Earlier quoted context omitted.
Is your argument that the world economy is running so close to theoretical maximum efficiency that AI can't improve it?
More likely that any ideas AI comes up with will be "mid" and face the same implementation obstacles that already exist, coupled with the majority of energy burnt up by AI will go to frivolous ends, making personalised dog whistles, derivitive elevator muzak, et al.
That's plainly false, as that's just not how markets work over the long term. We only see distortions like this iver the short term due to investor subsidies.