Michael Graziano works on the mechanisms by which the brain builds self-models, or predictive self-descriptions, and uses those models to choose actions (intentionality) and to interact with others (social cooperation). One of the most exciting lines of research that we are currently pursuing is to incorporate self-models into machine learning. We find that adding an attention schema in particular—a schematic, predictive model of one’s own attention—makes artificial agents more capable at collaborative tasks, just as an attention schema is hypothesized to do in the human brain. The goal is to understand how to take fundamental social mechanisms that allow for better cooperation in humans and incorporate those mechanisms into artificial systems. A second line of research in my lab is to study how human social health is affected by increasing interaction with socially capable AI. We are finding that, contrary to many fears, regular interaction with chatbots can in some cases improve social health by providing a safe, comfortable conversation partner, in a world where people are increasingly isolated from each other.