AI-OUTPUT-EVALUATION4 MIN READ
Sort The Eval Failure
Categorize common AI output failures into task fit, evidence, constraint, and risk buckets.
Sort each AI output failure into the category that best explains it. Name the failure before choosing a fix. Task fit Evidence Missing constraint Risk control Asked for a two-sentence answer; got a full essay. Cited a source that does not support the claim. Applied a US policy to a global contractor case. Promised a refund before approval. Summarized the wrong audience's priorities. Used last year's pricing as if it were current. fail-1 fail-5 fail-2 fail-6 fail-3 fail-4
Read the full lesson
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in