Artificial intelligence and Gluten Free Watchdog
Just an FYI that I have been misquoted in some recent articles and am working to get them corrected. The most recent incidents involve author use of artificial intelligence. Please know that AI “hallucinates” and sometimes makes up references and quotes. I try to be as responsive as I can to any questions from media organizations and individuals, so please don’t hesitate to reach out if you would like to ask a question or are seeking a comment! There is no need to use AI.
Comments (9)
Hello Tricia
Great points on AI generated responses to questions. You will never get a “I don’t know. I will look into it” response from AI. While useful, Ai still takes a subject matter expert like you to VALIDATE responses.
Cheers
Frank
Hi Frank. That’s a great point about AI and “I don’t know” responses. Thanks for commenting!
Frank, you made some excellent points, and I especially like the “I don’t know” response from AI (not).
This is a great reminder that if an individual, media outlet, book author, blogger, etc wants ACCURATE information, especially in the form of a QUOTE, they need to GO to the SOURCE. Otherwise, it’s unprofessional, irresponsible, unethical, fake information, and damaging to someone else’s work.
I 100% agree!
THANK YOU for talking about this!!
Generative ai is at best inconsistent with its responses and at worst, dangerous and even deadly. There’s been many times where it’ll spit out recipes that include ingredients like wood glue or chemical cleaners (there’s been at least one instance where using Draino as lye for pretzel making was suggested in an ai generated recipe) provide cooking times are far too low and people end up getting salmonella and e. Coli from ingesting under-cooked meat.
It’s even more of a problem when you recognize that half of the adults in the United States only have a 6th grade level reading comprehension.
Everyone should hold a good amount of caution when looking things up these days.
Hey Tricia
Thank you for bringing this to our attention. I have never used CHATGPT until today. You made me curious so I asked:
“Are oats for sale labeled “gluten free” really gluten free? How is a company allowed to use a gluten free label? What organization certifies the oats as gluten free? What is required to be certified gluten free?”
My goodness (that’s about the most innocuous response I could come up with).
I’m not sure if you have looked but CHATGPT uses many of your older posts as references.
After the initial lengthy response to my questions CHAPGPT said:
If you want, I can also:
Compare Purity Protocol vs. Mechanically Sorted oats
Check whether a specific brand is considered safe
Explain how to read gluten-free certification labels
So I asked
“ Are GF Harvest oats purity protocol and mechanically sorted? Are they safe?”
CHATGPT:
Good questions. Short answer: it’s complicated. The short version is: although GF Harvest Organic Oat Groats (and other GF Harvest oat products) claim to be “purity protocol” oats and historically many considered them among the safer gluten-free oat options, recent independent testing has cast real doubt on whether they are reliably free of gluten.
(Gluten Free Watchdog +2)
One of the references is to your Sept 28 2022
“Special Statement on Purity Protocol Gluten-Free Oats from Gluten Free Watchdog”
I can’t figure out the 2nd
>Here’s a breakdown of what we know — and what remains uncertain — about GF Harvest and “purity protocol” vs. “mechanically sorted” oats.
Again using 2 GFW refs
One is Mar 8 2021
“Oats are complicated! Updated Position Statement on Oats from Gluten Free Watchdog”
Couldn’t pull up the other
CHATGPT also uses the GlutenFreeOats website as a reference
https://glutenfreeoats.com/purityprotocol-org/
~~~~
I won’t continue to copy all this-people can use my questions & see what is written
After many paragraphs it ( I guess AI is an “it”) states:
So — Are GF Harvest Oats Safe?
It again uses you as a reference and also two Reddit posts (!)
At the very end it says
“ If you like — I can pull up a current (2025) list of oat brands that have recently (past 12–18 months) tested below 5 ppm and are able to demonstrate robust testing + transparency. That helps show which oat products are considered safer now (given evolving evidence). Do you want me to build that list for you now?”
(Sure!)
It responds with many of your articles, Reddit again, glutenfreeoats again & 8 other sources.
All but one of these are older articles.
It concludes the list with
“What Brands of Oatmeal are Safe (2025 guide)”
The link does not work. It timed out or was not available on 4 web browsers. There was a blog in 2018 using WordPress but it’s no longer available.
So much for the 2025 guide!
I did not ask anything more re specific brands after this gobbledygook. GFW was so often quoted using older blogs you have since updated.
Then all the CHATGPT disappeared. Does it time out? Does one need to pay to play?
I went to Reddit. Not AI but a lot of people use it as a reference and one responder mentioned the trouble with AI
On a recent post (26 days)
“What’s the deal with oats” one responder said they turned off Google AI at work bec it was quoting 10 yr old archived articles re gluten free oats that have since been disproven.
Wow
The plot thickens and not in a good way.
Thanks again, Tricia, for things like this and all you do.
Thanks so much for posting. It is all a bit concerning isn’t it. I asked AI (forget which one), to use only GFWD posts to answer the question how has Gluten Free Watchdog’s position on oats changed over the years (or some such wording). It did a decent job. BUT I became aware that AI was quoting made-up information/citations from GFWD that were being used by other groups in public posts/articles. From what I’ve been told by those well-versed in AI is that it sometimes hallucinates and creates quotes and citations that simply don’t exist. If anyone comes across a far-fetched statement from GFWD, please let me know!!! And to anyone using AI, check the actual citation–make sure the link works, there is no error message, and the information being cited is actually included in the article.
Sara, Wow! kudos to you for having the patience to research and follow up on this CHATGPT app. And thank you for sharing it with GFWD and all of us. I help support people with CeD through a national organization, and I am concerned that many of them search online, use AI, Facebook, etc. for their information rather than going to a scientifically based, reliable source. The research experience you share here shows how unreliable AI is.