We ran fifty patient queries through four AI systems. Doctors were named in under 20%.

We put fifty questions a real patient might ask — oncology, cardiology and orthopaedics — to ChatGPT, Gemini and Perplexity, and recorded what came back.

What the answers showed

Named clinicians appeared in fewer than 20% of responses. Hospitals were named readily; individual doctors rarely. Where a doctor was named they had a substantial personal web presence, indexed publications, and video content.

Educational pages were cited over institutional ones. Hospital About Us pages were almost never used as a source. Blog posts, FAQ pages and long-form clinical articles that directly answered the question were.

Schema presence tracked with citation. Every clinician named across the fifty queries had structured data describing them somewhere. Not one was cited on the strength of prose alone.

Review signals mattered independently of seniority. Doctors with substantial Google and Practo review volume surfaced more often than more senior colleagues without it.

What follows

These systems are biased toward findable clinicians — those whose credentials, published work and reputation are structured in a form a machine can parse. That is not a judgement about clinical quality and it will not correct itself. Clinical excellence without digital legibility is becoming invisible excellence.