Researchers from Stanford University and Northwestern University have developed generative agents that simulate human behavior based on interviews conducted with 1,052 individuals. Each participant underwent a two-hour interview, which was then used to create AI agents that replicate the attitudes and behaviors of the interviewees. These agents demonstrated an 85% accuracy rate in simulating their human counterparts during subsequent surveys and tasks. The study, titled 'Generative Agent Simulations of 1,000 People,' aims to provide a testbed for policy analysis and social science research, potentially revolutionizing how researchers approach the study of human behavior and societal changes.