AI for Mental Health and Well-Being

We are studying how LLMs can safely support therapist training and patients between sessions, and are evaluating these systems in clinical trials. We also study how LLMs may shape society, including their capacity for political persuasion.

We use NLP and large language data from the social web (Facebook, Twitter, Reddit) to measure the psychological states of large populations and individuals, to determine the thoughts, emotions, and behaviors that drive illness, depression, or support well-being.

This is especially relevant for measuring subjective well-being for populations around the world—in places where no traditional measures are available with sufficient spatial and temporal resolution to capture the impact of economic or social disruptions or to inform public policy.

We are part of the World Well-Being Project consortium which we co-founded in 2011, and part of the Stanford Institute for Human-Centered Artificial Intelligence.