Researchers from Google and Stanford University have created advanced AI models called “simulation agents.” These AIs can replicate a person’s personality with surprising accuracy after just a two-hour conversation.
The project, led by Joon Sung Park, interviewed 1,052 people from different backgrounds. Using the information from these interviews, the AI was trained to match how humans respond to personality tests and social surveys with 85% accuracy. However, the AI wasn’t as good at predicting decisions in money-related scenarios.
This technology could transform research by simulating human behavior for testing policies and products without involving real people, avoiding ethical concerns. However, it also raises worries about privacy and potential misuse, like creating fake identities. Companies like Tavus are already exploring this tech to develop digital versions of individuals.
