I try to be optimistic as well. But obviously horses are almost exclusively a hobby today. The work horse is gone. I think the problem is political to a part, if we manage to spread the wealth AI can create we are fine. If we let it concentrate power even more it looks very grim.

B2C businesses need consumers. If AIs take all the jobs, then most of the population-minus the small minority who are independently wealthy and can live off their investments-go broke, and can’t afford to buy anything any more. Then all the B2C businesses go broke. Then all the B2B businesses lose all their B2C business customers and go broke. Then the stock market crashes and the independently wealthy lose all their investments and go broke. Then nobody can afford to pay the AI power bills any more, so the AIs get turned off.

And that’s why across-the-board AI-induced job losses aren’t going to happen-nobody wants the economic house of cards to collapse. Corporate leaders aren’t stupid enough to blow everything up because they don’t want to be blown up in the process. And if they actually are stupid enough, politicians will intervene with human protectionism measures like regulations mandating humans in the loop of major business processes.

The horse comparison ultimately doesn’t work because horses don’t vote.

I am somewhat confident that horses are going to replace cars and tractors pretty soon, possibly within my lifetime and quite likely with my son's.

He's going to learn how to drive (and repair) a tractor but he's also going to learn how to ride a horse.

Apparently human chess grandmasters also need “oversight” from airplanes, because without those, essentially none of them would show up at elite tournaments.

In the case of horses and cars, you need the same number of people to drive both (exactly one per vehicle). In the case of AI and automation, the entire economic bet is that agents will be able to replace X humans with Y humans. Ideally for employers Y=0, but they'll settle for Y<<X.

People seem to think this discussion is a binary where either agents replace everybody or they don't. It's not that simple. In aggregate, what's more likely to happen (if the promises of AI companies hold good) is large scale job losses and the remaining employees becoming the accountability sinks to bear the blame when the agent makes a mistake. AI doesn't have to replace everybody to cause widespread misery.