Is AI clairvoyant? ChatGPT can create personality tests and predict responses, Israeli study finds
Translated from English, summarized and contextualized by DistantNews.
At a glance
- Scientists at the Hebrew University of Jerusalem have developed a method using ChatGPT to create personality assessment questionnaires from any English text.
- The AI could not only generate but also accurately predict population-level responses to these questionnaires.
- Researchers tested the method on the DSM-5 and an astrology textbook, suggesting AI can understand personality structures embedded in language.
Scientists at the Hebrew University of Jerusalem have discovered that artificial intelligence, specifically large language models like ChatGPT, can be used to generate personality assessment questionnaires and even predict how populations will respond. This groundbreaking method was detailed in the journal iScience.
The research team, including Dr. Rotem Monsa, Prof. Aviv Zohar, and Prof. Shahar Arzy, applied their technique to diverse texts, including the Diagnostic and Statistical Manual of Mental Disorders, 5th Edition (DSM-5), a key resource for mental health professionals, and an astrology textbook. They found that ChatGPT could create valid questionnaires based on these sources and accurately forecast responses at a population level before any surveys were conducted.
Since personality traits are reflected in language, LLMs may have learned the structure of human personality as a natural byproduct of their training.
Lead author Dr. Rotem Monsa explained that AI models are trained on vast amounts of internet data, including social media and websites. This extensive training allows them to learn the structure of human personality as a natural consequence of processing language. "Since personality traits are reflected in language, LLMs may have learned the structure of human personality as a natural byproduct of their training," Monsa stated. "So, while they were not taught specifically psychology or personality theories, these are already embedded in the language that LLMs learn from."
To validate their findings, the researchers utilized GPT-4, a sophisticated AI model. They generated questionnaires based on descriptions of personality disorders from the DSM-5, demonstrating the AI's capacity to identify and structure personality traits from complex texts. The study suggests that AI's understanding of human language extends to recognizing underlying personality structures, potentially opening new avenues for psychological research and assessment.
So, while they were not taught specifically psychology or personality theories, these are already embedded in the language that LLMs learn from.
Originally published by Jerusalem Post in English. Translated, summarized, and contextualized by our editorial team with added local perspective. Read our editorial standards.