How clinicians, AI systems, and patients make decisions together — and what makes those decisions safer, more equitable, and more accountable.

Most decisions in safety-critical clinical care are made by teams, not individuals: a high-stakes care decision typically involves multiple clinicians across roles, the wider care team, and the patient herself. My research studies how an AI system enters that team — when its flags are accepted appropriately, when they’re dismissed, how trust calibrates across hours of use, and whether the team’s decisions become safer and more equitable for that AI being in the room.

I focus on settings where decisions are time-pressured, multi-stakeholder, and where documented racial and language disparities in care are among the most consequential in medicine. Across these settings, I develop the measurement methods, the interaction primitives, and the deployed evidence base for AI that supports rather than replaces team judgment.

Two threads run across the work: (1) team-level appropriate reliance — extending HCI work on individual AI use to the distributed decision-making clinicians actually do; (2) trust trajectories in real clinical use, not lab vignettes.

The work is part of the broader Hybrid Intelligence research community, alongside collaborators studying human–AI teams across medicine, science, and engineering. The methods generalize to any team-based safety-critical decision setting — ICU, operating room, emergency resuscitation — where cohorts are dense, teams are clear, and equity claims are testable.