LO
Optometrist — tasks, one by one
The unit of analysis is the task, not the job title. Each one below carries its direction, whether the judgement rests on evidence or on platform inference, the reasoning, and what it does not establish.
Every task on this page#
Spotting eye disease in retinal images
Automating✓ Evidence-backedPhotographing and reading the back of the eye for signs of diabetic eye disease, glaucoma and macular degeneration.
This is where machines are already allowed to decide. The US regulator authorised a system that automatically, without physician assistance, detects more-than-mild diabetic eye disease in primary care clinics, and Medicare set a national payment for automated retinal analysis, noting that such technology is substituting for or augmenting physician work. Singapore's national pilot ran 38,215 cases through AI and found its accuracy comparable to the current standard. The UK is more cautious: its screening committee does not currently recommend automated grading, although a new review found some systems as good as the first human grader.
The US authorisation is for primary care, not eye clinics, and the Singapore pilot compared AI with technician graders; no record measures how much screening optometrists still do.
Measuring vision and prescribing lenses
Being augmented✓ Evidence-backedRefraction — working out the lens power a person needs — and writing the prescription for glasses or contact lenses.
Machines have measured refraction for years; the law and the evidence keep the prescription with the examiner. South Carolina says a prescription may not be based solely on the refractive error of the eye or be generated by a kiosk. The UK regulator's proposed statement says a spectacles prescription cannot be issued until every statutory part of the sight test is complete. A study of a self-test refraction device found it fell short on astigmatism and concluded that subjective refraction should complement rather than be displaced by it.
One US state's law, a proposed UK statement and one small study; other places may allow online refraction.
Managing and monitoring eye conditions
Being augmented≈ Platform inferenceFollowing patients with glaucoma, dry eye or other conditions over time, treating where allowed, and deciding when to refer.
The screening tools feed this work rather than doing it: a positive AI result still sends the patient to an eye-care professional. The US statistics bureau projects optometrist employment to grow 10 percent from 2025 to 2035, much faster than average, without mentioning AI.
No record measures AI in the management of eye conditions; the projection covers one country.
Fitting contact lenses
Still human-led≈ Platform inferenceMeasuring the eye, choosing and fitting lenses, and checking how they sit and how the eye tolerates them.
No record found shows machines doing this. It is hands-on work at the slit lamp, and the same prescription rules that apply to glasses apply to contact lenses.
The absence of a record is not evidence that nothing is changing; this judgement rests on the nature of the work.
Remote sight tests
Being augmented≈ Platform inferenceDoing parts of the eye examination at a distance, with an optometrist in one place and the patient in another.
New formats are arriving and regulators are writing rules for them. The UK regulator says teleoptometry brings opportunities for access but also concerns about risks to patients, and is consulting on a statement that keeps the full sight test before any prescription.
A consultation that has not been adopted; it does not say how widely remote tests are used.