Algorithmic Bias
Risk and SafetyEssentialAppliedNeeds source verification
Definition
Systematic differences in an algorithm’s outputs or errors that can create or reinforce unfair outcomes.
Context
Healthcare demo supplemental
Professional scenario
An AI screening tool has materially different false-negative rates across patient groups.
Governance implication
Evaluate subgroup performance and investigate disparities.
Question to ask
How does the error rate differ across patient populations?
Relevant to
AI / IT, AI Attorney, AI Project Manager, Data Scientist, HR, Healthcare Leader, ML Engineer, Nurse, Pharmacist, Physician, Researcher (Suggested — not reviewed)
Source: Conference demonstration draft; independent verification pending
Verification: Not independently verified
Version 0.1-draft · AGH-008