Lesson 9: Step-by-step reasoning (chain-of-thought)

When you ask the model for an instant answer to a problem that needs several steps, it sometimes 'blurts' a guess and gets it wrong. If you first ask it to think step by step — and only then give the answer — accuracy goes up. In this lesson we learn the simple phrasing that triggers this, and when

If you ask the model for an instant answer to a multi-step puzzle, it may slip. If you ask it to 'think step by step, then give the answer' — it shows the path, catches mistakes on the way, and lands it more often.

chain-of-thought: Asking the model to show its reasoning steps before the final answer, to improve accuracy on multi-step tasks.
reasoning steps: The intermediate steps the model writes on the way to the answer — each one building on the previous.
multi-step task: A task you must break into several steps to solve — a math word problem, logic, planning, or several constraints at once.