Chain of Thought Prompting

Chain-of-Thought (CoT) prompting is a technique designed to improve the reasoning capabilities of large language models (LLMs) by encouraging them to explicitly articulate step-by-step reasoning before arriving at an answer. Unlike standard prompting, which expects a direct response, CoT provides the model with examples that include intermediate reasoning steps, prompting it to replicate this approach for new queries. This method, introduced by Wei et al., has shown significant gains in accuracy on tasks requiring multi-step reasoning, such as arithmetic, commonsense, and symbolic reasoning. The technique is documented as part of the Learn Prompting educational platform, which offers courses, documentation, and community resources on prompt engineering. CoT prompting is typically applied through few-shot exemplars, and its effectiveness scales with model size—research indicates that models with around 100 billion parameters or more benefit most, while smaller models may produce illogical chains and degrade performance. The approach is free to use in principle, as it is a prompting method that can be implemented with any compatible LLM, though the Learn Prompting platform itself may offer paid courses or certifications.