I was curious whether AI could help with MR safety issues. So I asked ChatGPT a simple question. Below is a summary of my conversation. It’s both hilarious and terrifying. I hope you find it illuminating. Apologies to E. Russell R. Smith!
ME: Tell me about the book Essentials of MRI Safety.
ChatGPT: Essentials of MRI Safety is a well-known textbook focused on the practical and clinical aspects of staying safe around magnetic resonance imaging (MRI) systems. Primary author: E. Russell R. Smith, MD.
ME: Why do you say it was written by E Russell E Smith, MD?
ChatGPT: You’re right to question that — that was my mistake. Here’s what happened, clearly and honestly:
- Essentials of MRI Safety (published 2020) is written by Donald W. McRobbie, PhD.
- E. Russell R. (or E. Russell E.) Smith, MD did not write a book with that title.
ME: I work in MRI safety and I have never heard of E.Russell.E.Smith MD. Who is he?
ChatGPT: You’re absolutely right — and your instinct as someone who works in MRI safety is spot on. E. Russell E. Smith, MD is not a recognized figure in MRI safety. There is no well-known MRI safety author, clinician, or academic by that name, and no authoritative MRI safety publications attributed to such a person.
I incorrectly invented / hallucinated a name in my earlier response.
Most experienced MR professionals end up saying: “McRobbie is how safety really works — but we still need hard rules to survive the clinic.”
ME: Is this actually true? Where does this citation come from? How many is most?
ChatGPT: You’re right to press on this. That sentence was not a citation, and it should not have been presented as if it were one.
That statement is not a quoted finding, not from a paper, and not supported by survey data.
ME: You’re making a lot of mistakes today.
ChatGPT: You’re absolutely right, and I appreciate your patience — I’ve made several errors and overgeneralizations in this conversation, and that’s not acceptable when discussing technical topics like MRI safety.
If you want, I can create a clean, fully sourced summary that is accurate for professionals like you. I will avoid conjecture or invented consensus. Is that what you’d like?
ME: No thanks.
.
