Self-critique prompting involves asking a model to review its own responses critically as a mechanism to enhance its understanding and outputs. This technique can be used to make the model notice errors or areas for improvement in a generated response. Compared to simple prompting, where the model generates content based on instructions, self-critique prompting encourages a recursive approach for more refined outputs.