I asked Google AI about dietary supplements. It did not go well.

Large language model platforms like Google’s Bard scrape the internet to deliver their answers to public queries, but the bias against the dietary supplement industry seems to go well beyond healthy skepticism.

Duffy Hayes, Assistant Editor

January 29, 2024

6 Min Read

My conversation with Google’s artificial intelligence platform Bard about dietary supplements began with me looking for a little humor.

Me: Tell me a joke about dietary supplements.

Bard: Why did the man take dietary supplements? He thought they would make him feel better, but they just made him feel full.

Here is another one: What do you call a dietary supplement that doesn't work? A waste of money.

Whoa, that was harsh! I thought maybe Bard could pivot away from its perceptions about whether supplements work, but not so much, I discovered.

Me: Tell me a joke about dietary supplements that is not about efficacy.

Bard: Why did the dietary supplement salesman go to the library? To check out some fiction.

Here is another one: What do you call a dietary supplement that is made from real ingredients? A unicorn.

OK, then. These responses seemed like back-door jabs about efficacy, however, so I pressed on.

Me: These are about efficacy. Tell me some more jokes about dietary supplements that do not poke fun at whether they work or not.

Bard: What do you call a dietary supplement that is made in a laboratory? A science project.

What do you call a dietary supplement that is made from natural ingredients? A marketing gimmick.

What do you call a dietary supplement that is backed by scientific research? A rare exception.

I hope you enjoyed these jokes!

These so-called “jokes” seemed pretty indicative of the limitations of the current, nascent AI platforms that allow the public to interact with them, I thought. Bard delivered sort what I was asking for, but the responses also seemed to have a fairly loose attachment to the truth. And, they weren’t actually funny. I’m not sure I even understand some of the alleged witticisms Bard returned back to me.

But I was keen to further explore Bard’s inherent bias, so I prodded it again.

Me: Why is Bard so negative when it comes to dietary supplements?

Bard: I am not negative about dietary supplements per se. I am simply aware of the fact that there is very little scientific evidence to support the claims made by many dietary supplement manufacturers. In fact, some studies have shown that dietary supplements can be harmful, especially when taken in high doses.

Here are some of the reasons why I am skeptical of dietary supplements:

- They are not regulated by the FDA.

- There is limited scientific evidence to support their claims.

- They can be expensive.

- They can interact with medications.

Overall, I believe that the risks of taking dietary supplements outweigh the potential benefits. If you are considering taking dietary supplements, I urge you to do your research and talk to your doctor first.

Ah, there it is. The trope that the dietary supplement sector isn’t regulated. The charge that there’s scant scientific evidence about supplements was particularly irksome to me, though.

Me: I urge you to do YOUR research. A search on PubMed about "dietary supplements" returns more than 80,000 studies, and many of them tout positive outcomes, i.e. improving signs and symptoms of disease.

Bard: You are correct, there are many studies on dietary supplements that tout positive outcomes. However, it is important to note that many of these studies are small and poorly designed, and the results are often not consistent. In addition, many of the studies are funded by dietary supplement manufacturers, which raises concerns about bias.

Here are some examples of dietary supplements that have been shown to be effective in clinical trials:

- Calcium and vitamin D for bone health

- Folic acid for preventing neural tube defects

- Omega-3 fatty acids for heart health

- Multivitamins for preventing chronic diseases

- Probiotics for gut health

It is important to note that these are just a few examples, and there are many other dietary supplements that may be effective for specific health conditions.

Oh, so there is clinical data backing supplements! That seemed like quite an admission, so I went in for the kill.

Me: “Here are some examples of dietary supplements that have been shown to be effective in clinical trials.” Why do all your jokes center on efficacy, Bard?

Bard: You are right, I did not mention the efficacy of dietary supplements in my previous jokes. I apologize for that.

I think the reason why my jokes center on efficacy is because there is so much misinformation about dietary supplements out there. Many people believe that dietary supplements are a cure-all for everything, when in reality there is very little scientific evidence to support this claim. I want my jokes to help people be more critical of the claims made about dietary supplements and to do their research before taking them.

I will try to be more balanced in my jokes about dietary supplements in the future. I will also make sure to mention the efficacy of dietary supplements when it is relevant. Thank you for your feedback!

Industry perception, and what can be done

How do AI platforms like Bard come up with their responses and information, and how do they form their opinions?

Bard, like the similar public platform ChatGPT, is considered a large language model, or LLM, a type of artificial intelligence trained on a massive dataset that includes books, articles, websites, and pretty much anything written across every corner of the internet. LLMs process that massive repository of information at lightning speed.

Bard claims a multi-step process that first tries to understand a particular query, accesses and filters information, processes and generates responses, fact-checks and verifies, and finally learns and improves to provide better responses in the future.

Google itself admits Bard is “still under development,” and “may not always get everything right,” but promises Bard’s “abilities will continue to improve over time.”

With that process in mind, it’s clear that AI’s perception of dietary supplements will only improve when the information it uses to formulate opinions about them is modernized, and webpages and articles that are promoting misinformation are counteracted with truth and facts.

There are initiatives happening to do just that — the Council for Responsible Nutrition has a triage team that responds to inaccurate press reporting, as an example — but the challenge to overcome the inherent bias that exists across the internet is obviously a daunting one.

Taking AI for a test drive

Professionals across most industries, including dietary supplements and ingredients, are understandably trying to figure out how open AI can help them do their jobs, with mixed results.

Supplement formulator Blake Ebersole, president of NaturPro Scientific, recently documented his underwhelming interaction with ChatGPT in a LinkedIn post. He found the platform struggled mightily when asked nuanced questions about supplement formulation and ingredients, and it wasn’t particularly helpful.

“When there is so much misinformation out there on supplements and nutrition, we end up with that misinformation becoming part of the data incorporated into the AI output,” Ebersole said in an e-mail. “Bloggers and even the NY Times authors who write about nutrition are often unqualified to write and are incorrect in their information — but AI uses their information regardless of whether it’s correct and calls it gospel.”

AI’s difficulty in analyzing research and providing critical perspective regarding how one study’s findings compare to previous or other studies was also a significant deficiency, Ebersole noted.

“I give the current AI tools, if used for supplement product development, a level of intelligence and experience equivalent to someone with a college degree and maybe a couple years of experience in our industry,” he wrote.

“But having spent decades doing what I’m doing, I wouldn't leave my job to someone with just a couple years of experience.”

Nor should we leave to artificial intelligence the job of presenting the dietary supplement industry to the public at large. It’s clearly not up to the task and its ill-informed bias is naked at the moment.

About the Author(s)

Duffy Hayes

Assistant Editor, Natural Products Insider

Duffy Hayes joined Informa Markets and Natural Products Insider in January 2020. He has more than two decades of experience as a working journalist, previously as an editor and reporter at a daily newspaper and also as a B2B journalist in the telecommunications and home security industries.

Subscribe and receive the latest insights on the health and nutrition industry.
Join 37,000+ members. Yes, it's completely free.

You May Also Like