We're recording on the 23rd. Today's guest has spent years thinking about where AI is taking us, not just in technology, economy and politics, but even on the philosophical side. In his book, The Last Economy, he argues that AI isn't just another innovation cycle, it's the beginning of a transition that could reshape work, wealth, and what it means to be human. So today, we're exploring what happens next and how investors should prepare for it with Emad Mostaque, founder of Stability AI and Intelligent Internet, and a contributor on the Moonshawts Podcast. And hey, if you're looking for your next AI investment idea, our analysts have called Nebius, Micron, AMD, and even Bloom Energy way back in Q1. You can see what they're calling next in Milk Road Pro, which is just a dollar for the next seven days. The link is below. Today's episode is brought to you by BitGet, stocks 2 with real liquidity, real dividends. Emad, welcome to the show, man.

**Emad Mostaque** (1:12)
Nice to meet you. Thanks for having me on.

**LG Doucet** (1:13)
Okay. So I just want to dive right into it because one thing that really stood out about you to me is that you were early, years ago, when you were warning governments, policymakers, that it's like, you need to get a handle on this before a lot of these new models come out. And since then, we have models that now are being restricted by the government because they are just too potent. Is this kind of the timeline and the result you expected years ago as you saw what was going on?

**Emad Mostaque** (1:44)
Yeah, pretty much. Like I think in 2022-20, I think it was 23, I said no more programmers in five years. So we're looking pretty good on that.
You know, when I left Stability a couple of years ago, we did the world's leading giant of AI models for media. I was like, we got to decentralize now, we've got to start because otherwise you'll have systemic control and you won't have access to this intelligence once it gets to a certain level. And you will start to see slowly and then all at once massive disruption in the way we live, our own meaning of life and society itself. And I think we're at that tipping point now.

**LG Doucet** (2:22)
Is this, are you surprised that there's been models that need to be restricted? Like is that, is that, has that come as a shock to you? Is that earlier than you thought? Because I mean, it's just pretty close to your timeline.

**Emad Mostaque** (2:32)
It's pretty much just one of my timeline.
Like you could just extrapolate, like we were building code models in 23 and we saw the capabilities coming. And you know, most of the Internet is like built on spaghetti. You know, like you're either compromised or you don't know that you've been compromised is one of the jokes in cyber defense.
So it was always going to be the case that you would have this like consumer class of models that we all get as they were testing and not optimizing and you were going up the capability curve. But then when you get to actually competent intelligence, which is what I'd call Claude Sonnet 4.5 and more. So the anthropic model from the end of last year, that kicked off the open-claw wave, and you get to actually good intelligence. Well, if you can replicate a good worker, then you have a qualitative and quantitative advantage over everyone else. Obviously, cybersecurity is one of the key areas on that. But then beyond that, it is your means of production. So it's not surprising to me at all. I think the first shot across the bow was when OpenAI achieved a gold in the math Olympiad a year ago, and we still haven't really seen that model. Like you have this bifurcation of the really big models, and then the models that you'd get for everyone else, for the other billion, like let them eat tokens.

**LG Doucet** (3:46)
Is this, or do you have any new predictions that we can look at or anything you're thinking about? And just for people listening, that is largely the theme of today of talking to you. But is there anything currently on your plate that you're looking at of thinking like by 2030 we'll have this similar to your predictions from a few years ago?