Contents 9. AI-Mediated Interaction
How users interact with systems whose behavior is probabilistic or changes through learning.
These constrain intelligent systems to support human judgment without replacing it.
Mission statement
Ensure intelligent systems support human judgment responsibly by making system behavior understandable, controllable, and trustworthy.
Subcategories
1.Human OversightWho remains accountable.“Is human responsibility in your system explicit and exercisable at all times?”10 directives2.Learning ConsentWhat a system may learn and apply.“Do users of your system authorize what it learns from them and how it applies it?”9 directives3.PredictabilityWhy stability matters more than surprise.“Can users of your system reliably predict its behavior?”9 directives4.ExplainabilityHow results are understood.“Can users understand and contest outcomes in your system?”7 directives5.Confidence SignalingHow certainty and uncertainty are conveyed.“Does your system signal uncertainty to users honestly and actionably?”9 directives6.Graceful FailureWhat happens when intelligence breaks.“Does your system behave responsibly when intelligence fails?”9 directives7.ProvenanceWhat data influenced outcomes.“Can users of your system see what data shaped an outcome and its scope?”9 directives8.Bias ManagementWhere distortions may arise.“Are sources of bias in your system visible and mitigable to users?”8 directives9.Trust CalibrationHow reliance is adjusted over time.“Is user trust aligned with actual system capability over time?”9 directives10.Non-AnthropomorphismHow to prevent false human attribution.“Does your system present itself honestly to users as a tool, not a human agent?”9 directives