I don't think so, AIs are going to freeze the tooling to what we have today since that's what's in the training corpus, and it's self reinforcing.

Nah, the good LLMs can generally web search and read documentation well enough that the fact that pre-training isn’t up to the minute is not a serious concern. Badly-documented projects are more of a concern, but they weren’t likely to get much pre-AI usage either.