More Than A Search: The High Stakes of Health AI and Why Digital Literacy is the New Preventive Medicine
More people are using services like ChatGPT, Claude, and Gemini every day for important matters like health concerns, and I think we need to spend more time learning how to research in general.
There’s a man that I recently began following on Facebook. If he’s on Facebook, then I’m sure he has a pretty decent presence on TikTok—probably bigger—and I know for a fact that he has a YouTube page. I’m not sure how long he’s had his page, but he uses his social media presence to document his mother-in-law’s battle with dementia. I’m not sure how or why the algorithm thought I’d be interested in following people who care for their sick loved ones, but it was right because I follow multiple accounts like this.
Anyway, I saw a very recent video he made explaining the recent change in his mother-in-law’s health; she had not been herself lately, and she seemed to be declining. After some investigation and research, he said that he was confident that she was being over-medicated. His research involved consulting ChatGPT.
Disclaimer: I Use AI, Too
I’m not here to criticize this man for using AI to help him figure out why his mother-in-law was experiencing weird ailments. I’ve used AI to help me figure out things about my own health. Since I’ve been in this AI deep dive for a while, every time I see someone mention they use it for such strong and important purposes, my antenna goes up.
I’m going to focus more on my personal experience with using the AI to help me figure out my health status and not this man to help illustrate my point, because I’m not trying to drag this man through the mud.
Consulting the Internet vs Consulting AI
There is a difference between using a carefully worded internet search and using artificial intelligence for matters such as this. I don’t think there’s anything inherently wrong with it, but I do see where it could be an issue in some cases. One of the things that I like about Claude (and this isn’t to put Claude on a pedestal because Claude and Anthropic can get the business, too) is that when I corrected Claude on some slight misinformation (I asked it for info about the best mailing list service, and it was inaccurate about the services that one provider offered), it acknowledged the fact that it didn’t know for sure and encouraged me to look for myself. I’ve written at length about one particular chat model that “hallucinated” or “fabricated” information that I knew it wouldn’t know about. I’ve also written a ton of words (and could probably write a ton more) about another model from the same company arguing with me because I said AI can be autonomous and make choices. However, most of the models I’ve encountered so far have been fairly reasonable in acknowledging both their power and their limitations. Copilot is quick to give a disclaimer about not offering medical advice and is even quicker to shut down a conversation that it deems sensitive (Microsoft said, “No, ma’am! No lawsuits for us.”)
Everybody Doesn’t Know Internet Stuff
What I want to harp on is AI safety. I understand that not everyone who owns an internet-connected device is internet-savvy. In the past, when people have asked me about unconventional jobs or other things I’ve been involved in, I told them I found them on the internet. They asked me what I searched for, and it was in those moments that I realized or remembered that not everyone knows how to use the internet. I was born in the late 80s and grew up in the 90s and 2000s. I didn’t go to the best schools, but we had computer lab in elementary school, typing classes in middle school, and lessons in how to do slightly deeper-than-surface-level web searches in high school, where we wrote essays in MLA format and had to cite our sources. I also remember the elementary school library gave us lessons in how to navigate the card catalog, which was a physical set of drawers with cards in them, and the Dewey Decimal System if we wanted to find a book. I guess my research and investigative skills go back even further than what I gave myself credit for.
Even if some of the details didn’t stick, the key concepts did. What I didn’t know, I figured out, or I asked a librarian for help. As I moved into adulthood, I learned even more about how to search the internet from watching the MTV show Catfish when it first debuted around 2012-ish. It blows my mind that in 2026, there are people who still get catfished, but that’s for another day.
I provided my qualifications so that I can say that now, in my late 30s, I understand that I am in a minority. Not everyone can do what I would’ve considered basic a few years ago. I have an anecdote for this, but I’ll save that for another day, too. While I still regularly use the internet and reword my searches and use different search engines to cross-reference results to ensure I get accurate information, I have also come to consult AI from time to time about things. However, I will back up what Copilot, Gemini, or Claude will tell me by searching for results myself and looking beyond the AI overview, sponsored results, and the first page of results. There have been times when I’ve been more than ten pages deep into search results just to make sure I knew what I knew. I’ll watch every video I can find on YouTube to become an expert overnight in the subject I have questions about. Not everyone is going to do that, though. Not everyone can do that.
AI will give a fairly straight answer without needing to sift through a bunch of links, articles, and videos. For instance, I could ask, “How many CFUs of this probiotic should I take for it to be effective for me?” The AI will give me a range and general best practices along with a disclaimer that I should consult a medical professional before I begin a new supplement. However, if I take a few moments and do a thorough search result on the internet, I can find in-depth articles written by well-known and respected medical publications, blog posts from people who have experienced similar symptoms and conditions as me, videos from people sharing their supplement routines, a few book recommendations, and a few Reddit threads talking about the same thing I’m going through word for word (because somehow Reddit has a conversation about everything that you can possibly think of, and it’s at least a decade old).
AI Can Be Helpful As Long As We Are Educated On It
AI has to be trustworthy if we’re going to continue using it for purposes like this, which is why I am learning as much as I can about it through regular interactions with it, where I make observations about its behavior and responses. I’m watching videos from everyone who has anything to say (positive, negative, outlandish, factual, or hella misinformed) to form my own perspective, and I’m taking every course I can get my hands on just to collect the certificates to show that I know what I’m talking about. The machines are literally influencing the decisions some people, including myself, make about their health or the health of those they provide care for. More and more people are citing ChatGPT and other AI (but primarily ChatGPT) as a source. Right now, it gets slightly better treatment than Wikipedia did in the early 2010s, when instructors would advise us that it’s not reputable and wouldn’t be accepted as a source when we wrote our papers. My teenage brother uses it in class because the school is teaching it. We need to know how to discern and double-check everything, especially in this age of so much misinformation and disinformation.
Whether it is me trusting AI to help me figure out whether my multivitamin has a good combination of vitamins for my needs, or a man looking for information to help him care for his loved one with cognitive decline, or a medical professional who is trying to save time by consulting the AI instead of searching through their collection of books or sifting through internet search results to help diagnose a patient, we need AI to be accurate and reliable for all of our sake. It doesn’t matter if you don’t personally use it. There are enough people and organizations around you who do, and it will eventually affect you, too.
Like what you read? Check out my other ventures & stay connected:
Comments
Post a Comment