Frontier models, robots, and quantum machines only matter if they make human lives better. Our fourth research program studies exactly that — social science, civics, emotional intelligence, and psychology — done with people, not just about them.
AI doesn't land in a vacuum — it lands in workplaces, families, and neighborhoods. Our social scientists run longitudinal studies across 31 communities to measure how intelligent systems actually change work, trust, and opportunity, and publish every dataset openly.
Explore the social labWe build and test deliberation tools that help citizens reason together: assemblies assisted by open models, misinformation-resilience curricula, and civic-literacy studies with election officials — all evaluated in the field, not the lab.
See the civics programAffective computing done consent-first. We research how models can recognize and respond to human emotion without manipulating it — with hard commitments: no covert inference, participant ownership of affect data, and open audits of every emotion model we train.
Read the affect charterOur psychologists study memory, attention, and trust in human–AI collaboration: when people over-rely on models, when they under-trust them, and how interface design changes both. The findings feed straight back into how we build Chancellor models.
Browse cognition studiesSocial science can’t be done from a screen. Our teams spend as much time in town halls, classrooms, and clinics as they do at keyboards.
Every program starts with structured interviews — 2,400 hours so far — with the workers, teachers, and officials our tools will touch.
Two-year deployments in 40 schools study how open models change learning — co-designed with teachers, governed by parents.
With clinicians and counselors, we test where emotional-support AI helps, where it harms, and where it simply shouldn't be used.
This program is hiring across all four frontiers — and every study you run here gets published, openly.
