Chain-of-Thought
Also called CoT, Chain-of-thought prompting, Step-by-step reasoning
Asking or training an AI model to work through intermediate reasoning steps before giving its final answer.
What is chain-of-thought?
Chain-of-thought (CoT) is a way of getting an AI model to show its working. Instead of jumping straight to an answer, the model writes out the steps in between, such as the sums in a word problem or the checks in a decision, and then gives its conclusion.
It started as a prompting technique: adding worked examples with reasoning, or simply asking the model to "think step by step", made models noticeably better at maths and logic. Today, reasoning models are trained to do this on their own before they answer.
How does it work?
A language model produces one token at a time, and each token can only build on what has already been written. Writing out intermediate steps gives the model more text to build on, so later steps can use earlier results instead of guessing the whole answer in one go. Problems that need several steps benefit most; simple lookups barely benefit at all.
What is it commonly confused with?
- A faithful explanation. The written steps don't always reflect how the model actually reached its answer. A model can produce tidy reasoning for a wrong answer, or the right answer after flawed steps, so treat the chain as helpful, not as proof.
- Prompt chaining. Prompt chaining splits a task across several separate requests; chain-of-thought happens inside one response.
Why does it matter?
Chain-of-thought improves accuracy on multi-step tasks, and it makes mistakes easier to spot because you can see where the reasoning went wrong. The cost is length: more output tokens mean slower and more expensive answers. With a reasoning model, adding "think step by step" usually helps little, because the model already does it.
