Bias and inclusion
AI critique must include inclusion
- AI output can be generic, biased, or homogenized.
- Ask who is missing, burdened, stereotyped, or misrepresented.
- Check language, accessibility, cultural assumptions, and edge cases.
Quality lens
Missing userWho is not represented in the answer?
Hidden normWhat assumption might not hold?
Access gapWhere might the design exclude people?
Evidence gapWhat should be tested with real users?