prompting

Chain-of-thought: why making an AI think out loud makes it smarter Lesson

Asking a model to work through a problem step by step, instead of blurting an answer, dramatically improves it on hard tasks. Here is why that simple trick works, what it really buys the model, and where it backfires.