Is AI clairvoyant? ChatGPT can make personality tests and predict responses, Israeli study finds

A Hebrew University of Jerusalem study found that ChatGPT can generate personality assessment questionnaires and accurately predict population-level responses. Scientists Dr. Rotem Monsa, Prof. Aviv Zohar, and Prof. Shahar Arzy published their findings in iScience. They used GPT-4 to create questionnaires based on the DSM-5 and, as a control, an astrology textbook. The DSM-5-sourced questionnaire showed high internal consistency, similar to the validated Big Five personality questionnaire, while the astrology-based one did not. Surprisingly, ChatGPT predicted the mean responses and correlations between questions with high accuracy before participants took the surveys. Monsa explained that LLMs learn personality structures as a byproduct of their training on vast amounts of human language data, stating, “Since personality traits are reflected in language, LLMs may have learned the structure of human personality as a natural byproduct of their training.” The researchers noted that while AI doesn't have human understanding, its responses are very human-like. This method could potentially be used for diagnosis or treatment in the future, especially in areas with limited access to psychologists, though cultural biases from Western-centric training data remain a concern.