There is a new kind of health nut, and this one does not juice. The datamaxxer wears the ring, wears the watch, straps on the sleep tracker, and then — here is the new part — pipes all of it into an AI chatbot. Heart rate, sleep stages, workout logs, the calendar, sometimes the text messages. Then they ask the machine the question people have whispered at ceilings for centuries: what, exactly, is wrong with me?
One devotee described the practice as her love language. We are not qualified to evaluate that. But we understand the impulse: for decades your body generated data and simply threw it away. Now it goes in a dashboard, and the dashboard talks back.
And sometimes it says something interesting. Our favorite case: an engineer fed his stress readings to the AI alongside his work calendar, and the machine noticed his body went haywire whenever meetings included one particular colleague. “Claude said this guy is the prime suspect,” he reported. Centuries of medicine, and it took a chatbot to formally diagnose a coworker.
Not a fringe hobby
The numbers say the datamaxxers are just the enthusiast wing of a mass movement. Roughly a third of American adults now use AI for health advice. According to the health-research group KFF, 19% have had it interpret actual lab results — one in five people handing their bloodwork to a chatbot. OpenAI says its models field more than 300 million health questions a week. The world’s most-consulted health adviser, by volume, is now software that has never once made you sit in a waiting room with a 2019 golf magazine.
You can see why. The AI is available at 3 a.m., which is when health anxiety keeps office hours. It never rushes you. It has read everything. And it is endlessly, pathologically agreeable — which is exactly the problem.
Doctors, for the record, have mixed feelings. The good ones will tell you a prepared patient is a better patient. The tired ones have started receiving printouts.
The two catches
Catch one: the machine is confidently wrong at unpredictable moments. Bad medical advice from an AI does not arrive with a warning tone. It arrives in the same fluent, confident voice as the good advice. A doctor who is wrong tends to be wrong inside guardrails — training, colleagues, liability. A chatbot that is wrong just keeps typing.
Catch two: privacy. The datamaxxer pipeline routes your sleep, your heart rhythms, your calendar and occasionally your arguments with your spouse through servers you do not control, under policies you have not read. We are not saying don’t. We are saying know what you are mailing, and to whom, before you call it a love language.
The rule
Here is where we admit the datamaxxers are onto something real. Spotting patterns across your own life — sleep versus stress versus that one recurring meeting — is genuinely something machines do well and a doctor, in a 15-minute visit, cannot. The engineer’s prime-suspect discovery is silly and also, quite possibly, correct.
So the rule is not abstinence. It is the same rule we use for portfolios, where clients now regularly arrive with AI-written analyses of their own accounts. We love it. The dashboard sharpens the questions, surfaces the patterns, and makes the meeting better. What it does not do is replace the licensed human who knows your history and carries accountability.
Bring the dashboard to the doctor, not instead of the doctor. Let the AI draft the questions; let the human make the call. The engineer, to his credit, seems to have grasped this. He consulted an actual physician about the stress — and merely started declining meetings with the prime suspect. Both, we would note, are valid interventions.