> Current AI is never going to do better than us in every category, it can only do better in us in categories where it's possible to generate an infinite amount of verifiable synthetic data (math and programming).

Even if this turns out to be true, I don't think it will necessarily matter.

So much money was already burned, by so powerful people in incestuous relationship with governments, that I don't think this is an important point anymore.

> AI is still just as bad at sales, marketing, writing and anything involving human subjective appeal as it was a year ago.

All of these have objective metrics (units sold, clicks, likes, etc) and are thus verifiable in principle.