Chain-of-Thought
Prompting a model to work through its reasoning step by step before giving a final answer.
It noticeably improves accuracy on math, logic, and multi-step problems.
Chain-of-Thought in practice
The technique comes from a 2022 Google research paper by Wei and colleagues, which found that showing intermediate reasoning steps improved complex reasoning. Many current models now reason internally, so an explicit "think step by step" matters most when that is switched off. It is one tool in the wider prompting guide; for the basics, see what a prompt is, or compare it with prompt chaining.
Frequently asked questions
What does chain-of-thought mean for AI?
Chain-of-thought means prompting the model to work through its reasoning step by step before giving a final answer, rather than jumping straight to a conclusion. Walking through the steps tends to catch mistakes a snap answer would miss.
How do I get an AI to reason step by step?
You can simply ask it to think through the problem step by step or to show its work before answering. This is especially helpful on math, logic, and multi-step questions.
Does chain-of-thought actually improve answers?
Yes, it noticeably improves accuracy on math, logic, and multi-step problems because the model reasons its way to the answer instead of guessing in one leap. The trade-off is that it takes a bit longer and uses more words.
New to all this? Start with what an AI agent really is, browse the full glossary, or explore the learning hub.
← Back to the glossary