Matthew Leitao, behavioral scientist
People trust AI less than doctors, even when the advice is identical.
I design experiments on how people decide whether to trust AI, then report what the data support, including the null results. I am a postdoctoral researcher at the National Institutes of Health.
- 13+years in behavioral science
- 6experiments in my dissertation
- 3,000+participants studied
- 10papers published, accepted, or under review
I turn questions about people and AI into evidence
I am a behavioral scientist with a PhD in psychology from Georgetown and a master's in public policy. I study how people trust, interpret, and make decisions with AI, especially in health. At the National Human Genome Research Institute, I study how people read AI-generated health information, how uncertainty changes trust, and how virtual reality can fit into clinical research.
My dissertation ran six experiments with more than 3,000 people. I build the study, analyze the data in R, and explain the result in plain words. I have also led teams of up to 15 and managed six concurrent projects.
What I do
Design experiments
Randomized studies, registered in advance, from the first hypothesis through recruitment, analysis, and write-up.
Model messy human data
Multilevel, longitudinal, and causal models in R, applied to surveys, daily diaries, and clinical-trial data.
Lead research teams
I managed five research assistants across six or seven projects at Georgetown. I now manage six projects at NIH.
Build with AI
I build research tools on LLM APIs, test AI output with blind ratings, and use AI to speed up analysis and code.
Where I have expertise
- Trust in AI and algorithms
- Human-AI collaboration at work
- Anthropomorphism and mind perception
- Medical AI and health decisions
- Genetic risk and health beliefs
- Digital health behavior change
- Psychology of technology
- Smartphones and wellbeing
- Virtual reality in research
- Clinical drug trials, behavioral outcomes
- COVID-19 and social behavior
- Parasocial relationships
- Gaming and motivation
- Obedience to authority
- Open science and reproducibility
The gap shows up in every kind of medical decision
In Study 2 of my dissertation, 106 adults rated how much they trusted a doctor and an AI making the same medical judgment. Each dot is the average difference, and the line is its 95% confidence interval. Hover over a row or tab to it for the exact statistics.
AI trusted more Doctor trusted more
View as a table
| Domain | Difference | 95% CI | t | df | Cohen's d |
|---|---|---|---|---|---|
| Physical health | 3.98 | 3.40 to 4.57 | 13.55 | 103 | 1.33 |
| Mental health | 3.91 | 3.30 to 4.53 | 12.67 | 103 | 1.24 |
| Bodily injury | 3.97 | 3.38 to 4.56 | 13.32 | 103 | 1.31 |
| Acute sickness | 3.93 | 3.33 to 4.54 | 12.85 | 104 | 1.25 |
| Chronic sickness | 4.00 | 3.40 to 4.60 | 13.2 | 104 | 1.29 |
The gap is large in all five domains, with effect sizes from d = 1.24 to 1.33. It barely moves with the type of decision, so the cause is more likely in how people see the decision-maker than in the medical content. The rest of the dissertation tests that idea.
Source: Leitao (2024), Georgetown University dissertation, Study 2, Table 7. Read the full case study.
Keep exploring
- ProjectsCase studies built from the data, with every figure rebuilt so you can check it.
- Skills87 skills, each tied to the work that shows it.
- ResearchTen papers on trust in AI, smartphones, and clinical research.
- CVThe short version of my training and roles.
Looking to talk? Get in touch.