PRACTICAL EXAMPLE - 02
Financial
Services
Credit, fraud, and compliance decisions affect millions. Regulators demand accountability at every automated step.
SCENARIO
A bank uses AI to flag real-time fraud transactions and a separate model to score mortgage applications. Both affect customers financially.
WITHOUT MIHR
A declined mortgage triggers a complaint. The bank cannot isolate whether the AI made, assisted, or merely flagged the decision. Full discovery is required.
WITH MIHR — TRANSPARENCY LABELS
HOW
Anomaly Detection (Unsupervised ML)
OUTCOME:
AI flags transactions instantly. Human compliance officer reviews within 4 hours. Accountability chain is documented and auditable.
HOW
Predictive Scoring (Gradient Boosting)
OUTCOME:
AI generates a risk score. Loan officer makes the final credit decision. Any bias claim is isolated to the scoring model — not the full stack.
