What does ChatGPT say about you? How AI assistants describe researchers, and how to correct the record
AI assistants now answer questions about physician-scientists every day. Here is where their information comes from, why it is often wrong, and six practical steps to make it accurate.
Try it yourself
Open ChatGPT, Perplexity or Gemini and ask: “Who is Dr. [your name], and what do they research?”
Many researchers who try this are surprised. The answer may be roughly right, but it often mixes in an old appointment or misses your current focus. Sometimes it attributes a colleague’s paper to you, or a paper by someone with the same name. A collaborator, a journalist or a prospective patient asking the same question gets the same answer.
Where AI assistants get their information
AI assistants build their answers from what is published on the open web. When they can search the web, they typically rely on the pages that rank for your name. Those usually include:
- your institutional faculty profile, which is often updated only occasionally;
- Google Scholar, ResearchGate and PubMed author listings, which can merge people who share a name;
- physician directories and ratings sites, which describe your clinic, not your research;
- news coverage and press releases, which capture one moment in time.
When these sources disagree, or when none of them clearly states who you are, the assistant has to guess. The fix is to give it one clear, current and authoritative source.
Six steps to correct the record
1. Claim and clean up your ORCID record
ORCID is the closest thing research has to a universal identifier. Make sure your record lists your current employment, education and works, and connect it to Scopus and Crossref so new publications are added automatically.
2. Audit your Google Scholar profile
Remove papers that aren’t yours, merge duplicates, and turn off automatic additions if they keep pulling in other authors’ work. Make the profile public and add your verified institutional email.
3. Use one name, consistently
“J. A. Doe,” “Jane Doe” and “Jane A. Doe, MD” can look like three different people to a machine. Choose the form of your name you want to be known by and use it everywhere you control.
4. Ask for your faculty profile to be updated
Most institutions have a simple process for this, and the profile carries the institution’s authority. Make sure it states your current title, research focus and a link to your own website.
5. Publish a website on your own name
Search engines and AI systems both treat a website at your own name, such as janedoe.com, as a strong signal about who you are. Unlike every other source on this list, you control it completely. It should state, in plain language, your current role, your research focus and your key publications, and it should link to your ORCID, Scholar and institutional profiles so machines can connect them.
6. Make it machine-readable
Behind the scenes, a well-built site uses structured data (Schema.org markup) to describe you as a Person with an affiliation, credentials and authored works. It can also include an llms.txt file, a plain-language summary written for AI systems. People never see these details, but they help software understand your record correctly.
A note on guarantees
No one controls what an AI assistant says. Be wary of anyone who promises a specific answer or a number-one ranking. What you can do is make the accurate version of your record the easiest one to find, which is what these systems are built to reward.
Researcher’s Toolbox builds and maintains professional websites for physician-scientists, structured for search engines and AI assistants from the first line of code. See how it works.