Authors: S Sabouri, Y Daryani, S Haghighi, J May, M Dehghani, S Chattopadhyay Status: Under submission, 2026
Summary
Examines how anthropomorphic and sycophantic model behavior distorts expert judgment of therapeutic AI systems — i.e. the people best positioned to evaluate these systems are themselves affected by the traits they are meant to assess.
Related
Research Agenda — covers the “patients navigating conflicting medical advice” domain. Sepand Haghighi — co-author.