The company that taught AI to generate images now wants to scan your body in a golden pool — and the proposal reveals everything that goes wrong when science fiction becomes a consumer product.
There is something strangely beautiful about the proposal. You descend into a pool of golden light, underwater sensors emit ultrasound waves from every angle — like dolphins using echolocation, in the company's own description — and, without realizing it, your entire body is mapped with MRI-level resolution. According to Midjourney's announcement post, the scanner is part of a full spa, complete with saunas, hot tubs, and cold plunges, open 24 hours. The exam is a byproduct of relaxation. You go for the pleasure; the diagnosis is on the house. A cancer detected on a Tuesday afternoon? No drama — it was caught so early it's a non-issue. The vision is so seductive that the instinctive reaction is to ask for the Wi-Fi password and get in line.
But then the "too good to be true" bug bites. And it bites hard.
The detail no one is discussing is this company's origin. Midjourney isn't Medtronic, or GE Healthcare, or even a healthtech startup with regulatory compliance. It's the company that became famous for generating pretty images from text prompts. The leap — from "makes funny pictures" to "revolutionizes diagnostic medicine forever," as one LessWrong post summed it up — isn't a pivot; it's a blind jump across a bridgeless canyon. There's a difference between training a diffusion model to invent a dog samurai and building medical hardware that will inform life-or-death decisions. The proposal's aesthetics — golden lights, dolphins, spas — suggest Midjourney is thinking of the product as a sensory experience, not as a clinical device. This is equally enchanting and terrifying.
Here is the most uncomfortable point: the business model of the scanning spa is, in practice, a machine for producing the very problem it promises to solve. Diagnosing everything, always, early on, means also detecting everything that would never have harmed you. This is the paradox of overdiagnosis, well-documented in medical literature: routine MRIs on asymptomatic people find anomalies in somewhere between 15% and 30% of cases — the majority being false positives or lesions that would regress on their own. Each finding leads to biopsies, anxiety, and unnecessary interventions. A daily full-body scanner doesn't eliminate cancer; it redefines what counts as a disease. Suddenly, we are all patients all the time.
And then there's the data. Midjourney mentions almost as a footnote that you would have "a massive library of data about your health." Libraries aren't neutral. Who stores it? Who accesses it? Health insurers, employers, advertising platforms — all have incentives to want this data, and few of those incentives align with yours. A company whose original business is generating images from data now proposes collecting the most intimate dataset imaginable: the inside of your body, updated daily. The regulatory transparency this would require makes GDPR look like a newsletter signup form.
The lesson of recent years, as the LessWrong text observes, is that things that were science fiction, upon becoming real, tend to be far more controversial and far less cool than they seemed in the shows. Midjourney's golden spa is the most elegant version of this trap: it takes the hardest problem in medicine — the human and financial cost of late diagnosis — and turns it into a wellness experience. But medicine isn't a spa, and a spa isn't medicine. Confusing the two might be the most beautiful mistake technology has ever conceived.
The main risks include overdiagnosis and false positives. Routine scanning of asymptomatic people often detects harmless anomalies, leading to unnecessary anxiety, biopsies, and medical interventions. It also creates severe data privacy concerns regarding who accesses such intimate health information.
Midjourney is originally an AI image generation company without a background in medical devices or regulatory compliance. Transitioning from creating images to building clinical hardware for life-or-death diagnoses is a massive leap, especially since the product is marketed as a sensory spa experience rather than a clinical tool.
The paradox is that scanning everyone continuously to catch diseases early actually creates illness. It detects anomalies that would never have harmed the patient, turning healthy individuals into permanent patients and redefining normal bodily variations as diseases requiring treatment.