Unless some math friend of mine wants to tell me they're actually bullshit, these (what seem like) genuine advances that LLMs are making recently are both (a) really impressive, and (b) put some pressure on those of us (including myself) who are skeptical of the cognitive capacities of these things
