Why the exam lane still belongs to a person
Will AI replace ophthalmic medical technologists? The honest reason behind the answer above is the shape of the work. Software can read a retinal photo. It cannot seat a nervous patient at the slit lamp, steady a child’s head for tonometry, or coax a clean visual field out of someone who keeps losing fixation.
Most of the day is physical and conversational. Measuring visual acuity, instilling dilating drops, capturing OCT and fundus images, taking an eye history, assisting the physician during minor office procedures, teaching a first-time wearer how to insert and clean contact lenses. Each step needs hands, judgment about whether the capture is usable, and a read on the patient in front of you.
Scale matters too. The Bureau of Labor Statistics counts about 182,610 people in this occupation, with median pay of $50,290 and projected employment growth of 5.9% between 2025 and 2035 (BLS, 2025 projections). Aging eyes keep clinics busy, and the imaging volume that AI helps read is the same volume someone has to capture first.
What software handles, what it assists, and what stays human
A share of the task time sits with people on this page: 82%. That group covers the work that needs contact and presence, such as positioning patients for imaging and assisting during in-office procedures. It also covers the small judgment calls, like deciding a field test has to be repeated because the patient drifted.
Software already does some narrow pieces on its own: 0%. Think automated refraction output, scheduling and recall notes, and screening algorithms that flag a diabetic retinopathy image for review. None of that ends the appointment; it changes what the appointment produces.
The assisted slice is 18%. Here a technologist keeps the task and the tool speeds it up: pre-reading OCT thickness maps before the physician sees them, drafting the history note from the intake conversation, catching an image artifact the system marks as low quality. Our overall figure for how much task time AI can handle today is 9, and the coverage method page explains how that is built.
How strong is the evidence?
Weaker than the headlines suggest for this job specifically. The evidence grade here is D, which means no study has tested AI against ophthalmic medical technologists doing their actual work end to end. Plenty of published work compares algorithms with clinicians on single image-reading tasks. That is a different job and a different question.
So we give no quality-parity number for this occupation. What would settle it is a trial in a working clinic: the same patient list, the same equipment, measuring usable capture rate, repeat-test rate, chair time and patient cooperation with and without automated support. Until something like that exists, the fair statement is that image interpretation has been measured and exam-lane practice has not. Our grades and what each letter requires are set out in the quality parity method, and the wider approach sits on the methodology page.
When the picture could shift
Most likely after 2042 (8 in 10 of our scenarios). For what that window does and does not claim, see how we model replacement year.
Two things could pull it earlier. Self-operated imaging booths that let a patient complete acuity, pressure and retinal photos without a hand on the chin rest. And payer or regulator acceptance of autonomous screening results, which would cut the number of in-person visits for stable monitoring cases.
Two things hold it back. The robotics tier this page lists is a dexterous humanoid, and no such machine is sold or deployed in eye clinics today. Second, the cost comparison shown above still favors a trained human for the physical half of the job, because the hard part is not the software license but the hands, the room and the liability for a bad measurement.
Good to know: automated screening tends to raise the number of patients who need a follow-up visit, which is work that lands back in the exam lane.
How to stay needed in this role
Lean into the parts that are hardest to hand over. First, difficult-patient testing: pediatric, low-vision and poorly cooperative cases where getting a usable result is a skill, not a setting. Second, procedure assistance, including instrument prep, sterile technique and anticipating what the surgeon needs next. Third, patient teaching, from contact lens hygiene to drop schedules that people otherwise abandon.
Two skills pay off. Deep fluency with your clinic’s imaging platforms, including knowing when an automated reading is unreliable and saying so. And documentation quality, since the note a physician can act on is still written by someone who was in the room.
If you are weighing nearby paths, the closest work sits with ophthalmic medical technicians, orthoptists and optometrists. You can see how this job sits against its peers on the health technologists and technicians family page, or across the wider healthcare sector.
The headline figure above is 83 out of 100 (higher is safer). To put that in context, compare two jobs side by side or browse the jobs that mostly need a person list.