I actually like that analogy. It's somewhere in between. Enough that LLMs can help in many ways, but the current models are still far away from doing everything.