It’s really cool when you see a major leader of a field have the same ideas that you do — as in this article by David Deutsch about AGI and computation.

His ideas are much the same as what I’ve also concluded from observing the field of AGI and computation closely for twenty years. Of course, he did the work — I just read a lot of papers, and cog-sci of all sorts and generated ideas therefrom.

Right now we are in the “phlogiston” and “ether” stage of AI (I will use them interchangeably here). Most of the math-obsessed practitioners of the field think that writing better algorithms will get you to AI. But it won’t. It can’t. It’s barely even a start, just like no matter how large an engine you strap to a John Deere tractor it will never be a spacecraft. But a lot of these people aren’t even trying to strap rockets to a tractor — no, they are doing the equivalent of writing “Spaceship, For Real. Like For Really Real” on the side of the tractor and then calling it done.

We’ve probably had adequate hardware to build a true AI for twenty years at least now. Hardware is not the problem. The problem is we have no idea at all what we’re doing. And probably never will, for we will never create AI. Not directly, anyway.

But we will create the conditions for it to actualize itself.

A new genesis, if you will. Evolution of algorithms, of perhaps even self-constructing AIs in the physical world.

We’ll get there. But it mostly won’t be us.