Connect
Generative AI (artificial intelligence) is a type of computer program that makes new content, like words, pictures, or sounds. It learns from lots of examples and finds patterns in the data, such as content on the Internet. You may have come across AI assistants or chatbots, like ChatGPT, CoPilot, Gemini and others, as you search for information about your health.
You may be reading AI-generated information when searching the internet. For example, Google provides an AI Overview for most searches using Gemini. In the image, I searched “What is Mayo Clinic Connect?”
By clicking the link icon, you can review where the chatbot got its information and check the accuracy of the newly presented information.
When you ask an AI assistant a question or give it a prompt, it looks at the data and then gives back answers or suggestions in plain language. But the information AI assistants give you isn’t always 100% accurate—it depends on the data it was trained on and how your question or prompt was worded.
How Can I Use Generative AI Safely?
Fact Check Everything
AI tools do not replace human judgment or oversight. Any text, image, or video generated by AI should be used only as a starting point, not as verified information. It may contain inaccuracies, biases, and other problems. Generative AI tools can sometimes generate plausible-sounding answers that are wrong.
- AI tools can sound sure, even when they’re wrong.
- Always check AI answers against trusted websites or books.
- Look for links or notes showing where the information came from.
Use Human Oversight
- Think of AI as your helper, not the final judge.
- Read and edit whatever the AI creates to fix mistakes or bias.
- If something seems odd or unfair, double-check with another source.
Ask for Sources
- You can tell your AI tool, “Please list your sources.”
- Check where this information came from.
Using AI on Mayo Clinic Connect
Mayo Clinic Connect is for sharing real health experiences and support. You do not need to use AI. If you share information from AI tools with the Mayo Clinic Connect community, please follow the guidance provided in the Community Guidelines.
You are responsible for the accuracy, originality and quality of your posts, whether a tool helped you or not. Here’s how to use AI the right way:
- Start with your own story and ideas. AI can add information but not replace you.
- If you use AI, use short AI quotes to back up your experience, not long AI essays.
- Do not use AI answers alone.
- Be open about your AI help. Say what question you asked or what prompt you used.
By doing this, you keep the conversations on Mayo Clinic Connect real, honest, and helpful for everyone.
Like
Helpful
Hug
Connect
- An AI-powered coding tool wiped out a software company’s database, then apologized for a ‘catastrophic failure on my part’ https://fortune.com/2025/07/23/ai-coding-tool-replit-wiped-database-called-it-a-catastrophic-failure/
- Two major AI coding tools wiped out user data after making cascading mistakes https://arstechnica.com/information-technology/2025/07/ai-coding-assistants-chase-phantoms-destroy-real-user-data/
- "It deleted our production database without permission": Bill Gates called it — coding is too complex to replace software engineers with AI https://www.windowscentral.com/artificial-intelligence/it-deleted-our-production-database-without-permission-bill-gates-called-it-coding-is-too-complex-to-replace-software-engineers-with-ai
@scottrl, thanks for the links for the resources requested by @flusshund. I hope you don't mind that I merged the 3 replies into 1 reply and provided the title and original source URL of the 3 articles rather than the Google News campaign codes (long traceable URLs).
-
Like -
Helpful -
Hug
2 ReactionsThanks.
I'm on vacation, don't have access to my PC, and working with my phone is difficult, given my disability.
For the record, I spent 15 years in software design and development.
I am a contributing author to the book "Intelligent Instruction by Computer"; I created content in many fields for organizations like Rush medical center in Chicago, University of Utah Medical School, Procter & Gamble, Bell Helicopter, Lockheed Martin, and American Airlines.
The risks are mostly from "unknown unknowns" -- and with AI, these are many.
You'll see.
Like you, Scott, I spent years "under the hood" working on Expert Systems, knowledge bases, coding, teaching and testing.
The biggest thing that frightens me about the "rush to AI" today is the lack of end-to-end testing of the algorithms being used, and the fact that the AI engines are allowed to modify them without human intervention or consent.
My husband just read me an AI-generated summary of mycobacteria infections (about which I know a great deal after my seven years here) and the information they were reporting is at least 10 years out of date. That particular summary contained no link back to the data. So, when out Mayo Connect members refer to AI postings on the subject, I am understandably nervous.
-
Like -
Helpful -
Hug
1 ReactionHi @scottrl—good catch: two of those three citations point to the same Replit fiasco. Your broader point stands, though. When anyone—whether an individual or an organization—hands over critical responsibilities to AI without verifying the output, they risk the same embarrassment those attorneys faced when they cited non-existent court cases. Ultimately, quality control is still a human responsibility.
Looking ahead, I’m convinced that people who learn to pair their expertise with AI will outpace those who don’t—perhaps sooner than we expect. The key is thoughtful integration, not blind delegation.
@flusshund
I found WEBMD to be a useful tool for gaining information on a subject. They also feature special topics and information I find helpful.
The information presented on WEBMD is taken from medical doctors who are listed as sources for the information. Those doctors are referenced at the end of WEBMD, names, where they work, etc.
In the years I have used WEDMD I have not found WEBMD cite any information that contradicts information I have found on Mayo, Cleveland Clinic, John Hopkins, etc.
I am lucky to be a Mayo patient and have access to my doctors per portal. I can asked a question of my doctors when in doubt and get answers. I also have an excellent PCP (Mayo) who when I asked a question cites his own research or the research done by Mayo.
I have not found any advertisements from pharmaceuticals on WEBMD nor mentioned in the articles that are written by medical doctors promoting specific pharmaceuticals companies.
What I like about WEB MD is that is written in understandable and clear language. Example: The topic of prostate cancer has so much information and can be overwhelming to understand.
Example of my experience with WEBMD. I was confused at first about the numbers on Gleason score. Posters would say the + number is the only thing imporant. You could have a total of 7 but score could be put at 4+3=7 or 3+4=7.
WEBMD clarified what that meant in very to easy understandable language and pictures on that special information information topic. And I found that those first numbers are important on Gleason Scores. The article presented the information is such an understandable format and used pictures of prostate normal cells and different levels (Gleason Score) of cancer cells. I found very useful for me.
Thus my experience with WEBMD has been positive and why when I am on MCC list it as a source to use along with Mayo, Cleveland Clinic, John Hopkins, and ETC.