Nearly every student uses AI; many programs effectively ban it. We are running a multi-year study of what actually develops when AI collaboration is taught as a skill rather than policed as a problem.
How do people teach an agent that cannot be told what to do? We study teaching by physical state intervention — rearranging the environment so the learner discovers the right behavior.
A grading tool whose purpose is feedback. It proposes grades, but every point deducted has to name an issue and cite a source you can open — and across fourteen audit rounds, no finding has ever changed a grade.
Probabilistic and transformer-based models that recover the order in which a chronic disease changes the brain and behavior — learned from data collected at a single point in time. Led by Hongtao Hao.
An in-progress tutorial that builds from set-based probability to working Bayesian nonparametric models, with interactive notebooks and probabilistic programming throughout.
One hundred frog figurines are trying to get home across Japan, and none of them can move on their own. Kaeru-chan is a nationwide field study of how cooperation and trust spread through an object passed hand to hand.
This is a tool for generating semantic networks from fluency data
