@siracusa Hearing you argue with Marco on ATP about the quality of AI for non-programmer tasks. I’m afraid that a lot of people don't care as much as you (and me) about the last 20% of quality and will adopt AI for almost anything as it is « good enough » to them.

There is probably a Venn diagram to draw between people skeptical of AI and those who know everything about the pixel technology in their TV.

@guifort Unlike TV picture quality, that “last 20%” can suddenly become very noticeable to even the people who “don’t care” when some LLM output ends up being wrong in a way that has an impact on their life. (E.g., “This mushroom is perfectly safe to eat”)