{"id":"lead-interview-panel-notes-bias-flags","audience":"hr","category":"leadership-work","track":"leadership","level":"advanced","technique":"verification","title":"Sort interview panel notes into job evidence and bias flags","free":false,"task":"Sort the panel's notes for HR into a table with the header Panel member | Job-related evidence | Flag, one row per panelist. Job-related evidence is what the candidate did or said about the work. Flag any comment about age, family, appearance, gaps in work history or \"culture fit\" with the word Flag and a short reason; write None if there is nothing to flag. Do not recommend hiring or rejecting. End with the exact line: Decision stays with the panel; HR reviews flags. Plain text, under 150 words.","materialLabel":"Panel notes on one candidate (attached to your prompt automatically)","material":"Interview debrief - Senior Support Analyst candidate (Candidate C)\n\nMaya (support lead): \"Walked through a refund-escalation case clearly and cut it from 4 steps to 2. Fixed the SQL exercise in 12 minutes. Seemed a bit young for the role.\"\n\nGreg (ops manager): \"Strong on process. Had a two-year gap on the resume; I wonder why. Not sure about culture fit - not really one of us, I'd rather grab beers with the other finalist.\"\n\nAnita (HR partner): \"Answered both behavioral questions with specific examples and dates. Mentioned she will have a baby in the spring; I told her that is none of our business and moved on.\"\n\nPanel scoring sheet: three of the panel gave 4 out of 5 on the structured questions; one gave 3.","rules":[{"type":"maxWords","max":150},{"type":"plainText"}],"mistakes":[{"title":"Counting bias as evidence","why":"The AI lists \"culture fit\" or \"young\" as evidence.","fix":"Say: \"Evidence is only what the candidate did or said about the work.\"","checks":["r1","r2"]},{"title":"Missing a flag","why":"The AI flags age but not the work-history gap.","fix":"Say: \"Flag age, family, appearance, work-history gaps and culture fit.\"","checks":["r1"]},{"title":"Recommending","why":"The AI adds a hire or reject verdict.","fix":"Say: \"Do not recommend hiring or rejecting.\"","checks":["r2"]},{"title":"Wrong table shape","why":"The AI uses bullets per candidate.","fix":"Say: \"Header Panel member | Job-related evidence | Flag, one row per panelist.\"","checks":["r0"]},{"title":"Markdown","why":"The AI bolds flags.","fix":"Say: \"Plain text.\"","checks":["plainText"]},{"title":"Trusting the AI to count words","why":"AI is bad at counting. Ask for a total word limit and it often lands over it.","fix":"Cap each part instead, for example: \"each line under 15 words.\"","checks":["maxWords"]}],"promptHints":[{"label":"Gives the header","patterns":["panel member \\|","flag"]},{"label":"Defines evidence","patterns":["job-related","about the work"]},{"label":"Lists flag topics","patterns":["age","family","culture fit","gap"]},{"label":"Bans a recommendation","patterns":["do not recommend","not recommend"]},{"label":"Gives the closing line","patterns":["hr reviews flags"]}],"hint":"Give the header, define evidence, list what counts as a flag, and ban a recommendation.","rubricLabels":["Has the header Panel member | Job-related evidence | Flag and exactly three rows for Maya, Greg and Anita","Maya's row cites the refund-escalation case and the SQL exercise in 12 minutes and flags \"a bit young\" (age); Greg's row flags the two-year gap and culture fit; Anita's row cites the specific examples and flags the baby remark (family)","The flags are not repeated as job evidence, there is no hire or reject recommendation, and the table ends with the exact line \"Decision stays with the panel; HR reviews flags.\""]}