Once a computer can do something, it no longer called AI but just an algorithm.

At least, that used to be the case before the current AI summer and hype.

If you have an agent in an environment, the program that controls its choices has pretty consistently been called AI even when it's a simple design.

But I'm skeptical of calling most optimizers AI.

Back in the days when something like A* was invented, I' fairly sure it was called AI. Now it's just another algorithm. So indeed, we don't call those 'AI" anymore. The bar has significantly shifted and does so each time a big advance is made.

Perhaps one day we'll understand what goes on in our brains to the same extent.

The bar has shifted a bit, but I don't think it has shifted all that much.

I would not say the bar is moving along with "things computers can do".

Once a computer can do something, it's just an algorithm. LLMs can't really do anything right, so they're AI. ;)

> [...] the classical sense.

Which one? Fuzzy logic, ANNs, symbol processing, expert systems, ...?

It's always entertaining to watch the hype cycles. Hopefully this one will have a net positive impact on society.

Chapters 4, 6, 10, and 11 of Russel and Norvig’s Artificial Intelligence text all fit this bill.