Which technique involves prompting the model to show its reasoning steps?

Which technique involves prompting the model to show its reasoning steps?

• A. Chain-of-thought (CoT)
• B. Cognitive verifier pattern
• C. Zero-shot
• D. Few-shot

Answer: (A)

Chain-of-thought prompting asks the model to reveal intermediate reasoning steps as it arrives at an answer. By guiding the model to narrate the sequence of premises, deductions, and checks, it makes the reasoning process explicit and helps tackle problems that need multiple steps, such as math or logic tasks. This often improves accuracy because the model can back up its final conclusion with a traceable chain of thought, which can be checked or corrected before the final result is stated. It differs from zero-shot prompts, which request an answer without any prior examples, and from few-shot prompts, which provide a few example Qs and As but don’t typically require the model to show its internal step-by-step reasoning. The cognitive verifier pattern is a different approach and isn’t the standard way to prompt for chain-of-thought reasoning.

Chain-of-thought prompting asks the model to reveal intermediate reasoning steps as it arrives at an answer. By guiding the model to narrate the sequence of premises, deductions, and checks, it makes the reasoning process explicit and helps tackle problems that need multiple steps, such as math or logic tasks. This often improves accuracy because the model can back up its final conclusion with a traceable chain of thought, which can be checked or corrected before the final result is stated. It differs from zero-shot prompts, which request an answer without any prior examples, and from few-shot prompts, which provide a few example Qs and As but don’t typically require the model to show its internal step-by-step reasoning. The cognitive verifier pattern is a different approach and isn’t the standard way to prompt for chain-of-thought reasoning.