Personality Auto-Scoring with Large Language Models Using a Realistic Accuracy Model of Behavioral Cues in Chatbot Interviews (Updated December 29, 2025).

The article focuses on the integration of Funder's Realistic Accuracy Model (RAM) into large language model (LLM)-based autoscoring for personality assessment through text-based chatbot interviews. The study examines the alignment between LLM-derived personality scores and established measures using two archival samples, revealing that RAM-based prompts showed stronger convergence with human ratings in a job-focused behavioral interview compared to zero-shot prompts. However, in narrative identity interviews, the advantage of RAM prompts diminished, indicating similar convergence with zero-shot prompts. The research highlights the potential for theory-guided LLMs to identify behavioral cues relevant to personality evaluation, while also discussing limitations and implications for AI-based assessments. [Extracted from the article]