Media release

From: PNAS

Comparing humans and AI in psychological tests

A study explores behavioral similarity between humans and AI. As some roles for AI involve decision-making and strategic interactions with humans, it is imperative to understand AI behavioral tendencies. Qiaozhu Mei, Matthew Jackson, and colleagues evaluated the personality and behavior of a series of AI chatbots. The authors asked variations of ChatGPT to answer psychological survey questions and play interactive games that assess trust, fairness, risk aversion, altruism, and cooperation. Next, the authors compared ChatGPTs’ choices to the choices of 108,314 humans from more than 50 countries. ChatGPT-4 passed a Turing test, displaying behavioral and personality traits that could not be statistically distinguished from randomly selected human responses. For example, both humans and chatbots became more generous when told that their choices would be observed by a third party and modified their behaviors after experiencing different roles in a game or in response to different framings of the same strategic situation. However, the chatbots’ behaviors tended to be more cooperative and altruistic than the median human behavior, exhibiting increased trust, generosity, and reciprocity. According to the authors, the findings suggest that such tendencies may make AI well-suited for roles requiring negotiation, dispute resolution, customer service, and caregiving.