Keeping Humans Safe in the AI Era: Why Psychological Science Belongs at the Center of Model Evaluation

As LLM-powered products move into mental health, youth support, healthcare, education, and other high-stakes domains, traditional AI evaluation methods are no longer sufficient. These systems do more than generate correct or incorrect answers, they shape human experiences, respond to distress, influence decisions, and interact with people in moments of vulnerability. 

This keynote explores why psychological science must be central to AI safety and model evaluation. Drawing on their expertise as a psychologist and AI safety leader, Dr. Alison Cerezo will discuss the role of psychological science in AI safety, an ongoing, multidisciplinary practice grounded in human behavior, clinical judgment, cultural context, and real-world risk. Attendees will leave with a clearer understanding of how psychologists can shape safer AI systems and why their expertise is essential for keeping humans safe in the AI era.