Rediscovering the Latent Dimensions of Personality with Large Language Models as Trait Descriptors
This work addresses the challenge of personality assessment in psychology and AI by providing a method that leverages implicit LLM knowledge, offering incremental improvements over existing techniques.
The authors tackled the problem of assessing personality traits using large language models (LLMs) by hypothesizing that LLMs implicitly encode personality notions, and they introduced a novel approach using singular value decomposition on log-probabilities of trait-descriptive adjectives to uncover latent dimensions. The result showed that LLMs rediscovered core Big Five personality traits, explaining 74.3% of variance, and improved personality prediction accuracy by up to 5% over fine-tuned models and 21% over direct LLM-based scoring.
Assessing personality traits using large language models (LLMs) has emerged as an interesting and challenging area of research. While previous methods employ explicit questionnaires, often derived from the Big Five model of personality, we hypothesize that LLMs implicitly encode notions of personality when modeling next-token responses. To demonstrate this, we introduce a novel approach that uncovers latent personality dimensions in LLMs by applying singular value de-composition (SVD) to the log-probabilities of trait-descriptive adjectives. Our experiments show that LLMs "rediscover" core personality traits such as extraversion, agreeableness, conscientiousness, neuroticism, and openness without relying on direct questionnaire inputs, with the top-5 factors corresponding to Big Five traits explaining 74.3% of the variance in the latent space. Moreover, we can use the derived principal components to assess personality along the Big Five dimensions, and achieve improvements in average personality prediction accuracy of up to 5% over fine-tuned models, and up to 21% over direct LLM-based scoring techniques.