Now: September 2026
Last updated: September 7, 2026
This is a now page. It changes when I do.
I want to spend more of my working life thinking about patient-directed AI, putting Critical AI Health Literacy into practice, and collaborating with people who want to investigate these questions together.
I'm looking for a funded research home for that work. A part-time research appointment, fellowship, or sustained paid collaboration could provide the time and continuity to develop ideas, build practical tools, and learn with patients and colleagues.
From framework to practice
Critical AI Health Literacy (CAIHL) connects questions about AI to questions about agency and power. Can patients understand what a system is doing, question whose interests it serves, and direct it toward goals they define themselves? What gets in their way even when they have the knowledge and confidence to act?
The National Academy of Medicine commentary I co-authored with Liz Salmi develops this argument. The CAIHL skill explores how to bring it into practical AI use. My next priority is to investigate what happens when people use these ideas in concrete situations.
Several projects help me approach that work from different directions.
OpenKP connects my Kaiser Permanente Northern California health record to AI tools under my control. It gives colleagues a concrete example to examine as we investigate how patients direct AI around their own questions and data.
HugoScore is a public prototype for evaluating health AI through patient agency. It needs testing, criticism, and refinement. I’m interested in where people agree about what supports agency and where their judgments differ.
My Heart Data brings years of health records into a longitudinal dashboard I can examine. It connects this agenda to my longstanding work on patients’ access to and control over their data. Access matters because people need information they can actually use to pursue their own goals.
Three questions I want to investigate
1. Can CAIHL help people recognize whose interests an AI response serves? We could develop realistic scenarios with patients, compare how people interpret responses before and after a short learning activity, and examine the reasoning behind their judgments.
2. Can patients redirect AI toward their own priorities? We could observe people using a tool for a goal they choose, document where it helps or constrains them, and use those findings to revise the design.
3. Can we evaluate patient agency without flattening important differences? Patients and researchers could apply draft HugoScore criteria to the same health AI tools, discuss disagreements and gaps in the evidence, and refine the criteria together.
These are proposed directions. The research questions, methods, and definitions of success should develop with the people involved. I also want to examine the institutional rules and unequal resources that can limit a person’s ability to act.
Work with me
I’m interested in colleagues working across health literacy, AI, design, patient partnership, and implementation research.
A useful arrangement would support my time, compensate patient collaborators, and provide administrative help. It would also leave room to question assumptions, examine failures, and share what we learn.
If your team could host or fund this work, or you know someone who might, please get in touch.