Context Sensitive Frames and AI Alignment

Artificial Intelligence alignment requires more than encoding human values as traits. It demands representational alignment-the ability to frame situations in ways that mirror human sensemaking. Trait-based models capture general tendencies but underfit real-world variation, ignoring how values shift across contexts. This paper argues that context-sensitive framing is essential for trust in high-stakes domains like medical triage. We draw from the Frame Problem, Relevance Realization, and the Person-Situation Debate to highlight the limits of trait-based alignment and propose research into medic sensemaking.