Jul 4, 2025
A groundbreaking new artificial intelligence system called Centaur is redefining what it means to "read minds." Developed by researchers and published in Nature, Centaur can accurately predict human behavior across a wide range of psychological scenarios—going far beyond shopping habits or social media engagement. Trained on over 10 million decisions made by more than 60,000 people, the AI can anticipate how we learn, reason, and make choices. What sets Centaur apart is its versatility: unlike older, specialized cognitive models, this AI performs across almost every kind of psychological experiment, even in completely new contexts or unfamiliar situations.
Built using Meta’s Llama 3.1 language model and a massive dataset called Psych-101, Centaur was fine-tuned to understand human behavior by converting experimental scenarios into natural language. Remarkably, it not only outperformed decades-old models, but also displayed internal processes that mirrored real human brain activity. The system even uncovered new decision-making strategies that psychologists hadn’t identified before. Though this AI breakthrough offers huge potential in areas like education, mental health, and scientific discovery, it also raises serious questions about privacy, digital manipulation, and ethical oversight. For the first time, an artificial system has begun to model the intricacies of human cognition with startling precision—ushering in a new era where predicting behavior may become as routine as observing it.
Stay Awake. Keep Watch.
SOURCE: Israel Hayom