STAT News: AI, Food Pyramids, and the Future of Health Information
How trustworthy is the new U.S. food pyramid? It’s a complex question, according to the government website devoted to it. Kyle Diamantas, head of the Human Foods Program at the Food and Drug Administration, recently alerted the public to a generative artificial intelligence tool added to the government’s “transformational” realfood.gov site. The tool, headlined “Use AI to get real answers about real food,” boasts “AI integration to provide parents and consumers with clear and concise answers at the click of a button,” as Diamantas wrote on X.
However, that click of a button directs users to Grok – the generative chatbot that’s part of the X social media platform owned by Elon Musk. When asked if the new food pyramid is backed by high-quality research, Grok responds: “Many nutrition scientists and organizations have raised concerns about the evidence quality and process for the final version.” While acknowledging the guidelines’ recommendations on limiting added sugars and ultra-processed foods are research-backed, Grok points out, “the emphasis on saturated fats and animal proteins contradicts longstanding evidence.”
This accurately reflects the nutrition community’s response to the new guidelines and pyramid, which prominently features a rib-eye steak and stick of butter alongside healthier options like broccoli, salmon, and olive oil. However, researchers are generally cautious about relying on AI for nutrition advice.
“I think the use of AI holds promise for providing tailored nutrition advice in a way that is convenient and low-cost,” says Alyssa Moran, a nutrition policy researcher and epidemiologist at the University of Pennsylvania. But she notes that generative AI, like human healthcare providers, can perpetuate stereotypes about eating and weight, potentially stigmatizing obesity. She emphasizes the need for extensive testing of these models before widespread public use – “and certainly before these tools are promoted by the government, which is supposed to be looking out for the public’s health.”
The National Design Studio, responsible for the dietary guidelines website, has not yet responded to requests for comment. The Grok tool on the site also includes sample questions, frequently using the phrase “REAL FOOD.” One asks about feeding a vegetarian, Indian-food-loving family “REAL FOOD” on a $200 weekly budget. Grok recommends lentils, rice, and buying in bulk.
Another sample question concerns “REAL FOOD” during pregnancy. Grok recommends “Folic acid/folate (400–800 mcg/day, often via prenatal vitamin),” listing food sources like fortified cereals and leafy greens. However, medical organizations specifically recommend folic acid supplements to prevent birth defects, rather than relying solely on food sources. Grok doesn’t make this crucial distinction, potentially reinforcing unfounded fears about folic acid supplements spread by some wellness influencers.
Regardless of these concerns, Americans are increasingly turning to large language models for health information. A recent study published in Nature Medicine highlights that how people phrase their questions and the information they emphasize impacts the accuracy of chatbot answers. “Evaluation mistakes are often made when systems are assessed in artificial environments,” explains co-author Adam Mahdi, an associate professor at Oxford Internet Institute, via email. This explains why chatbots may excel under ideal conditions but falter in real-world scenarios.
This means user experiences with Grok may vary. The STAT newsroom tested the tool, asking, given butter’s prominence on the food pyramid, how many sticks of butter one should eat daily. Grok’s response? Zero. The pyramid’s visual hierarchy shouldn’t be interpreted literally. “Ah, thank you,” the reporter replied. “That was very confusing. Can you suggest a better way to lay out food guidelines visually to avoid this kind of mistake?” Grok suggested abandoning the inverted food pyramid and reverting to MyPlate.
Further Reading:
STAT’s coverage of chronic health issues is supported by a grant from Bloomberg Philanthropies. Our financial supporters are not involved in any decisions about our journalism.
