I think Yann Lecun was right about LLMs (but perhaps only by accident)

The degree to which AI can generalize previously trained networks to novel tasks is growing with time.

The degree to which I care about that has not.

It just means we can get better inference with less targeted models. Whoopdy doo