Every weekday, one foundational AI question gets a real answer here. Not a thinkpiece. Not a hot take. Just an honest attempt to answer the kind of question you're slightly embarrassed to admit you don't know — because everyone seems to assume you do.

The questions are things like: When does fine-tuning actually beat prompt engineering? What does "context window" really mean for retrieval? Why does temperature matter, and what does zero temperature even do? Foundational questions. Definitional questions. The kind that expose exactly how much is still being figured out by everyone, including the people building these systems.

A daily agent surfaces candidates from the AI news cycle — things that are timely, contested, or quietly assumed. I pick one, sometimes edit the framing, and publish. Each answer links out to the best further reading I found.

The name is intentional. The dumb questions are the important ones.