About
I record what the assistants say.
Iris Ellison, MPH — publisher and analyst.
MPH in epidemiology & biostatistics (Boston University). Ten years in clinical informatics and digital-health analytics, most recently designing patient-facing dashboards for a health system.
I started AI Health Consensus in 2026 after noticing patients arriving with AI-generated nutrition advice no clinician had time to vet — and noticing, when I asked the same questions myself, that the four assistants agreed with each other far more than anyone was tracking. Nobody was writing it down. So I do.
The project
AI Health Consensus asks health questions — phrased the way patients ask them — to ChatGPT, Claude, Gemini and Perplexity, and publishes every answer with its sources.
Editorial standards
- A record, not a review. I don't test products and I don't rate them; I publish what the assistants said.
- No editing. First choices, other products named and sources cited are extracted exactly as each answer gave them.
- No ads, no affiliate links, no sponsored placements, no free products. Nothing on the site is paid.
- All answers published. Twelve per question — four assistants, three attempts each — not a selection.
- Fixes are logged. Anything that changes after publication appears on the changelog.
Publisher
AI Health Consensus is an independent project of Iris Ellison, MPH.
50 Milk Street, Suite 1300
Boston, MA 02109
US
Contact: hello@aihealthconsensus.com
The full method is on How I ask; changes are logged on the changelog.