It does strike me as surprising that a lot of pundits and populists keep blabbering on about the utopian/dystopian future AI will bring.

Yet most experts in the fields are stuck. Constantly trying to improve the accuracy of relatively simplistic models that can accomplish relatively basic tasks.

Something like a GO playing AI is simple to create, since GO is a very “inhuman” thing to do, it’s much more suitable for a computer program.

But we are no closer to “general AI” than we were 40 years ago, not in any significant ways. We are not even closer to defining what “general AI” means in a meaningful way.

That’s not to say that some capable people aren’t working on the subject. But some very intelligent people had been working on finding an energy-free way of turning iron into gold for thousands of years… and guess what, a lot came of that research, but not a way to turn iron into gold.

Instead (if I can insert a shameless plugin), I’d recommend that we focus more on the God machines that exist right now, filled with old software and running on old hardware. They could change the world as we know it in a matter of minutes.