Exploring technology, AI, and possible futures
This is probably draft thirty-something of this essay. I have written it as a UBI essay, an immigration essay, a capital-versus-labour essay, a housing rant, and every time it either blew up into a book or collapsed into a rant. So I parked it and went off to publish easier things instead.
Conversations I keep having with friends about the West keep revolving back to the same frustration. We talk about housing, trains, energy, defence, robots, supply chains, data centres, even basic public services, and then we hit this wall where everyone agrees the thing should exist, everyone can describe the thing in a PowerPoint, and then the thing either takes ten years longer than it should…
Alignment talk usually lives in technical papers and LessWrong threads, corrigibility and reward modelling and all that. What I keep turning over in my head is much less about clever schemes for steering models and much more about the way those schemes look in ordinary lives once you push them into products and laws and the tools people use every day.
A while ago I was swapping screenshots with a designer friend. We were both poking at the new generation of model based tools. His complaint was simple: “Everything looks like a SaaS landing page from about five years ago.” Purple gradients, friendly avatars, soft cards, the usual pattern from old dribbble boards. If you gave the system very careful art direction, you could drag it out of that…