On 5 October 2026, Utah let an AI start issuing prescriptions — topical acne treatments, adults only, under a one-year state pilot. Social media called it the moment medicine changed hands.
It is a genuine milestone. It is not the first one, and for anyone building immersive technology, it is not the most interesting.
The line was crossed in 2021
Five years before a state let an algorithm write a prescription, the FDA authorised a virtual reality product as a medical treatment. AppliedVR received De Novo authorisation in November 2021 for a VR therapeutic for chronic lower back pain — the first medical extended-reality product cleared in the United States. Its pivotal trial was randomised, double-blind and sham-controlled across 179 patients: 66% of the VR group reported pain reduction greater than 30%, against 41% of controls. A further RelieVRx clearance followed in August 2025, and a 2025 review in Frontiers in Virtual Reality traces the path from development through authorisation into clinical use across more than thirty publications.
Read that sequence again. Immersive technology reached regulated, prescribable clinical use before AI did.
The evidence is, in places, better
AI's medical footprint is enormous but shallow in exactly the place that matters. The FDA's list of AI-enabled devices passed 1,600 authorisations by September 2026 and more than 80% of physicians now use AI at work — but only about 17% use it for assistive diagnosis. The rest is documentation, summarisation and detection.
Immersive clinical evidence is narrower and, within its narrow band, often more direct. A randomised trial of virtual reality training for total hip arthroplasty found VR-trained participants completed 33% more key procedural steps, placed components 12 degrees more accurately, and worked 18% faster than controls. These are performance measures on cadaveric assessment rather than patient outcomes, and the trials are small — but they measure a person getting better at something.
That distinction is becoming the whole argument.
The deskilling problem is an immersive opportunity
The most important medical AI finding of the past year was not a capability. It was a cost. At four endoscopy centres, researchers measured what happened to clinicians' unassisted performance after AI detection arrived: detection rates fell from 28% to 22% in three months, among experienced clinicians with thousands of procedures each.
Set the two technologies side by side and the shapes are opposite. AI, deployed as a crutch, can quietly erode the skill underneath it. Immersive simulation exists to build that skill through repetition that costs nothing and harms no one. One substitutes for practice; the other is practice.
The most defensible position in clinical technology may turn out to be the combination — AI absorbing the clerical load, immersive environments maintaining the human competence that automation quietly spends.
Why this matters beyond medicine
Health care is the most heavily regulated environment immersive technology has ever entered, and it is already through the door with authorisations, randomised trials and reimbursement conversations.
For anyone building or owning immersive property, that is the signal worth reading. A walkable, interactive space is not a novelty waiting for permission — it is infrastructure that has already cleared the hardest bar there is. Training floors, therapy rooms and clinical simulation are commercial spaces with regulated demand behind them, and they are built the same way a showroom or gallery is built.
The studio behind this site, Smith App Studio, builds immersive environments and clinical software — their companion piece, Is AI Taking Over Medicine?, reads the same evidence from the practitioner's side.
Medicine is not being handed to machines. It is being divided between them — and the half that requires a human to be good at something is the half immersive technology serves.
Educational content, not medical, investment or legal advice.
