Extrovert, Socially Desirable, Not Conscientious, and My Language Is Surprising: Surprisal from Large Language Models As an Individual Difference

We found that surprisal from LLMs can be an individual difference that quantifies how subjects produce their discourse and their predictability, regardless of the language-based task. Its relationship with the OCEAN model was evaluated in a Spanish sample of 643 undergraduate students, being predicted by extroversion, social desirability and conscientiousness.