Chain-of-thought prompting: Chain-of-thought prompting asks an AI model to work through a problem step by step before giving its final answer. Showing its reasoning tends to improve accuracy on math, logic, and multi-step tasks and makes mistakes easier to spot.
A language model produces its answer one token at a time, and it cannot go back and revise. If you ask for a final number directly, it has to commit before it has 'thought.' Asking it to lay out intermediate steps first gives it room to work, and the final answer is conditioned on that work.
The technique was described by Google researchers in a 2022 paper (Wei and colleagues, 'Chain-of-Thought Prompting Elicits Reasoning in Large Language Models'). The simplest form is adding 'think step by step' or 'show your reasoning before the answer.' Richer versions give the model an explicit procedure to follow.
Newer reasoning models do this on their own, spending extra time working through a problem before responding, so you may not need to ask. The habit is still valuable: a visible chain of reasoning is the easiest thing to check when the stakes are high, and it exposes wrong assumptions that a bare answer would hide.
Example at work
A financial analyst asks: 'Before recommending a lease-versus-buy option, list the assumptions you are making, calculate the five-year cost of each option showing each step, then state your recommendation and the one assumption that would flip it.' The visible math lets her catch that the model applied the wrong discount rate.
Why it matters
For anything involving numbers, multi-step logic, or policy interpretation, asking for the reasoning is cheap insurance. You get better answers and, more importantly, a trail you can audit before you act on them.
Related terms
- Reasoning modelA reasoning model is a large language model trained to work through a problem step by step before answering, spending extra computation (often called thinking) to improve accuracy on math, code, logic, and multi-step tasks.
- Prompt engineeringPrompt engineering is the practice of writing and refining instructions to get reliable, high-quality output from an AI model. It covers giving context, showing examples, specifying format, and iterating on what did not work.
- Few-shot promptingFew-shot prompting is including a handful of examples of the input and output you want inside your prompt so the AI model can copy the pattern. It is the fastest way to get consistent formatting, tone, and judgment calls.
- HallucinationA hallucination is when an AI model produces information that is false or fabricated but presents it confidently, such as an invented statistic, a citation to a paper that does not exist, or a made-up product feature.
- Zero-shot promptingZero-shot prompting is asking an AI model to do a task with instructions only, without showing it any examples of the output you want. It works well for common tasks the model has seen many times, such as summarizing or translating.