If it is just a tool, it isn't AI. ML algorithms are tools that are ultimately as good or bad as the person using them and how they are used.
AI wouldn't fall into that bucket, it wouldn't be driven entirely by the human at the wheel.
I'm not sold yet whether LLMs are AI, my gut says no and I haven't been convinced yet. We can't lose the distinction between ML and AI though, its extremely important when it comes to risk considerations.
Because no one defines AI the way you seem to do here. LLMs and machine learning are in the field of artificial intelligence, AI.
How is ML a subset of AI?
Machine learning isn't about developing anything intelligent at all, its about optimizing well defined problem spaces for algorithms defined by humans. Intelligence is much more self guided and has almost nothing to do with finding the best approximate solution to a specific problem.
> Machine learning (ML) is a field of study in _artificial intelligence_ concerned with the development and study of statistical algorithms that can learn from data and generalise to unseen data, and thus perform tasks without explicit instructions.
https://en.wikipedia.org/wiki/Machine_learning
You are free to define AI differently but don't be surprised if people don't share your unique definition.
Silent down votes, any explanations or counter points?