What's this about?
Ethan Mollick's blunt advice — "Prompt crafting is over-rated, ask for what you want" — paired with his pointed question about the conjecture disproofs: who deserves authorship, the person who typed 58 words of prompts, or GPT-5.6 Pro?
What exactly is "prompt crafting"?
The folk art of magic phrasing — "think step by step," role-play instructions, elaborate templates — that mattered for weaker models. Current frontier models mostly just need a clear statement of what you want; this echoes Anthropic's own guidance this week that over-engineered prompts now *hurt*.
Why is it worth noting?
The two tweets together sharpen the day's theme: if lazy prompts can disprove 30-year conjectures, the scarce skill shifts from phrasing to knowing what to ask for — and credit allocation gets genuinely murky.