LLM data classification: a prompt is only the beginning
Build a label handbook, preserve supporting evidence, and route ambiguous outputs to a reviewable decision process.
Read the guide ↗DATA & DECISIONS
Decision Model Neuro Browser
Turn a broad decision into explicit alternatives, assumptions, tradeoffs, and review responsibilities before involving an AI model.
TOPIC GUIDE / 03 · READ, QUESTION, EXPLORE
Write the decision before calculating
Start with the choice being made and the person responsible for it. Record the available alternatives, non-negotiable constraints, and evidence supporting each criterion. A weighted score is a representation of priorities, not an objective answer independent of those choices. Keep the assumptions available to the person who must act on the recommendation.
Try a sensitivity review
Change one assumption at a time and ask whether the preferred option changes. In a model-selection exercise, compare review workload, data permissions, and operational demands rather than only a demonstration score. If small changes reverse the decision, document that instability and identify the evidence needed next. The goal is a more understandable decision, not an automatic one.
Three questions to take with you
01Who owns the final decision?
02Which assumptions could reverse the recommendation?
03What evidence is still missing?
Reference for this topic: NIST guidance on AI risks and trustworthiness. Read the original context alongside the guide.