Some of this AI criticism is ridiculously hyperbolic. Sure, sometimes it makes mistakes and I don't believe that simply making a bigger LLM will somehow result in a godlike superintelligence in the near future. I don't think that the stock market believes that either.

But regardless of that you've got to admit it is actually useful for lot of tasks and therefore has huge economic potential. You could only say that it's useless "for most tasks" when you define that as tasks that nobody would think of using it for - it's not going to dig you a ditch. But how can she say that it's both no good at most things and simultaneous dangerous because of the job losses it will cause?