Human-AI Interaction

The study and design of how people interact with artificial intelligence systems, including how AI communicates its outputs, uncertainty, and limitations to users. Key principles include making AI behavior transparent, supporting user correction of errors, acknowledging uncertainty, and allowing users to control and customize AI behavior. In accessibility contexts, human-AI interaction presents unique challenges: users with disabilities may not be able to verify AI outputs through the same channels the AI uses for sensing (e.g., Deaf users cannot hear the sounds a sound recognition AI is classifying), requiring alternative feedback mechanisms and trust-building strategies. Guidelines from Microsoft and Google PAIR emphasize graceful degradation, user agency, and efficient error correction.