Eh, we've gone through a lot of hype cycles before and the field still exists. For example, deep learning was hyped to replace all feature engineering for all problems and then NLP would be trivialized. In practice, that was overhyped and you still need to understand NLP to get value out of deep learning for NLP. And in practice, there's still quite a bit of feature engineering (and practices like it).

I think LLMs will turn out to be similar. They'll change the way we approach many problems, but you'll still need to understand both LLMs and more problem-specific aspects of ML.

Back to your question, if you enjoy AI/ML and you're worried about jobs in a few years, I think it's still worth pursuing your interests.

If anything, the bigger challenge in jobs in the next year or two is the current job market.