The physics prize seems to have gone to ML/AI work …

I misunderstood your post. It was work *on* machine learning, not work created by LLMs... Machine Learning has many useful applications so that's entirely valid. I'm not sure why it's not the maths prize, though...