Optometrist

United States. The statistics bureau projects employment of optometrists to grow 10 percent from 2025 to 2035, from 45,100 to 49,400, much faster than the average for all occupations, with about 2,000 openings a year. The page does not mention artificial intelligence or automation. It counts jobs in one country.

A named person with standing publicly predicted something, on a date, in an attributable statement. It is recorded so that who said what, and when, stays checkable — and it never moves a task's assessment, because a prediction is not an observation. Its value arrives later: the record sits on the same page as the evidence about that occupation, so anyone reading the forecast reads the record of what happened next beside it. That is the reckoning; this site publishes no verdict on whether a forecast came true.

United Kingdom. The optical regulator says new formats of sight testing, notably teleoptometry, bring opportunities for improving access but also concerns about risks to patients. Its proposed updated statement reinforces the legal requirement for sight tests to include both eye health and vision checks and says a spectacles prescription cannot be issued until all statutory components of the sight test are complete. The consultation closes on 21 October 2026 and the statement is not yet adopted.

A rule that would have constrained or required automation was formally proposed and did not come into force — vetoed, voted down, struck down, or allowed to lapse. It establishes something real and checkable about the legal environment: a guardrail many people assume exists does not. It never moves a task's assessment, because it says nothing about what employers can do or are doing.

United Kingdom. The screening committee says it does not currently recommend incorporating automated grading in the diabetic eye screening programme. A 2021 review found the evidence insufficient; a new review concludes that some systems are as good as the first human grader and could be tested in real NHS services. The committee opened a consultation, closing on 3 August 2026, before making an updated recommendation. Pictures are currently read by trained graders, with a second and third grader where needed.

A rule that would have constrained or required automation was formally proposed and did not come into force — vetoed, voted down, struck down, or allowed to lapse. It establishes something real and checkable about the legal environment: a guardrail many people assume exists does not. It never moves a task's assessment, because it says nothing about what employers can do or are doing.

A descriptive study of 80 eyes from 40 patients comparing a portable self-refraction device with an autorefractor and with subjective cycloplegic refraction by an experienced examiner. The device was comparable to the autorefractor on spherical refraction but less precise on cylindrical power and axis, and showed inconsistencies in astigmatism correction. The authors call it promising for large-scale screening and conclude that subjective refraction should complement it. A small study of one device.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

United States. The Medicare rule says some new codes cover services where innovative technology is substituting for and/or augmenting physician work, citing code 92229 for point-of-care automated analysis of retinal images for diabetic retinopathy. After leaving its price to regional contractors for 2021, the agency finalised national payment values for the code. A payment rule funds the service; it does not require anyone to use it.

Regulation, subsidy or public procurement is requiring or funding adoption — the mirror of a constraint. It shows adoption is being required, not that it has happened, so one mandate is never enough on its own; two independent ones are.

Singapore. The health ministry says its national AI pilot processed 38,215 patient cases at six polyclinics and compared the results with grading by non-physician technician readers; on preliminary analysis its accuracy in detecting vision-threatening diabetic retinopathy, glaucoma suspects and late-stage macular degeneration was comparable to the current standard of care. A second phase from December 2020 extended the trial to six more polyclinics to study whether technician readers detect more with AI support, and whether automating grading improves efficiency for mild to moderate cases.

Small-scale trial in a real setting. Tells us the deployment conditions are being tested, not that they hold — so one pilot is never enough on its own; two independent ones are.

United States. The regulator granted a De Novo request for software that analyses retinal photographs and reports whether more-than-mild diabetic retinopathy is present. Its pivotal study of 900 participants showed sensitivity of 87.4% and specificity of 89.5%, and the summary says it demonstrated safe and effective performance when used to automatically, without physician assistance, detect the condition. Its labelling is for primary care clinics, not eye clinics; a positive result means referral to an eye-care professional.

A demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.

South Carolina, United States. The law says a person may not dispense glasses or contact lenses without a valid prescription from a provider, that the prescription must take into consideration the medical findings and refractive error from the eye examination, and that it may not be based solely on the refractive error of the eye or be generated by a kiosk. Violation is professional misconduct. It covers one US state.

Failure, rollback, regulation or cost is suppressing adoption. Can lower an assessment or widen its uncertainty.

If you are starting out, expect retinal photographs to be read first by software in many settings, and build what it does not do: examining the whole eye, managing conditions over time, fitting lenses, and explaining results to worried patients.

Expect more patients to arrive already screened by AI in primary care, and more of your work to be confirming, managing and referring. Writing the prescription and judging the whole eye stay with you where the law requires it.

Reshape the role

Move from screening to managing eye disease

Detection from retinal photographs is the part regulators and payers have opened to AI; the follow-up after a positive result still needs an eye-care professional.

Stay and strengthen

Stay in full eye examinations

Rules keep the prescription tied to a complete examination, and a self-test device fell short on astigmatism.

Adjacent move

Move into reviewing AI-flagged cases or running screening services

AI screening programmes need someone to check flagged images, handle ungradable ones and oversee the service.

Not on present evidence, though it is taking over part of the screening. AI may detect diabetic eye disease without a doctor in US primary care and Medicare pays for it, but the UK screening committee does not yet recommend AI grading, rules keep prescriptions tied to a full examination, and the US projects optometrist jobs to grow 10 percent from 2025 to 2035.

We do not answer that with a number of years. Watch whether screening programmes where you work adopt automated grading, and whether regulators allow prescriptions from remote or automated refraction. Those tell you more than any date.

For diabetic eye disease, yes, in approved settings: the US regulator authorised a system that detects it without physician assistance, and Singapore found AI accuracy comparable to its current grading. A positive result still sends the patient to an eye-care professional.

It depends on where you live. South Carolina bans prescriptions generated by a kiosk or based only on refractive error, the UK requires a complete sight test first, and a study found a self-test device less accurate on astigmatism than an examiner.