> This is an extremely common fallacy ... they use the fact that "some humans can't do X" to claim that LLMs are, in fact, basically conscious/human-like/AGI already.

Claiming that LLMs are conscious or human-like because humans can't do X seems a very strange way to argue for LLM intelligence.

Usually, it goes the other way around: an LLM sceptic says "LLMs are dumb because they can't do X" and soon someone has to remind them that also most of the population can't, in fact, do X.