Most of the health and nutrition advice you're getting from AI is wrong. Not "maybe" wrong. Confidently, convincingly, and dangerously wrong.
I'm not saying AI is useless. I use it every day. But here's the problem.
When I ask ChatGPT about nutrition, it gives me answers that sound right. It cites studies. It uses confident language. It tells me things like "a balanced diet includes whole grains and lean protein" because that's what the internet says.
I've been self-studying nutrition for over 10 years. I've reversed pre-diabetes and cured 14 years of panic attacks through diet alone. So I can spot when it's wrong.
Most people can't.
AI Doesn't Know Anything
After working with AI for a few years as an SRE (Site Reliability Engineer) at a major software company, here's the thing you have to understand about how AI actually works.
It doesn't know anything. It doesn't have beliefs. It doesn't reason about what's true.
What it does is predict the next word.
You give it a prompt, and it asks itself: given everything I've been trained on, what word is most likely to come next?
Then it picks that word. Then it does it again. And again.
That's basically it. That's the whole thing.
So when the result sounds intelligent, it's not because the machine understands nutrition, it's because the machine has been trained on an enormous amount of text written by humans, and it has gotten very good at predicting what a human would probably say next.
This is what I call the 'Averaging Problem.'
AI doesn't give you the truth. It gives you the average.
It takes everything the internet has ever said about a topic, finds the statistical center of all that text, and hands it back to you in polished, confident-sounding sentences.
If the internet's consensus on a topic is correct, the average will be useful.
If the internet's consensus is wrong, the average will be confidently, convincingly wrong.
And you will have no way to tell the difference unless you already know the subject well enough to catch the errors.
The Internet's Nutrition Consensus Is Broken
When it comes to health and nutrition, the internet's consensus is not just occasionally wrong. It's systematically wrong.
The training data AI pulls from includes decades of industry-funded studies designed to shift blame away from sugar and toward fat.
It includes dietary guidelines written by committees that were, in many cases, advised by people with financial ties to processed food companies.
It includes the same "balanced diet" and "everything in moderation" messaging that has coincided with the worst metabolic health crisis in human history.
And don't think this is a conspiracy theory. That would be lazy of me and unfair to you (which I talk about in an up and coming newsletter).
It's well-documented.
The sugar industry paid Harvard scientists in the 1960s to downplay sugar's role in heart disease and point the finger at saturated fat instead.
That single piece of corrupted science shaped five decades of dietary advice.
It's still embedded in the training data. It's still in the average.
So when you ask ChatGPT what to eat, it doesn't consult biochemistry.
It doesn't look at comparative anatomy. It doesn't check the latest rigorous metabolic research.
It averages everything it's ever read about nutrition, and the average says: whole grains, lean protein, everything in moderation, calories in versus calories out.
The average got us here.
What the Average Misses
Here's what the average will never tell you.
It won't tell you that the same dietary advice that helped you get lean might be destroying your mental health.
It won't tell you that some people reverse their panic attacks by eliminating certain foods, because the average doesn't account for individual response.
It won't tell you that the studies it's citing might have been funded by the same companies that profit from the advice those studies support.
It also won't tell you that all human health studies are based on inferential statistics, which CAN NOT establish cause and effect.
The average doesn't know what worked for you. It doesn't know what failed.
It can't ask you to audit yourself honestly and check whether what you're doing is actually producing results.
That's the fundamental limit of AI in health. It can give you the consensus. It cannot give you the truth. And the distance between those two things, in nutrition, is enormous.
What Actually Works
If you can't trust AI, and you can't trust the internet's consensus, what do you trust?
Here's the framework I've settled on after 10 years of self-study, after reversing pre-diabetes, after curing panic attacks that no doctor connected to diet.
First, do your own due diligence.
Do not outsource your health decisions to a language model.
AI can help you articulate what you already know. It can help you organize your thinking. It should not give you the answers.
If you don't know the subject before you ask the question, you have no way to evaluate what comes back.
You'll take confident-sounding nonsense as truth because it feels right.
Second, follow the hard science.
Not the epidemiology. Not the food-frequency questionnaires where people try to remember what they ate six months ago.
I mean biochemistry. Comparative anatomy. Mechanisms.
The kind of science that asks: what actually happens in the body when you eat this? What are humans biologically adapted to eat? What does the metabolic pathway look like?
Third, audit yourself honestly.
A lot of health advice out there will help you get lean. It will not necessarily help you feel good. It will not necessarily improve your mental health.
You have to be honest with yourself about whether what you're doing is actually working, across all the dimensions that matter. Not just the mirror.
Your energy, your mood, your sleep, your anxiety, your ability to think clearly.
If the advice says you should feel great and you don't, the advice might be wrong.
For you, specifically.
Fourth, reject social proof and authority as substitutes for evidence.
The fact that a million people follow a certain diet does not make it correct. The fact that a credentialed expert endorses it does not make it correct.
The fact that it's the official recommendation does not make it correct. Popularity and authority are not evidence. They are shortcuts, and shortcuts in health are expensive.
The Filter Is the Value
There's a bigger point here, and it goes beyond nutrition.
Everyone has access to the same AI now. The same ChatGPT. The same information. The same averages.
If information were the thing that mattered, everyone would be equally capable of making good decisions about their health, their work, their life.
But they're not. And the reason they're not is that the value was never in the information. It was always in the filter.
The filter is your years of study. Your lived experience. Your mistakes. The things you learned the hard way.
The patterns you can recognize because you've seen them before. The bad advice you can spot because you followed it once and it made everything worse.
AI can't replicate any of that. It can't live in your body. It can't make your mistakes.
It can't spend a decade following a thread and arriving at a conclusion that contradicts the consensus.
It can only average what the consensus has already said.
The people who use AI best are the ones who already deeply understand their craft.
AI sharpens their thinking. It doesn't replace it. It helps them articulate what they already know. It doesn't give them the answers.
Finding the Truth Is Hard
I'm not saying you should ignore AI. I use it every day. I'm saying you should treat it like a research assistant who has read every book in the library but has never actually lived in a human body.
It can summarize. It can organize. It can help you see connections. It cannot tell you what's true.
Finding the truth about your own health requires more than a prompt.
It requires study. Self-audit. The willingness to be wrong. The willingness to go against the consensus when the evidence and your own experience point somewhere else.
It requires getting away from the default setting: going with the flow, following the crowd, doing what's easy.
Those instincts are exactly what the Averaging Problem exploits.
The average is comfortable. The average is popular. The average is what everyone else is doing.
The average is also what got us here.
Work on your critical thinking. Build your own filter.
Learn to trust your own mind and body, not because they're infallible, but because they're the only filter you have that AI can't replicate.
Finding the truth is hard. But it's worth it.
If you want more on cutting through the noise and learning to trust your own mind and body, subscribe below. This is just the beginning.