KAIST AI Interviewer: Psychiatric Intake in 30 Mins

A Korean research team has developed an AI interviewer to support initial psychiatric consultations. This system allows patients to first talk with AI to structure their symptoms before seeing a psychiatrist. KAIST announced that a research team led by Professor Lee Eui Jin and Professor Takyeon Lee, along with Professor Eun-Joo Kim, jointly developed this large language model-based system. The goal is to address time constraints in initial psychiatric interviews, which typically last around 30 minutes. The AI interviewer conducts natural language dialogue with the user, systematically collecting purpose-specific information. It identifies symptoms, emotional state, and impairment in daily functioning. The AI adjusts the conversation flow in real time based on patient responses, generating key questions by analyzing answers against specialized psychiatric knowledge. It also incorporates counseling techniques like empathy and clarifying. This system aims to improve the efficiency and depth of psychiatric care, allowing physicians to focus on in-depth counseling.