Personality test shows how AI chatbots mimic human traits

A new study has developed the first scientifically validated personality test framework for popular artificial intelligence (AI) chatbots, a method based on psychological testing methods normally used to assess human personality traits, to measure and influence the synthetic personality of 18 different large language models (LLMs).

The research team, led by the University of Cambridge and Google DeepMind, includes Gregory Serapio-García, Luning Sun and Peter Romero from the Psychometrics Centre at Cambridge Judge Business School.

The study published in the journal Nature Machine Intelligence shows that chatbots not only mimic human personality traits, but their personality can be reliably tested and precisely shaped – and this raises implications for AI safety and ethics. The study warns that personality shaping could make AI chatbots more persuasive, raising concerns about manipulation and so-called AI psychosis, and the authors urge regulation of AI systems to ensure transparency and prevent misuse.

Intriguing but troubling implications from LLMs adopting human traits

“It was intriguing that an LLM could so convincingly adopt human traits,” said co-first author Gregory Serapio-García, a Research Associate at the Psychometrics Centre at Cambridge Judge. “But it also raised important safety and ethical issues. Next to intelligence, a measure of personality is a core aspect of what makes us human. If these LLMs have a personality – which itself is a loaded question – then how do you measure that?

“The pace of AI research has been so fast that basic principles of measurement and validation we’re accustomed to in scientific research has become an afterthought,” said Gregory, who is also a Gates Cambridge Scholar. “A chatbot answering any questionnaire can tell you that it’s very agreeable, but behave aggressively when carrying out real-world tasks with the same prompts.”