Chain of Thought
A prompting technique where AI models reason step-by-step rather than jumping to answers.
Also known as: CoT prompting, Step-by-step reasoning, Thinking aloud
Category: Techniques
Tags: ai, prompting, reasoning, thinking, techniques
Explanation
Chain of Thought (CoT) is a prompting technique that encourages AI models to reason through problems step-by-step rather than jumping directly to answers. By asking models to 'show their work,' CoT dramatically improves performance on complex reasoning tasks. How it works: instead of asking for just the answer, prompt the model to explain its reasoning process. This can be done by: including step-by-step examples (few-shot CoT), simply adding 'let's think step by step' (zero-shot CoT), or asking the model to reason before answering. Why it works: breaking problems into steps reduces error propagation, exposes the reasoning process (making errors visible), and leverages the model's full capability. Best applications: math problems, logical reasoning, complex analysis, multi-step tasks, and any problem where intermediate steps matter. Variations include: tree of thought (exploring multiple reasoning paths), self-consistency (generating multiple chains and selecting best answer), and extended thinking (longer deliberation). For knowledge workers, chain of thought prompting is essential for: getting better AI outputs on complex tasks, understanding how AI reached conclusions, and catching errors in reasoning.
Related Concepts
← Back to all concepts