Stanford Seminar - On human-machine interaction games

The podcast features a speaker presenting research on human-machine interactions, focusing on predicting and shaping outcomes when humans control machines. The speaker discusses models people learn when controlling machines, using paradigms from fighter pilot studies to modern interfaces like joysticks and myoelectric controls. The research explores how humans adapt to different dynamical systems, aiming to understand and leverage these models to improve machine learning and adaptation. The speaker details experiments involving scalarized human-machine interaction games, examining various game theory equilibria such as Nash, Stackelberg, and conjectural equilibria, and how machine learning algorithms can influence user behavior. The podcast concludes with a discussion on applying these concepts to brain-machine interfaces, predicting user effort, and shaping outcomes in rehabilitation contexts, advocating for the integration of game theory in human-machine interaction studies, followed by Q&A.