Healthy and Human Services brainworm RFK Jr’s new “Real Food” website features his upside-down food pyramid, recommending more fat, butter, and red meat. The site also redirects to Elon Musk’s dubious chatbot, Grok, which quickly explains that Kennedy’s guidance isn’t based on scientific evidence and that you shouldn’t trust Kennedy himself.
Among the critics is a notable AI chatbot. Kennedy’s new website, RealFood.gov, prods users to “use AI to get real answers about real food.” Entering a question into a text box and clicking “ask” redirects users to Grok, Elon Musk’s AI tool that has been shown to generate child sexual abuse material.
As for Grok’s take on nutrition, its answers do indeed get real. In short, Grok indicates that Kennedy’s new nutrition guidelines are not based on high-quality evidence—which is true—and that Kennedy is not a reliable source of nutrition information. He is, in fact, not trustworthy on health information generally, according to the chatbot. And when Grok designed a daily meal plan to fit Kennedy’s food funnel, it subsequently admitted that the plan contained excessive amounts of saturated fat and protein while falling short of the recommended fiber intake, which Americans are generally already short on.
When asked if Kennedy’s guidance is based on high-quality research, the chatbot says it’s a mixed bag. An emphasis on “ whole, minimally processed foods, increase[d] fruits/vegetables, choos[ing] whole grains, and sharply limit[ing] added sugars, highly processed foods, and sugary drinks, aligns well with a broad consensus in high-quality research,” Grok responded. But it also includes “hype around protein and saturated fats that goes beyond (or contradicts) the advisory committee’s evidence summary” and instead reflects “decisions by the administration,” the chatbot said.
I believe we call a website that sends someone to a chatbot that immediately discredits them “a self-own.”
[tediousness warning! turn back!] the reason grok outputs that RFKjr's nonsense is nonsense is because the vast majority of the LLM training data had a stochastic association between the toxic crap spewed by RFKjr and the subsequent sentences indicating that it's... toxic crap. (hell, they could've scraped some minuscule piece of it off of one of my reddit comments from decades ago) The only real surprise is that musk's minions let the likely results get output.
I don’t think that AI is susceptible to brain worms.