Refuse People-Ranking Checklist

Use with Refuse to Rank People With AI.

The four-question check

Before asking AI to rank, score, or grade people, ask:

1. Would I be comfortable explaining the exact criteria to the person being ranked?
2. Is there an accountable human who owns this decision and can be questioned about it?
3. Am I using AI to organize evidence, or to produce the verdict itself?
4. Would this ranking hold up to an audit?

If any answer is no, stop and restructure: use AI to organize criteria and evidence, and make the ranking or decision yourself.

Context-specific checklists

Hiring / candidate screening

• Criteria were written down before any candidate was reviewed.
• AI was used only to organize documented evidence (resume facts, work samples), never to output a ranked shortlist.
• A named human owns and can defend the final decision.
• If using any automated scoring tool, it has been independently bias-audited (required by law in some jurisdictions for employment tools).

Performance reviews

• Criteria are consistent across everyone being reviewed, written before review season starts.
• AI checked my own written evaluations for consistent language and standards across people, not the other way around.
• The final rating is mine, based on evidence, not an AI-generated score.

School grading

• The rubric was set before any student’s work was seen.
• AI assisted with organizing rubric criteria or flagging inconsistency, not assigning the grade.
• I can explain the grade to the student and parent without reference to an AI output.

Everyday informal comparisons

• I am not asking AI to tell me who is the “best” friend, partner, family member, or employee.
• If I am using AI to think through a relationship decision, I am using it to organize my own observations, not to generate a verdict about a person.

What AI can still help with

• Drafting consistent, specific evaluation criteria before looking at any candidate or person.
• Organizing documented evidence into a clean comparison format for human review.
• Checking your own written evaluations for consistent language and standards across people.

What this cannot do

An AI-generated score or ranking of a person is not neutral, verified, or safe to treat as a decision on its own — informal use carries the same underlying bias risk as a formal system, without any audit process to catch it.