Every human has implicit biases, or involuntary associations about members of different social groups. But these biases are not fixed. Rather, they’re malleable reflections of our everyday experiences.

It’s harder to change the biases of people living in big cities, with millions of people, however, than it is to change the biases of people living in small cities, with around 150,000 residents, a new study shows. This is “the first strong proof of how contextual biases are,” says Andrew Stier, a Complexity Postdoctoral Fellow at the Santa Fe Institute who performed the work.

Publishing August 6^th in the Cell Press journal iScience, scientists outline how they developed a method for generating personality assessment questionnaires with ChatGPT from any source text. To test the method, they applied it to both the DSM-5 and, as a deliberately unconventional example, an astrology textbook. Not only could ChatGPT be used to create and validate these questionnaires, but it could also accurately predict population-level responses before the surveys were administered.