Why so little discussion on the capability of AI models in investing subreddits? Isn't it the most important question right now?

The performance of the S&P and just about every stock talked about on this sub in large part comes down to if AI models will be able to replace/augment economically useful work. It doesn't matter if MSFT is below it's historical PE or that profits were up X% last quarter. If LLM's are a dead end and we at best only see incremental improvement that is incredibly costly to train, all the stocks being recommended by Redditors will implode and to a lesser extent, the S&P along with it. Even the companies selling the shovels will have a rough 5 years if we don't continue to see absolutely massive growth of AI adoption, and we're only going to see that adoption if models improve. If LLM's are likely to dramatically improve however, then you should be dumping your money into semis/hyperscalers.

It's weird since Reddit is generally very anti LLM, but is everyone here just taking it for granted that models will significantly improve when they recommend these companies? Or is it that people here don't realize that just how fucked these companies are if LLM's are a dead end? If we're going to spend 1.5 trillion in cap ex on semis, then hyperscalers need to sell that compute for say 2 trillion, and the companies buying that compute need significantly more than 2 trillion in revenue from it.

I feel like this sub should be focusing on reports of AI adoption, METR reports on AI capabilities etc.