Tag
This paper examines how human interventions at fault points in multi-agent medical systems affect diagnostic accuracy, showing improvements with correct interventions and degradation with incorrect ones.
This paper introduces a framework for rhetorical misalignment in language models, where presentation can induce harmful cognitive biases in human decision-making, and demonstrates this through experiments in clinical scenarios.
The paper develops an agentic evaluation framework for clinical reasoning in hematologic oncology, finding that LLMs primarily fail due to systematic information-seeking deficits rather than insufficient knowledge, with error patterns resembling cognitive biases in novice clinicians.
The paper identifies the limitation of assuming rational agents in strategic classification and proposes the Prospect-Guided Strategic Framework (Pro-SF), which incorporates cognitive biases from prospect theory to model behaviorally realistic strategic manipulations.