arXiv Artificial Intelligence· Stefano Epifani (Digital Transformation Institute, Rome, Italy, Department of Communication and Social Research, Sapienza University of Rome, Italy), Giuliano Castigliego (Digital Transformation Institute, Rome, Italy, Psychoanalytic Academy of Italian-Speaking Switzerland, Chur, Switzerland), Giuliano Razzicchia (Digital Transformation Institute, Rome, Italy), Elisabeth Seiwald-Sonderegger (Psychiatric Services of the Canton of Grisons, Chur, Switzerland), Laura Kecskem\'eti (Psychiatric Services of the Canton of Grisons, Chur, Switzerland)·· 18 小时前AI 评分31
GPT-4.1 能否复现心智化的语言结构：一项方法学研究
Algorithmic Reflectivity in Large Language Models: Methodological Study of Linguistic Structure in Mentalization
AI 导读
研究用 GPT-4.1 在标准化条件下生成 50 段人机对话，并由 5 名受过 MBT 训练的精神科医生按 4 个 MBT 轴和 2 个高阶维度独立评分。各维度均分 3.63-3.98，评价域 ICC(3,k) 为 0.60-0.84，评价与论证一致性相关系数 r 为 0.826-0.967，错配对话-画像对的阴性对照评分极低。
来源：arXiv Artificial Intelligence · arxiv.org