Psychometrics Benchmark
Psychometrics benchmarking aims to rigorously evaluate the psychological attributes of large language models (LLMs), assessing capabilities like personality, emotion, and theory of mind. Current research focuses on developing comprehensive benchmarks encompassing diverse psychological dimensions and utilizing various assessment methods, including those inspired by established psychological examinations and incorporating psychometric principles for improved reliability and validity. These efforts are crucial for understanding LLMs' behavior, identifying limitations, and informing the responsible development and deployment of these increasingly influential technologies across various fields, including mental health applications.
Papers
July 8, 2024
June 25, 2024
May 16, 2024
April 2, 2024
December 2, 2023
October 25, 2023
June 1, 2023