What Are We Then?

Last year the British AI founder and writer, Emad Mostaque, put a number on how long most of us have before our jobs become economically irrelevant: about a thousand days. He did that in his book The Last Economy, which has since become an Amazon bestseller. By his count, that leaves roughly six hundred days before we are all out of a job. When I sat down with him at TechBBQ in Copenhagen in August, I expected him to soften that estimate. Much to my surprise, he went the other way.

Emad is a mathematician out of Oxford who ran a hedge fund for years before founding Stability AI in 2020, the company behind Stable Diffusion. Stable Diffusion was one of the first open image models, and it put generative AI on everyone's laptop before ChatGPT became a household name. He left Stability in 2024 and started Intelligent Internet, with the explicit goal that countries and citizens should be able to own and run AI themselves. To get there, the company builds the open pieces that sovereign AI needs: infrastructure, agents, datasets and smaller specialised reasoning models built on open base models.

His book The Last Economy came out last year, and this spring he followed it with an essay series called the Commonwealth. Most of it weaves together mathematics, finance, deep learning and philosophy into one vision of where the world is heading as AI enters our schools, workplaces and everyday lives.

As we sat down I wanted to make sure I got Emads perspective on core issue, which I wrote about in the Nowable Manifesto: what happens to the capabilities we have built our identity on, when machines become better at them than we are?

Lars: What do you see happening to the capabilities we've built our society on, our human capabilities and our institutions?

Emad: We've moved up the scale in our work, from labour on the land to industrialisation to the information age, and now this final intelligence inversion. The question is, where do we go from here? A lot of people think the upper limit of AI is the human corpus it's trained on. Unfortunately, the upper limit is a human who doesn't need a coffee, is already operating at the best of their performance, and works a thousand times faster. In my book last year I said there were about a thousand days before your job became economically irrelevant. I might have been a bit too optimistic on that one.

Lars: Even though it's still a jagged technology? Good at some things, bad at others?

Emad: The jaggedness is almost disappearing. I'm a mathematician, and you see breakthroughs on conjectures that have puzzled mathematicians for fifty years. Some of my friends are PhDs working on these topics, and they're like: what do we do now? The AI can do it better.

This is the question I raised in The Nowable Manifesto. For most of history, being capable meant knowing things. The lawyer knew the law, the doctor knew the symptoms. With the internet, it came to mean finding and combining knowledge faster than others. If machines now do both better than we do, what is left for us to be good at? I asked Emad how he sees that playing out.

Lars: One thing is what the technology can do. But how do we handle it as humans?

Emad: The diffusion of technology is the key thing that stops it from proliferating everywhere, given how capable it is. When artificial intelligence meets human reality, it can be a bit messy. But this is such a general-purpose technology that it will permeate every part of our lives. And we have to consider: what do we do when we're the dumbest people on the team? The real meaning of life is something we need to re-examine, when your work identity can't necessarily be part of it, because the AI can do it better than you, a thousand times cheaper.

Lars: For a long time we've built our identity on what we do. How do you see that changing? What are we then?

Emad: Classically, you would have Emad, son of Khaled, son of this. We had network-based identities. We had strong communities. We need to return to that and tell better stories. Because if you define yourself by a role, what school and our current job system do is turn you into a robot. And so it's not surprising a robot can do your job better.

But you're more than that. People ask me, what should I teach my child? That they have agency. Get them to build with this technology, and they realise they're capable of so much more. Before, it was a bit scary to say, I can't make music. Now you use Suno and make an album, and turn it into a music video. The floor is being raised for everyone, which means we can achieve more, and we can connect more as well.

If the floor is being raised for everyone, the next question is who owns it. Nobody wants a future where the intelligence we all depend on is rented from a handful of American labs.

Emad's answer is that Europe should stop chasing the frontier and focus on what he calls competent intelligence: smaller, open, specialised models that citizens and institutions can actually own. Intelligent Internet has a name for the structure that would make this happen in each country, the Champion, and a plan that ranges from a personal AI agent for every citizen to who should own the robots.

In the full episode of the Nowable podcast, Emad explains how the Champions would work, how he would measure value when GDP stops making sense, and why he thinks you can't rent your cognition. We also explore his concept of “improvement is capture”, the idea that every improvement we hand to the machine can be a small transfer of control. 

[Listen to the episode →]