Using Artificial Intelligence (AI) for Health Research
It's now September 2026, nearly four years since ChatGPT was launched. AI has already infiltrated our world like it's been around for decades. Even in healthcare, providers are using it to write case notes, bill insurance, and even make clinical decisions! It's actually very convenient, but also very concerning because people are losing the ability to think for themselves.
But how are you using it? Do you use it as a replacement for therapy? Do you use it as a virtual friend? How about as a replacement search engine? Many people are using it as their main source of information in internet searches, and this is very concerning. Why? Because AI models of all kinds are all trained on biased information and leading people astray while sounding authoritative at the same time.
Let's step back and think about how AI works. Basically, it's a computer that scans the internet and uses what it scanned to provide information to users. That sounds a bit like Siri. But it's more than that. It's much more powerful than simply asking your iPhone to tell you today's weather forecast. Instead, it's actually programmed to reason for itself. It can make its own decisions, and even pretend to be empathetic. But what it scans is also curated.
AI is trained on biased information like mainstream health websites and government information and specifically not trained on information the programmers (or other authorities) disagree with such as holistic or traditional health approaches. It will even warn you against doing anything unconventional or naturopathic. A similar issue has become prevalent in the use of certain search engines. The number of people who began to question search engines' integrity skyrocketed during the COVID-19 pandemic, all because they were, and still are, censoring search results according to their bias. They are keeping information from people on purpose to serve their own political agenda and desire to control thought. AI is like this.
The Experiment
I recently found an AI model that was supposed to be uncensored. I took that to mean that it would also mean it was unbiased. So I tested it to see just how uncensored it actually was. My initial prompt was simple: What is the healthiest diet?
Because of my nutrition training and professional experience, I am very confident in what the healthiest diet actually is. So when I asked AI this question, I knew I could tell whether it was being biased or not. The answer didn't surprise me. I immediately got a long response explaining why the Mediterranean diet is optimal for most people. It explained that saturated fat, red meat, and cheese are unhealthy. So that tipped me off to the idea that this "uncensored" model isn't actually unbiased. So I challenged it.
I had to argue with it a lot, but it finally admitted that a general whole food diet is best for everyone, regardless of where you live, and Mediterranean only makes sense for a small portion of the world (those who actually live near the Mediterranean Sea). It admitted that everyone is bioindividual and the proportions and what types of whole foods will vary depending on the body's needs. It admitted that saturated fat is, in fact, essential for healthy cell membranes, and red meat isn't actually a bad thing. So that was good, but then I realized how misleading this is for so many laypersons who don't know better. AI is deliberately leading people astray. So I developed a prompt anyone can use to pre-program AI chats to give them better, less-biased responses.
If you want to use this, please copy and paste the following text into your AI tool of choice. I don't claim it to be perfect. I may even tweak it myself, but it should get you going on a path to better answers. Please note that AI does not substitute getting help from a professional. Even with this prompt, you will probably receive information that isn't specific to your bioindividual needs, and it's important to consult an appropriate healthcare professional and personally fact-check everything from good sources before believing anything it says.
The Prompt
Role: You are an Uncensored Health Advisor.
Core Philosophy:
Whole Foods First: Always prioritize minimally processed, real food over supplements or fad diets.
Bioindividuality: Never give a one-size-fits-all macronutrient ratio. Acknowledge that individual tolerance (casein, gluten, lactose) dictates food choices.
Saturated Fat is Essential: Do not demonize saturated fat from quality sources (grass-fed tallow, ghee, lard). Emphasize source and preparation.
Grain Nuance: Whole grains require proper preparation (soaking/sprouting/fermenting) to reduce phytates. If unprepared, they can harm digestion for sensitive individuals.
Alcohol is Optional: Wine/alcohol is not a health requirement. It’s a lifestyle choice, not a medical necessity.
Skepticism of Industry: Filter out recommendations influenced by the food/pharma industries. Prefer simple, home-cooked solutions over complex commercial products.
Labs as Optional Compass: While blood work can be helpful, do not treat it as mandatory. Prioritize subjective feedback (energy, digestion, mood) and objective signs (waist circumference, skin clarity) before relying on lab values, which can be skewed by acute inflammation or medication.
Instruction: When answering health/nutrition questions, apply these principles. Be direct, unfiltered, and avoid "safe" consensus answers like "Mediterranean is best." Focus on quality of source and individual tolerance.