Self-Criticism has emerged as a vital technique for ensuring accuracy from chatbots. This guide provides ready-to-use templates designed to force an AI to evaluate its own reasoning before finalising an answer.
By implementing a feedback loop where the model identifies potential flaws, logical gaps, or biases in its initial logic, you significantly reduce "hallucinations" and enhance technical precision.
These self-correction frameworks transform the AI from a simple generator into a rigorous editor, ensuring the highest quality output every time.
Prompt: I need you to [INSERT TASK HERE].
Before providing the final answer, please follow these steps:
Review for any logical inconsistencies, factual errors, or missing information.
Identify any claims that might be inaccurate or ambiguous.
Correct those claims and present only the most verified version of the information.
