Chain of Thought (CoT)
A prompting technique where an AI model explicitly generates a step-by-step reasoning path before providing a final answer.
Chain of Thought (CoT) prompting is a technique that forces a Large Language Model to articulate its internal reasoning process step-by-step before arriving at a final conclusion.
Instead of jumping directly from a complex question to an answer, the model outputs "thinking tokens" that outline its logical progression. This mimics human problem-solving; when faced with a difficult math problem, a human works through it systematically rather than guessing the final number immediately.
Research has shown that Chain of Thought significantly improves a model's performance on logic puzzles, mathematics, and complex reasoning tasks. Modern "reasoning models" (like OpenAI's o1 or DeepSeek-R1) integrate CoT deeply into their training, generating hidden chains of thought automatically to ensure highly reliable outputs.