PA024
Log each AI decision's contributing factors at the moment it is made so auditors never need to reverse-engineer the model later.
CyberTRIZ analysis · Process contradiction PA024 · one of 8,235 worked contradictions published by CyberTRIZ.AI
Regulations
Greater AI-Driven Automation Judgment vs. Greater Auditability
Business Context. AI models embedded in automated workflows can make increasingly sophisticated judgment calls, but the more sophisticated the model, the harder it becomes for auditors to reconstruct exactly why a specific decision was made.
Process TRIZ Resolution. Rather than sacrificing judgment sophistication for auditability, organizations should require every AI-driven decision to log its key contributing factors at the time the decision is made, creating an audit trail that does not depend on reverse-engineering the model later.
Applicable TRIZ Principles
Principle 23 (Feedback) logs contributing factors at the moment of decision rather than reconstructing them later.
Principle 10 (Prior Action) builds audit logging into the model's decision process from the start.
Principle 24 (Intermediary) uses an audit layer that captures decision rationale independent of model complexity.
Expected Outcome
Sophisticated automated judgment
Strong auditability
Faster audit response times
Reduced regulatory exposure
Decision Indicators
Auditors cannot reconstruct the basis for past automated decisions.
Audit trails do not exist for AI-driven decisions.
Model complexity has increased without corresponding audit capability.
Regulatory inquiries about automated decisions take excessive time to resolve.
No decision-time logging occurs for AI-driven judgment calls.
If several of these indicators are present, the contradiction is likely active and the Process TRIZ resolution above should be evaluated.