LLM Self-Reflection

A technique in which a large language model is prompted to evaluate and critique its own output, identifying errors, gaps, or areas for improvement. In the context of accessibility, LLM self-reflection involves asking the model to assess whether the code or UI it generated meets accessibility standards and to suggest fixes. Research has shown that LLMs can achieve approximately 76% alignment with expert accessibility evaluations through self-reflection, excelling at identifying contrast and semantic structure issues but underreporting problems with keyboard navigation and interactive elements. This approach shows promise as a complementary tool alongside automated testing and expert review.