Ophthalmic Medical Technologists

29-2099.05
Median wage $50,290/yr182,610 employed (US)Rank #604 of 923 scored · top 65% by substitution

Assist ophthalmologists by performing ophthalmic clinical functions and ophthalmic photography. Provide instruction and supervision to other ophthalmic personnel. Assist with minor surgical procedures, applying aseptic techniques and preparing instruments. May perform eye exams, administer eye medications, and instruct patients in care and use of corrective lenses.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution23
Exposure22
Augmentation51

Substitution — the headline: capability discounted by cost, barriers and adoption.

Exposure — technical capability alone, regardless of whether anyone deploys it.

Augmentation — how much AI assists without replacing. High here + moderate substitution = a changing job, not a disappearing one.

Tasks on the substitution scale

31 rated tasks, binned by substitution score.

Position among all scored occupations

Distribution of 923 occupation scores; the marker is this occupation.

Tasks with substitution ≥ 70

3%

Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.

Why this score

The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.

Task automatabilityw 35%23

panel mean rating 1.9/5 → substitution pressure 23/100

Technical feasibility todayw 20%21

panel mean rating 1.8/5 → substitution pressure 21/100

Cost vs. human wagew 15%21

panel mean rating 1.8/5 → substitution pressure 21/100

Adoption barriersw 20%inverted — strong barriers lower the score27

panel mean rating 3.9/5 (barrier strength) → substitution pressure 27/100

Sector adoption velocityw 10%22

panel mean rating 1.9/5 → substitution pressure 22/100

Task breakdown (31 tasks)

Substitution pressure per task, weighted by O*NET importance in the composite. Expand a task for the full rater audit trail — every rating, every model, every rationale.

Calculate corrections for refractive errors.

71

CI 67–74 · exposure 75 · augmentation 88 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Ophthalmic practices are relatively high-digitization sectors; automated refraction and calculation tools are increasingly integrated into clinical workflows. Adoption is well above average for healthcare specialty roles, particularly in larger practices and hospital systems.
Sector adoption velocityclaude-sonnet-53/5Digital autorefractors and calculation tools are common in ophthalmology practices, but full end-to-end automation of the diagnostic workflow remains a mix of manual and automated steps, so adoption is moderate rather than fast/deep.
Augmentation potentialclaude-haiku-4-5-202510015/5AI-powered refraction and calculation tools substantially augment technician productivity by automating repetitive measurements, instantly generating correction estimates, and flagging anomalies, allowing technicians to focus on patient interaction and clinical troubleshooting.
Augmentation potentialclaude-sonnet-54/5AI-assisted calculation tools speed up and improve accuracy of refractive error corrections, letting technologists focus on patient interaction and verification rather than manual math.
Task automatabilityclaude-haiku-4-5-202510014/5AI can perform most aspects of refractive error calculations end-to-end by analyzing eye measurements (refraction data, keratometry, axial length) and computing corrective lens prescriptions, achieving significant time savings. However, some edge cases and complex scenarios may require human ophthalmologist review, preventing a full 5 rating.
Task automatabilityclaude-sonnet-54/5Refraction calculations (e.g., vertex distance adjustments, cylinder conversions, prism calculations) are formulaic and well within reach of software/calculators, especially when integrated with autorefractors and EHR systems.
Adoption barriersclaude-haiku-4-5-202510013/5While calculations themselves can be fully automated, regulatory frameworks (FDA clearance for automated refraction devices) and professional norms require ophthalmologist sign-off on prescriptions and final verification. Organizational practice and liability considerations create friction but not absolute legal barriers to automation.
Adoption barriersclaude-sonnet-53/5While the calculation itself isn't heavily regulated, the broader refractive assessment and prescription typically require oversight by a licensed optometrist/ophthalmologist, creating moderate procedural friction.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated refraction and calculation systems operate at near-zero marginal cost per patient once deployed, whereas ophthalmic technicians command $40,000–$55,000+ annual loaded wages. The cost ratio heavily favors AI-driven automation.
Cost vs. human wageclaude-sonnet-54/5Automated refraction calculation software is inexpensive relative to technician time once integrated into existing diagnostic equipment, giving a substantial cost advantage.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed ophthalmic software systems (e.g., automated refraction analyzers, optical biometry systems) reliably calculate corrections in routine clinical practice today. These products are used in production across clinics and hospitals, though integration with human review workflows remains standard.
Technical feasibility todayclaude-sonnet-54/5Autorefractors, phoropters with digital output, and EHR-integrated calculation modules already perform these computations reliably in many clinics, though final sign-off and refinement still involves technician/clinician judgment.

Measure and record lens power, using lensometers.

49

CI 30–67 · exposure 50 · augmentation 63 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of full automation in ophthalmic clinics has been slow. While lensometry is mechanically automatable, eye care remains a human-centric, regulated sector with relatively low digitization compared to finance or tech. Pilots exist, but production displacement is minimal and concentrated in only advanced optometric practices.
Sector adoption velocityclaude-sonnet-53/5Automated lensometers are common in optometry and ophthalmology practices, representing decent penetration of automation technology, though full AI-driven diagnostic workflows are still emerging in this sector.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-assisted lensometry—where automated systems measure and present results to the technician for review—can substantially raise productivity by reducing measurement time and flagging anomalies for human verification. This augmentative use case is already deployed in many modern optical labs and electronic health records integrations.
Augmentation potentialclaude-sonnet-53/5Digital/automated lensometers assist technicians by speeding up and improving accuracy of readings, but the task still requires physical setup and human interpretation in context.
Task automatabilityclaude-haiku-4-5-202510014/5Lensometry is a mechanical/optical measurement task with a clear input (lens) and output (numeric power values). Modern computer vision and automated optical systems can measure lens power reliably, though integration with existing workflows and occasional manual verification for edge cases (high prescriptions, coatings, decentration) limits full end-to-end automation, achieving ~60–70% time savings.
Task automatabilityclaude-sonnet-52/5Measuring lens power with a lensometer is a hands-on optical measurement requiring physical manipulation of the instrument and lens, which current general AI cannot perform end-to-end; automated lensometers exist but are a hardware solution, not an AI/software substitution of the task.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: eye care is regulated (prescription verification often requires licensed personnel oversight), liability and error costs are asymmetric (incorrect prescription measurements can harm patients), and many eye care organizations prefer human technicians for quality assurance and patient interaction. State optometry boards and medical directors typically require or strongly enforce human sign-off.
Adoption barriersclaude-sonnet-52/5There's no strict licensing requirement mandating a human perform this exact measurement, but it occurs within a clinical workflow where technicians are supervised by licensed practitioners, creating some organizational friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated optical measurement systems have high upfront capital cost but very low per-measurement inference cost once deployed. Compared to a technician's loaded wage (~$30–40/hour), the per-lens cost of an automated system amortizes quickly, yielding roughly 5–10× savings at scale.
Cost vs. human wageclaude-sonnet-52/5Automated lensometers reduce time per reading but require capital investment in specialized equipment and the technician is still needed for handling and calibration, so overall cost savings versus a trained technician are moderate at best.
Technical feasibility todayclaude-haiku-4-5-202510013/5Automated lensometers and spectacle analysis systems exist in research and limited production (e.g., advanced optical inspection systems), but widespread deployment in eye care clinics remains incomplete. Products perform well on standard lenses but have material error rates on complex prescriptions or non-standard frames, so reliability is partial rather than mature.
Technical feasibility todayclaude-sonnet-53/5Automated/digital lensometers already exist as deployed products that read lens power with a button press, though these are specialized hardware devices rather than generalizable AI systems, and technician oversight is still standard practice.

Call patients to inquire about their post-operative status or recovery.

47

CI 37–56 · exposure 38 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains conservative in automating patient contact, particularly post-operative follow-up where clinical accountability is high. Most ophthalmology practices still rely on human technicians for these calls.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially small ophthalmology practices, is a slower-adopting sector for conversational AI outreach compared to finance or tech, though patient engagement tools are growing.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by pre-populating call scripts, flagging high-risk patients, or summarizing chart data before a technician calls, improving efficiency without removing human judgment on symptom assessment and escalation.
Augmentation potentialclaude-sonnet-54/5AI-assisted call scripts, automated triage, and pre-call symptom screening can meaningfully speed up and standardize the technologist's outreach workflow while they remain available for complex cases.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate outreach calls and scripts, the task requires nuanced empathy, clinical judgment about concerning symptoms, and adaptive conversation flow based on patient responses. Current systems struggle with handling unexpected patient concerns or medical red flags that demand human escalation.
Task automatabilityclaude-sonnet-53/5Automated voice-call or chatbot systems can conduct structured post-op check-in calls and flag concerning responses, but nuanced clinical follow-up questioning and reassurance still benefit from human judgment.'
Adoption barriersclaude-haiku-4-5-202510013/5Healthcare privacy (HIPAA) and contact regulations apply, but no explicit licensure bars automation of post-op status checks. Patient preference and clinical liability concerns (missed complications) create moderate friction.
Adoption barriersclaude-sonnet-52/5No strict licensure requirement for this specific task, but patient safety concerns around missing surgical complications create moderate liability and clinical oversight friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-driven calling via automated services costs substantially less than human technician time ($0.50–$2 per call vs. $20–$40 loaded wage for a technician call), though oversight and quality control add overhead.
Cost vs. human wageclaude-sonnet-54/5Automated calling/chatbot systems cost far less per call than a technologist's time, especially for routine standardized check-ins, though escalation handling adds some cost back.
Technical feasibility todayclaude-haiku-4-5-202510012/5Automated call systems exist but have high failure rates in healthcare contexts due to poor symptom assessment, patient dissatisfaction, and inability to reliably escalate medical concerns. Deployed patient outreach remains largely human-driven in ophthalmic settings.
Technical feasibility todayclaude-sonnet-53/5Healthcare voice-AI and IVR-based patient outreach products exist and are used for post-op check-ins, but adoption in ophthalmology-specific clinics is limited and often supplements rather than replaces staff calls.

Measure corneal thickness, using pachymeter or contact ultrasound methods.

39

CI 16–62 · exposure 41 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Ophthalmology practices and imaging centers are moderately digitized, and some advanced clinics use automated measurement systems, but widespread deep adoption of fully autonomous pachymetry is still emerging. Most practices still employ technologists for device operation and positioning, indicating middling adoption rather than fast, deep transformation.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic imaging adoption of full automation for hands-on procedures remains slow; AI is used more for downstream image interpretation than for performing physical measurements.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assists technologists by providing real-time measurement feedback, flagging image quality issues, and automating calculation, which substantially improves throughput and reduces measurement error. The technologist remains in the loop for positioning and quality control, making this a strong example of productivity augmentation.
Augmentation potentialclaude-sonnet-53/5AI-enhanced pachymeters and ultrasound devices can improve measurement precision, calibration, and data logging, aiding the technician's workflow without replacing them.
Task automatabilityclaude-haiku-4-5-202510014/5The physical measurement via pachymeter or ultrasound requires device operation and positioning, but image acquisition and thickness calculation are largely automatable. AI can analyze pachymetry images and ultrasound data with high accuracy, though the initial device positioning and patient cooperation still require human involvement, preventing full 50% time-saving automation end-to-end.
Task automatabilityclaude-sonnet-52/5This is a hands-on diagnostic procedure requiring physical instrument contact/positioning on a patient's eye; AI cannot physically operate a pachymeter or perform ultrasound contact today.the interpretive part is small compared to the physical measurement task.
Adoption barriersclaude-haiku-4-5-202510013/5Clinical measurement results often require human verification and sign-off by the ophthalmologist; no legal mandate exists for human technologist signature on the raw measurement itself, but quality assurance and liability practices create friction. Regulatory bodies (FDA, clinical labs) have oversight of measurement accuracy, creating moderate friction to full automation.
Adoption barriersclaude-sonnet-54/5Direct patient contact with medical devices typically requires a trained/certified technician, and clinical liability for diagnostic accuracy adds strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI-based image analysis and measurement interpretation is substantially cheaper than human technologist labor once integrated, with inference costs near zero and minimal oversight needed. The primary remaining cost is the device hardware and initial setup, making AI clearly cheaper for high-volume measurement scenarios.
Cost vs. human wageclaude-sonnet-51/5AI cannot replace the physical measurement process, so there is no viable AI-only cost comparison; a technician is still required to perform the procedure.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed AI systems exist for corneal thickness measurement from pachymetry and ultrasound images with good accuracy in controlled settings, but real-world deployment varies with device integration and requires validation for clinical use. Products are available but often require manual alignment and verification rather than fully autonomous operation.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously performs corneal pachymetry measurement without a human operator; current devices are operator-driven tools, not autonomous systems.

Perform ophthalmic triage, in the office or by phone, to assess severity of patients' conditions.

37

CI 11–62 · exposure 33 · augmentation 63 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare telehealth and digital triage adoption accelerated post-2020, but ophthalmic practices remain mixed: large vision centers and telehealth platforms are piloting AI triage, while smaller practices lag; production deployment is growing but not yet standard of care.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially direct patient-facing clinical triage, has been slow to adopt autonomous AI decision-making due to safety and liability concerns, despite some telehealth symptom-checker pilots.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can present symptom severity summaries, flag red flags, and suggest urgency categories to technologists in real time, substantially speeding their review and reducing cognitive load, while the human remains accountable for final triage decision and complex judgment.
Augmentation potentialclaude-sonnet-53/5AI can help by suggesting possible conditions, prioritizing urgency flags from reported symptoms, or pulling relevant patient history quickly, supporting but not replacing the technologist's judgment.
Task automatabilityclaude-haiku-4-5-202510014/5AI can reliably extract key patient history, symptom severity indicators, and vision-threat red flags from patient reports (via phone or questionnaire), applying evidence-based triage protocols to recommend urgency levels with high accuracy and significant time savings. However, real-time clinical judgment around atypical presentations and direct physical assessment still require human oversight.
Task automatabilityclaude-sonnet-51/5Triage requires synthesizing patient-reported symptoms, visual acuity, history, and clinical judgment to assess urgency and risk of vision loss, which current AI cannot reliably do end-to-end without human oversight.
Adoption barriersclaude-haiku-4-5-202510013/5Triage decisions affect patient safety and liability; regulatory bodies (FDA, state medical boards) increasingly scrutinize AI-assisted clinical workflows, and liability asymmetry favors human sign-off. However, no hard legal mandate requires a licensed tech to perform initial triage in all jurisdictions, creating moderate friction rather than a hard block.
Adoption barriersclaude-sonnet-54/5Triage decisions carry significant liability for missed emergencies (e.g., retinal detachment) and are typically performed by trained/licensed technologists under physician supervision, creating strong regulatory and liability barriers.
Cost vs. human wageclaude-haiku-4-5-202510014/5Once integrated into an EHR or telehealth workflow, AI triage inference costs are negligible (cents per encounter), while ophthalmic technologist labor costs $25–40/hour; the cost advantage is substantial even accounting for oversight and integration.
Cost vs. human wageclaude-sonnet-52/5Even if AI could assist with initial screening, the need for licensed clinical staff to verify and take responsibility for triage decisions keeps overall costs comparable to human-only workflows.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI-powered symptom checkers and triage decision support exist in production healthcare settings (e.g., NHS 111, various telehealth platforms), but ophthalmic-specific deployed systems are less mature and often require human review; error rates on subtle conditions remain meaningful.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs autonomous ophthalmic triage in clinical settings today; symptom-checker chatbots exist but are not trusted or validated for this specific high-stakes decision.

Take and document patients' medical histories.

34

CI 25–43 · exposure 33 · augmentation 63 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmic practices remain fragmented and slower to digitize than primary care or finance; most still rely on paper or legacy EHRs. AI-driven intake automation is in early pilots rather than established production use across the sector.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI documentation tools slowly due to compliance, EHR fragmentation, and clinical workflow inertia, with ophthalmic specialty clinics lagging further behind fast-adopting sectors.
Augmentation potentialclaude-haiku-4-5-202510013/5Voice-to-text, auto-population of fields, and symptom-suggestion tools can assist technologists in faster documentation and completeness, raising throughput. However, the gains are incremental rather than transformative, since human judgment and questioning remain essential.
Augmentation potentialclaude-sonnet-54/5AI-assisted intake forms, transcription, and summarization tools meaningfully speed up history documentation while the technician remains responsible for accuracy and follow-up.
Task automatabilityclaude-haiku-4-5-202510012/5Collecting structured patient histories (name, age, symptoms) can be partially automated via forms or voice intake, but patients often provide complex, non-linear narratives that require human clarification and follow-up questioning. AI could capture basic demographic and chief-complaint data but not reliably extract nuanced clinical context without significant human oversight.
Task automatabilityclaude-sonnet-53/5AI can transcribe, structure, and summarize patient-reported history via conversational agents or ambient documentation tools, but eliciting relevant clinical detail and probing follow-up questions still often needs human judgment.asked, so partial automation is realistic today.
Adoption barriersclaude-haiku-4-5-202510014/5Medical documentation is regulated by HIPAA and state healthcare licensing boards; medical technologists must be certified/licensed and accountable for record accuracy. Liability for errors in medical history (missed diagnoses, wrong dosages) creates strong legal and organizational friction against full automation.
Adoption barriersclaude-sonnet-53/5No strict licensure requires a human to take histories, but clinical liability, EHR integration, and patient trust create meaningful friction against full delegation to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5Initial AI implementation (voice transcription, form automation) carries high integration and compliance overhead relative to the wage of a technologist, especially when human review and correction are factored in. Cost parity is approached only at scale with very simple standardized intake.
Cost vs. human wageclaude-sonnet-53/5AI scribe/intake tools have real licensing and integration costs plus required human review, so savings versus a technician's wage are moderate rather than order-of-magnitude cheaper.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbot-based history systems exist in limited pilot settings, but deployed products struggle with accuracy on complex symptom descriptions and miss clinical red flags. No mature production system reliably replaces human technologists at the thoroughness required for medical documentation standards.
Technical feasibility todayclaude-sonnet-52/5Ambient scribes and intake chatbots exist in some clinics, but ophthalmology-specific history-taking with structured clinical detail is not yet a widely deployed, reliable standard product.

Measure the thickness of the retinal nerve, using scanning laser polarimetry techniques to aid in diagnosis of glaucoma.

30

CI 28–32 · exposure 30 · augmentation 75 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Ophthalmology and optometry practices use advanced imaging widely, and many have adopted automated measurement tools as assistive aids, but adoption of unsupervised automated measurement for diagnostic purposes remains cautious and varies by practice size and risk tolerance.
Sector adoption velocityclaude-sonnet-53/5Ophthalmology has adopted advanced imaging and semi-automated diagnostic software steadily, but this specific polarimetry task remains a moderately adopted, equipment-dependent niche rather than a fast-diffusing AI use case.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-assisted measurement of retinal nerve fiber layer thickness significantly accelerates the technologist's workflow by providing instant quantitative guidance, trend tracking, and quality flags, allowing the human to focus on scan quality and clinical context rather than manual measurement.
Augmentation potentialclaude-sonnet-54/5AI-enhanced software within the device already assists by generating quantitative retinal nerve fiber layer maps and normative comparisons, significantly aiding the technologist's diagnostic contribution and speeding workflow.
Task automatabilityclaude-haiku-4-5-202510012/5While scanning laser polarimetry generates objective quantitative data that AI can process, the task requires real-time instrument calibration, patient positioning, and quality-control judgment that current AI systems struggle to automate reliably end-to-end. Automated measurement of thickness from valid scans is feasible, but ensuring valid image acquisition and interpreting borderline cases still require human oversight.
Task automatabilityclaude-sonnet-52/5The scan is performed via specialized ophthalmic equipment (GDx/scanning laser polarimeter) that requires hands-on patient positioning, calibration, and technical skill; AI can assist in image interpretation but not the physical measurement acquisition itself.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical measurement of retinal structures in glaucoma diagnosis is regulated by medical device standards and often requires sign-off by a licensed ophthalmologist or certified technologist for clinical validity and liability. The diagnostic context and potential for patient harm if measurements are incorrect create strong barriers to full automation.
Adoption barriersclaude-sonnet-54/5This is a clinical diagnostic procedure typically requiring a credentialed ophthalmic technician/technologist, with liability and accuracy concerns limiting full delegation to non-supervised automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5The cost of integrating automated measurement software into existing ophthalmic workflows, plus ongoing calibration and oversight, remains substantial relative to a technologist's labor for routine scans. The labor cost is moderate, and AI does not achieve order-of-magnitude savings given the need for human quality assurance.
Cost vs. human wageclaude-sonnet-52/5The device itself is a large capital investment plus a skilled technician is still required for patient interaction, positioning, and quality control, so cost savings versus a human operator are modest despite automated internal analysis.
Technical feasibility todayclaude-haiku-4-5-202510013/5Commercial ophthalmology software and some research systems can measure retinal nerve fiber layer thickness from polarimetry scans, but these operate within narrow, controlled conditions and require human technologists to validate image quality and flag artifacts. Deployed products assist but do not fully replace the technologist's judgment.
Technical feasibility todayclaude-sonnet-52/5Automated instruments already capture and quantify RNFL thickness with built-in software, but a trained technologist must operate the device, ensure proper alignment/fixation, and validate image quality—no autonomous product performs the full task independently in production.

Educate patients on ophthalmic medical procedures, conditions of the eye, and appropriate use of medications.

29

CI 25–34 · exposure 25 · augmentation 75 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI for patient-facing education remains slow and conservative; most eye care practices use AI minimally for this function, relying instead on printed materials and human technologists, reflecting sector-wide caution around medical liability and patient trust.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially outpatient clinical settings like ophthalmology, has been slower to adopt AI-driven patient communication tools compared to purely digital/information sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially assist technologists by drafting personalized educational summaries, generating condition-specific diagrams, and providing instant fact-checking on medications, enabling technologists to spend more time on patient interaction and complex case discussion rather than rote explanation.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist technologists by drafting patient-friendly explanations, translating medical jargon, and providing reference materials to support in-person education, improving efficiency and consistency.
Task automatabilityclaude-haiku-4-5-202510012/5AI can generate educational materials and provide standardized information about eye conditions and medications, but patient education requires individualized explanation, empathy, and real-time Q&A clarification that demands human presence and adaptability. Current systems lack the contextual sensitivity for medical patient communication at scale.
Task automatabilityclaude-sonnet-52/5AI can generate patient education content and answer general questions, but this task requires tailored, in-person explanation, assessing patient understanding, and addressing individual concerns tied to clinical findings, which current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Patient education is often legally tethered to licensed clinical staff (technologists, optometrists, or ophthalmologists) in many jurisdictions; informed consent and liability for medication counseling create regulatory and organizational friction that prevents full substitution.
Adoption barriersclaude-sonnet-53/5No strict licensure barrier prevents AI from providing general information, but clinical accuracy concerns, liability for medication guidance errors, and patient preference for human reassurance create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5While AI-generated content has low marginal cost, implementation requires clinical oversight, regular updates for accuracy, and integration into clinical workflows; total cost approaches parity with technologist wages when accounting for setup, validation, and human supervision.
Cost vs. human wageclaude-sonnet-53/5AI-generated educational materials are cheap to produce, but the interpersonal counseling component still requires paid staff time, keeping overall cost roughly comparable when factoring in oversight and patient interaction.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbots and educational videos exist but are narrow in scope and typically used as supplementary tools rather than primary educators; no deployed product reliably replaces a trained technologist in patient education across diverse conditions, literacy levels, and anxiety states in clinical practice.
Technical feasibility todayclaude-sonnet-52/5Chatbots and AI-generated patient handouts exist and are used for general health education, but no deployed product reliably conducts individualized in-person ophthalmic patient counseling as part of clinical workflow.

Take anatomical or functional ocular measurements, such as axial length measurements, of the eye or surrounding tissue.

29

CI 25–32 · exposure 30 · augmentation 63 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption remains slow in ophthalmology practices; while digital biometry systems are common, automation of the technician role itself is rare. The field remains relatively traditional with strong emphasis on licensed technician presence, and integration of AI agents into clinical workflows is still in pilot phases.
Sector adoption velocityclaude-sonnet-53/5Ophthalmology has adopted automated diagnostic imaging and biometry devices steadily, but full replacement of technologist involvement remains uncommon in current clinical workflows.
Augmentation potentialclaude-haiku-4-5-202510013/5AI offers useful assistance in post-acquisition measurement validation, outlier detection, and recommending repeat measurements when quality is suboptimal. AI can flag measurement inconsistencies and accelerate interpretation, but the core measurement act remains technician-dependent, limiting transformative potential.
Augmentation potentialclaude-sonnet-54/5Modern automated biometers and imaging systems substantially speed up and standardize measurement collection, letting technologists focus on quality control and patient care rather than manual measurement.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI cannot perform end-to-end axial length or anatomical ocular measurements with 50% time savings at equal quality. While AI can assist in image analysis and measurement interpretation, the initial acquisition and validation of precise measurements requires human technician operation of specialized equipment (A-scan, optical biometry devices) and hands-on patient positioning.
Task automatabilityclaude-sonnet-52/5Ocular measurement devices are automated instruments, but the task of positioning patients, operating equipment correctly, and ensuring measurement validity still requires trained human hands-on execution and clinical judgment.
Adoption barriersclaude-haiku-4-5-202510014/5Significant barriers exist: measurements must be taken by licensed ophthalmic technologists or under direct optometrist/ophthalmologist supervision in most jurisdictions, creating a hard licensing requirement. Clinical liability for measurement errors is asymmetric—incorrect axial length leads directly to wrong intraocular lens calculations and surgical outcomes.
Adoption barriersclaude-sonnet-54/5Direct physical contact with patients, medical device operation, and clinical accuracy requirements mean this task is typically restricted to credentialed technologists under regulatory and liability frameworks.
Cost vs. human wageclaude-haiku-4-5-202510012/5The AI cost per measurement (software licensing, infrastructure, oversight) plus the necessary technician labor for equipment operation and validation remains comparable to or exceeds the cost of a skilled technician performing the task directly, since human involvement cannot be eliminated.
Cost vs. human wageclaude-sonnet-52/5The equipment itself is costly capital investment, and a trained human operator is still needed for patient interaction, calibration checks, and troubleshooting, so cost savings versus a technologist are limited.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed products reliably perform the full measurement task autonomously in production. Some AI tools exist for post-hoc analysis of biometry images or OCT scans, but acquiring accurate measurements remains dependent on trained technicians operating equipment and validating patient-specific anatomical factors.
Technical feasibility todayclaude-sonnet-53/5Biometry devices (e.g., IOLMaster, optical coherence tomography) already automate much of the measurement capture, but a technologist is still required to operate the device, verify quality, and reposition the patient as needed.

Create three-dimensional images of the eye, using computed tomography (CT).

26

CI 21–30 · exposure 30 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare imaging adoption of AI is moderate and cautious, with most deployment focused on radiologist assistance rather than technologist displacement. Clinical imaging sectors move slowly due to validation, regulatory, and liability concerns.
Sector adoption velocityclaude-sonnet-52/5Healthcare imaging is a moderately regulated, physically-oriented sector where AI adoption for full imaging task automation is slow, though AI-assisted reconstruction tools are gradually being piloted.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automating image reconstruction, reducing noise, segmenting anatomical structures, and highlighting artifacts, which helps the technologist verify and improve image quality. However, the human technologist remains central to patient interaction and imaging protocol execution.
Augmentation potentialclaude-sonnet-53/5AI-enhanced reconstruction algorithms and image processing tools can improve image quality and speed for the technologist, providing moderate productivity benefits during the imaging process.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with CT image reconstruction and analysis, the full task of creating 3D eye images requires positioning the patient, operating specialized equipment, and ensuring proper image quality—steps that demand physical presence and real-time judgment. AI can accelerate post-processing but cannot perform the imaging acquisition itself.
Task automatabilityclaude-sonnet-52/5Operating CT imaging equipment and positioning patients requires hands-on physical work and clinical judgment that current AI cannot perform end-to-end; AI can assist with image reconstruction but not the full acquisition task.rationale
Adoption barriersclaude-haiku-4-5-202510014/5Medical imaging performed on patients involves regulatory oversight (FDA), licensing requirements for technologists, liability for image quality, and legal requirements that a qualified technologist must acquire and validate images. These hard barriers protect human employment in clinical settings.
Adoption barriersclaude-sonnet-54/5Operating imaging equipment on patients typically requires licensure/certification and involves direct physical patient contact and radiation safety protocols, creating strong regulatory and liability barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5CT equipment operation, patient positioning, and quality assurance remain expensive to perform, and AI inference for reconstruction adds cost to the clinical workflow rather than replacing the technologist's labor significantly. The all-in cost favors the human technologist over AI substitution.
Cost vs. human wageclaude-sonnet-51/5CT equipment and skilled technologist labor remain necessary; AI reconstruction software is a cost addition to existing workflows rather than a cheaper substitute for the technologist's role.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI-based CT reconstruction algorithms exist in research and some commercial products, but reliable end-to-end 3D eye imaging generation requires validated clinical-grade systems that remain limited in scope. Current deployments focus on image enhancement rather than independent image creation.
Technical feasibility todayclaude-sonnet-52/5While AI-based image reconstruction algorithms exist in some CT software, no deployed product autonomously performs the full patient positioning, scanning, and image generation task without a trained technologist.

Measure corneal curvature with keratometers or ophthalmometers to aid in the diagnosis of conditions, such as astigmatism.

26

CI 21–30 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmology practices and optical clinics have shown slow adoption of fully automated measurement systems; most workflows still rely on technician-operated instruments. Digital transformation in this sector lags information-heavy professions due to the hands-on, patient-contact nature of the work.
Sector adoption velocityclaude-sonnet-52/5Healthcare/ophthalmology is a moderately digitized but conservative sector where diagnostic equipment adoption is steady but slow, with automation limited to device-assisted measurement rather than full AI substitution.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by automatically flagging measurement anomalies, suggesting repeat measurements, or pre-analyzing corneal shape patterns to guide technologist attention, but the core measurement task remains technologist-dependent.
Augmentation potentialclaude-sonnet-54/5Automated keratometers and computer-assisted corneal topography significantly speed up and standardize measurement, letting technologists focus on patient positioning and data interpretation.
Task automatabilityclaude-haiku-4-5-202510012/5Keratometry involves precise physical measurement of corneal curvature using specialized optical instruments. While AI could analyze the resulting measurements or images, the task of operating the device and obtaining accurate readings remains dependent on skilled manual positioning and instrument calibration that current AI cannot independently perform end-to-end.
Task automatabilityclaude-sonnet-52/5Automated keratometers already capture readings, but positioning the patient, ensuring proper alignment, and integrating results into clinical judgment still require a trained technologist; only partial time savings are achievable today.
Adoption barriersclaude-haiku-4-5-202510014/5Keratometry is performed as part of licensed eye care workflows; results must be verified by optometrists or ophthalmologists before clinical use. The requirement for trained human technicians to position patients, adjust instruments, and ensure measurement quality creates organizational and regulatory friction against full automation.
Adoption barriersclaude-sonnet-53/5Performing diagnostic eye measurements typically falls under scope-of-practice regulations for certified ophthalmic technologists, creating moderate regulatory and liability barriers to full replacement.
Cost vs. human wageclaude-haiku-4-5-202510011/5The equipment and human technologist labor needed for reliable keratometry measurement is already highly cost-efficient. AI augmentation would require expensive imaging hardware and software integration with minimal cost savings over the current low-cost direct measurement approach.
Cost vs. human wageclaude-sonnet-52/5Automated keratometry equipment reduces some labor time but the capital cost of devices plus required technician oversight keeps costs comparable to, not dramatically lower than, human-performed measurement.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system independently operates keratometers or ophthalmometers to produce measurements. Theoretical computer vision approaches exist for analyzing corneal topography images, but no production system reliably replaces the technologist's hands-on measurement workflow.
Technical feasibility todayclaude-sonnet-52/5Autorefractor/keratometer devices exist and are widely used, but they still require skilled human operation for setup, calibration checks, and quality control, so full end-to-end automation is not deployed.

Conduct visual field tests to measure field of vision.

25

CI 25–25 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmology practices adopt automation slowly; most clinics still rely on human-administered perimetry with technician oversight as standard of care. Adoption is limited by high equipment costs, regulatory conservatism, and the clinical requirement for professional interpretation, keeping this in the pilot/early-adoption phase rather than deep production.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic testing sectors adopt automation slowly due to regulatory, clinical validation, and safety requirements, with AI mainly assisting interpretation rather than replacing test administration.
Augmentation potentialclaude-haiku-4-5-202510014/5Automated perimetry systems substantially augment technician productivity by handling stimulus presentation and response logging, reducing fatigue-driven errors and speeding test execution. Technicians focus on patient comfort, fixation monitoring, and quality control, with AI handling the repetitive measurement and preliminary reliability assessment.
Augmentation potentialclaude-sonnet-53/5AI can assist by flagging unreliable test patterns, standardizing protocols, and aiding in interpretation of results, improving technologist efficiency and diagnostic support.
Task automatabilityclaude-haiku-4-5-202510012/5Visual field testing requires patient interaction, precise instrument calibration, and real-time response to patient performance—tasks that demand human presence and judgment. While some data logging and result analysis could be partially automated, the core test execution (monitoring patient fixation, recording responses, detecting unreliable results) remains heavily dependent on human technician oversight.
Task automatabilityclaude-sonnet-52/5While automated perimetry machines already perform much of the visual field test mechanically, the task includes patient setup, instruction, monitoring reliability, and interpretation of results which still require trained human oversight rather than being fully AI-driven end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Visual field testing is typically performed or signed off by licensed optometrists or ophthalmologists, and results must be interpreted by licensed professionals for diagnostic accuracy. Regulatory oversight (FDA, state licensure) and liability concerns around missed vision defects create significant legal barriers to full automation without professional authorization.
Adoption barriersclaude-sonnet-54/5Visual field testing is a clinical diagnostic procedure typically requiring a certified/licensed technologist to administer and ensure valid results, with liability tied to diagnostic accuracy in ophthalmic care.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current automated perimetry equipment is expensive ($30K–100K+), integration into clinic workflows requires training and maintenance, and human technician oversight remains necessary. The all-in cost per test still favors human technicians in most clinical settings, especially considering equipment amortization and support.
Cost vs. human wageclaude-sonnet-52/5Specialized diagnostic hardware and clinical oversight still require a trained technologist present, so cost savings versus a human-operated device are modest rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably conducts full visual field tests without human administration. Automated perimetry devices exist but require technician setup, patient instruction, and result validation; they assist rather than replace the technician. Research into fully autonomous testing exists but lacks production deployment in clinical settings.
Technical feasibility todayclaude-sonnet-52/5Automated perimeters are standard equipment, but they are device-based diagnostic tools operated by a technologist, not autonomous AI systems performing the full task independently in production.

Conduct tests, such as the Amsler Grid test, to measure central visual field used in the early diagnosis of macular degeneration, glaucoma, or diseases of the eye.

25

CI 25–25 · exposure 25 · augmentation 50 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmic practices are moderately digitized but adoption of autonomous AI testing remains limited; most clinics still rely on technologist-administered tests with human-led quality control. Adoption is slower than in fully digital domains, reflecting the procedural and patient-contact nature of the work.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially hands-on clinical diagnostic testing, has historically been a slower-adopting sector for full automation despite growing interest in AI-assisted diagnostics.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automatically analyzing Amsler Grid responses, flagging potential abnormalities, and generating preliminary reports, allowing technologists to focus on patient interaction and anomaly verification. This meaningful assistance improves workflow efficiency without removing the technologist from the testing loop.
Augmentation potentialclaude-sonnet-53/5AI can help analyze test results, flag anomalies, and support documentation, offering useful assistance to the technologist without replacing the hands-on test administration.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze Amsler Grid responses and eye imaging, the task requires real-time patient interaction, calibration of visual field testing equipment, and clinical judgment to ensure valid test administration. Current systems cannot independently conduct the full procedural task (patient instruction, equipment setup, result interpretation in context) with the required quality and time savings.
Task automatabilityclaude-sonnet-52/5Administering the Amsler Grid and similar visual field tests requires direct patient interaction, physical positioning, and real-time observation of patient responses that current AI cannot independently perform end-to-end.It could assist with data recording and pattern flagging but not the physical test administration.
Adoption barriersclaude-haiku-4-5-202510014/5Ophthalmic testing has regulatory oversight (FDA regulations on diagnostic devices), requires proper licensing/credentialing of personnel administering tests, and involves clinical liability if results are misread. Patients also expect human supervision and contact during eye examinations, creating both regulatory and organizational friction.
Adoption barriersclaude-sonnet-54/5Diagnostic eye testing typically requires supervision by licensed clinical staff, and results feed into medical diagnosis requiring professional interpretation and liability accountability, creating substantial regulatory and licensing barriers.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of AI analysis tools adds software costs and oversight requirements; the human technologist remains essential for procedural execution, equipment management, and patient interaction. AI reduces some analysis time but does not eliminate the dominant cost of technologist labor.
Cost vs. human wageclaude-sonnet-52/5Specialized ophthalmic testing equipment and any AI analysis layered on top still require technician oversight and calibration, making all-in automated costs not dramatically cheaper than a trained technologist performing the test.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can assist in analyzing Amsler Grid images and flagging abnormalities, but no deployed product reliably conducts the end-to-end testing procedure autonomously. Existing clinical software supports interpretation but does not replace the technologist's role in administering and supervising the test in real clinical settings.
Technical feasibility todayclaude-sonnet-52/5While automated perimetry devices exist and some AI-assisted diagnostic imaging tools are deployed, fully autonomous administration of subjective visual field tests like the Amsler Grid in production clinical settings is not standard practice.

Conduct tonometry or tonography tests to measure intraocular pressure.

24

CI 23–25 · exposure 25 · augmentation 50 · importance 5.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Ophthalmic practices remain heavily dependent on in-person clinical procedures and credentialed human technicians. Adoption of automation in this domain is minimal; the sector is slower-moving on AI integration compared to information and financial services.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic settings adopt automation more slowly due to regulatory, safety, and workflow integration requirements, though semi-automated tonometers have seen some uptake.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by analyzing tonometry results, flagging anomalies, and suggesting clinical context, improving technician and physician workflows in result interpretation and documentation. However, assistance is limited to post-measurement analysis rather than the measurement procedure itself.
Augmentation potentialclaude-sonnet-53/5Automated and non-contact tonometers already assist technologists by speeding up measurement and improving consistency, though the technologist remains essential for administering and interpreting the test.
Task automatabilityclaude-haiku-4-5-202510012/5Tonometry and tonography require precise physical positioning of specialized equipment against or near the eye, patient cooperation, and real-time clinical judgment. While image analysis of results could be partially automated, the hands-on measurement and patient interaction cannot be reliably performed by current AI systems without human operation of the device.
Task automatabilityclaude-sonnet-52/5This requires physical instrument application to the patient's eye, patient positioning, and real-time adjustment based on patient response, which current AI cannot perform end-to-end without a human operator physically present.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical testing procedures like tonometry typically require a licensed or credentialed ophthalmic technician or physician to perform and sign off on results. Regulatory bodies and clinical liability frameworks generally mandate human oversight and responsibility for accurate intraocular pressure measurement, creating strong adoption barriers.
Adoption barriersclaude-sonnet-54/5This is a clinical diagnostic procedure typically requiring credentialed personnel, with liability concerns around misdiagnosis of glaucoma risk and regulatory oversight of medical devices and their use.
Cost vs. human wageclaude-haiku-4-5-202510012/5The equipment and sensor hardware required for tonometry/tonography are substantial capital costs beyond AI inference. An ophthalmic technologist's loaded wage remains lower than the total system cost for full automation, and the speed advantage is minimal given that the procedure itself takes only minutes.
Cost vs. human wageclaude-sonnet-52/5Automated tonometry devices reduce some labor but still require a trained technician present for setup, patient interaction, and quality control, so cost savings versus a human technician are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system independently performs tonometry or tonography testing today. While AI assists in result interpretation and analysis in some ophthalmology workflows, the physical act of conducting the test and positioning the equipment remains entirely operator-dependent in clinical practice.
Technical feasibility todayclaude-sonnet-52/5Some automated tonometers (e.g., non-contact tonometers) exist and are used clinically, but they still require a technician to position the patient, operate the device, and interpret readings in context; fully autonomous testing is not deployed.

Instruct patients in the care and use of contact lenses.

23

CI 16–30 · exposure 20 · augmentation 63 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Eye care settings have moderate digitization and remain relatively conservative on automation due to direct patient contact requirements and liability; adoption of AI for this instructional task is nascent and slow.
Sector adoption velocityclaude-sonnet-52/5Healthcare/vision care sectors are slower AI adopters generally, and this specific patient-facing hands-on task has seen limited real-world AI deployment beyond basic informational chatbots.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by generating personalized written instructions, video demos, or follow-up reminders, moderately improving technologist productivity, but the core hands-on instruction requires human presence and judgment.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by generating personalized care instructions, answering patient FAQs, and providing reminder/adherence tools, letting technicians focus on hands-on training and complex troubleshooting.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could generate instructional content about contact lens care, the task requires hands-on demonstration, patient-specific fitting feedback, and real-time adjustment based on individual patient comprehension and physical dexterity—elements that current AI cannot perform autonomously.
Task automatabilityclaude-sonnet-52/5Patient education on contact lens care involves hands-on demonstration, physical fitting checks, and responsive troubleshooting that current AI cannot fully replicate end-to-end, though written/video instructions can be AI-generated.
Adoption barriersclaude-haiku-4-5-202510014/5Patient safety and liability concerns create strong barriers; contact lens care errors can cause serious eye infections or injury, and legal/regulatory expectations hold the human technologist responsible for proper patient instruction and sign-off.
Adoption barriersclaude-sonnet-53/5No strict licensing mandate requires this specific instruction be done by a certified professional in all jurisdictions, but liability concerns (improper lens care causing eye damage) and clinical practice norms create moderate friction against full automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems cannot yet replace the direct, hands-on instruction required, and any AI-assisted approach would still require a human technologist present, making the all-in cost higher than direct human instruction alone.
Cost vs. human wageclaude-sonnet-52/5AI-generated instructional content is cheap, but the hands-on supervision component still requires a paid technician, so overall cost savings versus the human-delivered task are modest.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product reliably teaches contact lens insertion, removal, and care independently; this task demands tactile guidance, visual feedback on patient technique, and adaptive correction that exceeds current AI capabilities in production settings.
Technical feasibility todayclaude-sonnet-52/5Some clinics use AI chatbots or apps for general lens care reminders, but no deployed product reliably conducts in-person instruction, insertion/removal training, or hands-on correction at scale.

Assess abnormalities of color vision, such as amblyopia.

23

CI 20–25 · exposure 20 · augmentation 50 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmic practices are moderately digitized, but assessment of abnormalities like amblyopia remains a hands-on, patient-interactive clinical task with high quality and liability requirements. Adoption of AI for diagnostic assessment in this field remains limited to research pilots.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic subspecialties adopt AI cautiously due to regulatory and liability concerns, with pilots more common than production-scale deployment for this specific task.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist a technologist by flagging potential color vision abnormalities in test results or suggesting patterns consistent with amblyopia, reducing manual review time and improving consistency, but the technologist would remain responsible for clinical interpretation and patient interaction.
Augmentation potentialclaude-sonnet-53/5AI-assisted analysis of test results or imaging can help flag abnormalities and support interpretation, improving efficiency while the technologist remains central to test administration and judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze color vision test images and detect some statistical deviations from normal, comprehensive assessment of amblyopia requires integrating patient history, behavioral observation, and clinical judgment about causation and severity. Current systems lack the sensorimotor integration and real-time patient interaction needed to perform the full diagnostic task at 50% time savings.
Task automatabilityclaude-sonnet-52/5Administering and interpreting specialized color vision and amblyopia tests requires hands-on patient interaction, equipment operation, and clinical judgment that current AI cannot fully replicate end-to-end., though some scoring/interpretation aid is possible.
Adoption barriersclaude-haiku-4-5-202510014/5Medical diagnosis and assessment carry regulatory oversight under clinical practice standards; color vision and amblyopia assessment directly informs treatment decisions and must be validated by a licensed or certified technologist. Liability and the requirement for professional sign-off create substantial legal and organizational barriers.
Adoption barriersclaude-sonnet-54/5Clinical vision assessments typically require a credentialed technician/technologist and are embedded in regulated healthcare workflows with liability concerns, limiting substitution without licensed oversight.
Cost vs. human wageclaude-haiku-4-5-202510012/5Deployment would require integration into clinical workflows, quality assurance, and technologist oversight for patient interaction and result validation. The combined cost likely exceeds that of a trained technologist performing the task directly, particularly given liability and accuracy requirements.
Cost vs. human wageclaude-sonnet-52/5AI diagnostic tools require expensive imaging/testing hardware, integration, and human oversight, making all-in costs comparable to or higher than a technologist's wage for this narrow task.
Technical feasibility todayclaude-haiku-4-5-202510012/5Research systems can screen for color vision deficiencies using automated image analysis, but no production-deployed product reliably assesses the full clinical context of amblyopia without significant technician involvement and review. Existing tools support data collection but do not replace the technologist's interpretive role.
Technical feasibility todayclaude-sonnet-51/5No widely deployed product autonomously performs this specific clinical assessment in production; existing AI tools are research-stage or narrow diagnostic aids, not full replacements for the technologist's exam.

Measure visual acuity, including near, distance, pinhole, or dynamic visual acuity, using appropriate tests.

22

CI 19–25 · exposure 25 · augmentation 50 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Ophthalmic practices remain primarily human-centric and operator-dependent; while EHRs and automated refractors exist, actual displacement of technician-performed acuity measurement is minimal and adoption of AI agents in this domain is negligible.
Sector adoption velocityclaude-sonnet-52/5Healthcare/ophthalmology has been slow to adopt full automation of hands-on clinical testing tasks, though some digital acuity tools are gradually being piloted.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automatically scoring test results, flagging abnormal patterns, and recommending follow-up tests, which raises technician efficiency in documentation and triage, though the core measurement and patient interaction remain human-performed.
Augmentation potentialclaude-sonnet-53/5Digital acuity charts and semi-automated devices can speed up test administration and standardize results, aiding technologists without replacing them.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can analyze visual acuity test results and charts, the task requires physical administration (positioning the patient, displaying stimuli at precise distances, assessing responses in real-time) and clinical judgment that current systems cannot perform end-to-end without human intervention. Partial automation of analysis is possible but does not meet the 50% time-saving threshold.
Task automatabilityclaude-sonnet-52/5Some automated visual acuity testing devices exist, but the full task including patient positioning, technique adjustment, and clinical judgment for varied presentations still requires substantial human involvement.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: visual acuity measurement is a licensed clinical procedure typically required to be performed or directly supervised by an ophthalmic professional; liability and regulatory requirements (e.g., state licensure for technologists) create legal and compliance friction around unsupervised automation.
Adoption barriersclaude-sonnet-54/5This is a clinical diagnostic measurement often requiring a credentialed technologist under physician supervision, with liability concerns around misdiagnosis limiting full automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI infrastructure, integration, and oversight to partially automate this task (combined with retained human oversight) currently exceeds the cost of a trained technician performing it directly.
Cost vs. human wageclaude-sonnet-52/5Specialized automated equipment involves significant capital and maintenance costs, and human technologists remain in the loop for calibration, patient handling, and edge cases, keeping cost savings modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs the full measurement task independently. AI-assisted chart reading and analysis tools exist, but the hands-on clinical administration—especially dynamic visual acuity testing—requires a human technician and lacks production-scale automation.
Technical feasibility todayclaude-sonnet-52/5Automated acuity testing devices and computerized charts exist in some clinics but are not universally deployed and rarely replace technologist-administered testing for all acuity types (pinhole, dynamic).

Photograph patients' eye areas, using clinical photography techniques, to document retinal or corneal defects.

21

CI 16–25 · exposure 20 · augmentation 50 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmic practices remain heavily reliant on human technicians for image capture and quality assurance. Adoption of autonomous or near-autonomous imaging is limited to specialized research centers and advanced facilities, not mainstream clinical practice, indicating slow real-world displacement.
Sector adoption velocityclaude-sonnet-52/5Healthcare imaging technology adoption is steady but the physical hands-on nature of this specific task limits AI-driven displacement; robotic imaging systems are not yet mainstream in ophthalmology clinics.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist technicians by providing real-time feedback on image focus, positioning, and quality metrics, and by flagging potential defects post-capture for technician review. These tools enhance productivity and consistency but do not fundamentally transform the task when the human remains the primary operator and decision-maker.
Augmentation potentialclaude-sonnet-53/5AI can assist with image quality assessment, auto-alignment features in modern fundus cameras, and post-capture image analysis/annotation, improving efficiency without replacing the technologist.
Task automatabilityclaude-haiku-4-5-202510012/5While image capture itself is automatable, clinical photography of eyes requires real-time patient positioning, alignment, cooperation, and judgment to obtain diagnostic-quality images. Current systems cannot reliably handle patient variability, equipment adjustments, and quality assessment without human oversight, falling well short of the 50% time-saving threshold.
Task automatabilityclaude-sonnet-52/5Physical positioning of patients and operation of specialized ophthalmic imaging equipment (fundus cameras, OCT) requires hands-on technique and patient interaction that current AI cannot perform end-to-end; AI can assist in image analysis but not the physical capture process.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical photography for diagnosis is embedded in regulated medical workflows and often requires licensed or credentialed personnel to ensure quality and liability. Insurance and medical device regulations govern image capture standards, and diagnostic use demands clinician sign-off, creating meaningful organizational and regulatory friction.
Adoption barriersclaude-sonnet-54/5Requires direct physical patient contact, proper equipment calibration, and clinical training/certification; liability for missed or poor-quality diagnostic images is high, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current ophthalmic imaging equipment combined with AI integration, quality control, and operator oversight costs remain comparable to or exceed the loaded wage of an ophthalmic technician. The equipment infrastructure and per-image processing costs do not yet achieve significant economic advantage.
Cost vs. human wageclaude-sonnet-51/5AI cannot replace the physical act of imaging a patient's eye, so there is no substitutive cost comparison—human labor remains mandatory.
Technical feasibility todayclaude-haiku-4-5-202510012/5Automated eye imaging devices exist in research and limited clinical settings (e.g., retinal cameras), but they require human operators for positioning, focusing, and ensuring diagnostic quality. No mature end-to-end autonomous system demonstrates reliable, production-grade performance across diverse patient populations and pathologies without trained technician input.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously photographs patient eyes without a trained technician operating and aligning the equipment; this remains a manual clinical procedure.

Conduct binocular disparity tests to assess depth perception.

21

CI 16–25 · exposure 20 · augmentation 38 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmic practices remain moderately digitized and fragmented (many small offices), with strong institutional preference for human technologist judgment and direct patient contact. Adoption of AI-assisted or fully automated depth-perception testing is still in pilot phases in large academic centers; mainstream practices lag significantly in deployment.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic testing, especially hands-on ophthalmic exams, sees slow AI adoption due to physical and regulatory constraints, though AI is used more in image analysis than physical test administration.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could meaningfully assist ophthalmic technologists by automating stimulus presentation, real-time data logging, preliminary scoring, and flagging abnormal results for technologist review. This would raise productivity on data management and reproducibility, though human judgment and patient interaction remain central to valid test administration.
Augmentation potentialclaude-sonnet-52/5AI could assist with recording results, flagging abnormal patterns, or supporting interpretation, but offers limited help with the physical test administration itself.
Task automatabilityclaude-haiku-4-5-202510012/5Binocular disparity testing involves presenting stereoscopic stimuli and interpreting responses, which AI could theoretically automate stimulus presentation for. However, the task critically requires real-time patient interaction, calibration of equipment to individual anatomy, and clinical judgment about test validity—elements that current AI systems cannot reliably handle end-to-end without substantial human oversight, falling well short of 50% time savings at equal quality.
Task automatabilityclaude-sonnet-52/5This requires physical positioning of the patient, operation of specialized instruments (e.g., stereopsis testing devices), and interpretation of responses in real time, which current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Significant regulatory and liability barriers exist: ophthalmic testing is regulated by state boards and clinical standards (e.g., OSHA, medical device directives), results inform medical diagnoses and treatment decisions, and malpractice liability attaches to test validity. Patients expect and often require trained human technologists; clinical laboratories typically must maintain licensed personnel oversight of diagnostic test administration.
Adoption barriersclaude-sonnet-54/5This is a clinical diagnostic procedure typically requiring a trained/certified ophthalmic technician, involving direct patient contact and equipment operation, with liability tied to accurate diagnosis.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI solutions for ophthalmic testing are expensive specialized software/hardware bundles requiring integration into clinical workflows, technologist oversight, and equipment maintenance. The all-in cost per test remains comparable to or higher than the loaded wage of an ophthalmic technologist performing the task, given the narrow deployment base.
Cost vs. human wageclaude-sonnet-51/5No viable AI substitute exists for the physical test administration, so AI cost cannot be compared favorably; a human technician remains necessary.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system reliably conducts binocular disparity tests independently. While computer vision and psychophysical testing software exist in research, production ophthalmic systems do not autonomously administer, score, and interpret disparity tests without a trained technologist present to operate equipment and validate results.
Technical feasibility todayclaude-sonnet-51/5There are no deployed products that autonomously conduct binocular disparity/stereopsis testing in clinical settings; this remains a hands-on technician task.

Perform slit lamp biomicroscopy procedures to diagnose disorders of the eye, such as retinitis, presbyopia, cataracts, or retinal detachment.

16

CI 7–25 · exposure 13 · augmentation 50 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmology practices have been slow to adopt autonomous diagnostic automation, favoring AI as an ancillary aid to technologist-performed exams rather than a replacement. Adoption remains concentrated in large academic centers and specialized clinics, not widespread across the profession.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially hands-on clinical diagnostics, has been slower to adopt AI for physical exam tasks compared to purely digital/administrative healthcare functions.
Augmentation potentialclaude-haiku-4-5-202510013/5AI shows useful assistance on specific subtasks: automated image quality assessment, cataract severity grading, and referral flagging can improve technologist efficiency and consistency. However, the augmentation is partial—the human must still control the instrument, position the patient, and make judgment calls on examination completeness.
Augmentation potentialclaude-sonnet-53/5AI-assisted image analysis and decision support tools can help technologists interpret findings or flag abnormalities, but the core biomicroscopy procedure itself sees limited AI augmentation.
Task automatabilityclaude-haiku-4-5-202510012/5Slit lamp biomicroscopy requires skilled manual operation, precise physical alignment, and real-time interpretation of complex optical imagery to diagnose subtle pathology. While AI can assist in image analysis post-acquisition, the procedural execution—patient positioning, lamp adjustment, lens selection, and dynamic inspection—remains heavily dependent on human sensorimotor control and clinical judgment, limiting full automation below the 50% time-saving threshold.
Task automatabilityclaude-sonnet-51/5Slit lamp biomicroscopy requires hands-on physical instrument manipulation, precise positioning of a patient, and real-time interactive assessment that current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: ophthalmic procedures often require a licensed clinical technologist or optometrist to perform and certify findings; regulatory bodies (FDA, state boards) govern diagnostic device use; malpractice liability for misdiagnosis is high; and many patients expect direct human contact for eye examination. These legal and organizational frictions substantially protect human technologists.
Adoption barriersclaude-sonnet-54/5This is a clinical diagnostic procedure typically requiring licensed/certified personnel, direct patient contact, and liability considerations around eye health diagnosis.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI image analysis tools require expensive hardware integration, human oversight, and clinician validation. The all-in cost per diagnostic procedure remains comparable to or higher than employing a trained technologist, especially when accounting for integration and liability overhead.
Cost vs. human wageclaude-sonnet-51/5AI cannot yet replace the physical examination process, so there is no viable cost comparison; the human technician remains necessary for the procedure.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI systems can analyze slit lamp images offline and flag some pathologies (e.g., cataract grading), but no deployed product reliably performs the full procedure (positioning, imaging acquisition, dynamic examination) autonomously in clinical settings. Existing tools are narrow assistants to interpretation, not autonomous performers of the task.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously performs slit lamp exams; AI image analysis tools exist for interpreting captured fundus/anterior segment images but not for conducting the physical procedure itself.

Conduct ocular motility tests to measure function of eye muscles.

16

CI 7–25 · exposure 13 · augmentation 38 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Ophthalmology has moderate digitization but remains slower than information-intensive sectors; adoption of AI-driven diagnostic aids is still in pilot phases at academic centers, with limited production deployment in typical clinical practice.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic testing, especially hands-on ophthalmic exams, sees slower AI adoption compared to purely digital/information sectors, with automation limited to some imaging analysis rather than physical exams.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automatically flagging abnormal motion patterns, suggesting test variants, or accelerating data logging, but the technician must remain central to test execution, patient communication, and clinical decision-making for this safety-critical diagnostic task.
Augmentation potentialclaude-sonnet-52/5AI can assist with recording, quantifying eye movement data via video-oculography analysis or supporting documentation, but this augmentation is narrow and not yet widespread in routine motility testing.
Task automatabilityclaude-haiku-4-5-202510012/5Ocular motility testing requires precise measurement of eye movement and muscle function, which demands real-time visual feedback, patient cooperation, and physical alignment—elements that current AI systems struggle to automate end-to-end. While AI can assist with image analysis of eye movements, the full procedure (patient positioning, test execution, real-time adjustment, and clinical interpretation) remains heavily dependent on technician expertise and patient interaction.
Task automatabilityclaude-sonnet-51/5This requires hands-on physical manipulation, observation of eye movement, and patient interaction that current AI systems cannot perform end-to-end; no off-the-shelf system replaces this physical clinical exam.'
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory frameworks (FDA clearance for diagnostic devices) and clinical standards require licensed or certified personnel to conduct and validate ocular motility tests; the task often carries legal liability for misdiagnosis, and ophthalmologists typically require human technician sign-off before patient treatment decisions.
Adoption barriersclaude-sonnet-54/5Ocular motility testing is a clinical procedure typically requiring trained/certified personnel and is part of a regulated diagnostic workflow, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5The specialized equipment (eye trackers, video-based systems, integrations with EHR) and required clinical oversight currently cost more than or are comparable to employing a trained ophthalmic medical technologist for routine testing.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical exam, so the human technologist remains the only cost-effective option for this task.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs complete ocular motility testing autonomously in clinical settings today. Research exists on eye-tracking AI and motion analysis, but production systems require technician oversight, manual equipment positioning, and validation—falling short of full end-to-end automation at scale in real ophthalmology practices.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product autonomously conducts ocular motility testing; this remains a manual clinical procedure performed by trained technologists.

Clean or sterilize ophthalmic or surgical instruments.

15

CI 5–25 · exposure 13 · augmentation 25 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of automated sterilization systems has been slow in many ophthalmic offices and smaller surgical centers due to cost, space constraints, and regulatory compliance complexity. Only larger hospital systems and surgery centers with dedicated sterile processing departments have widely adopted such equipment.
Sector adoption velocityclaude-sonnet-51/5Healthcare instrument sterilization remains a manual, low-digitization physical process; adoption of AI or robotics for this specific task is minimal and not part of current sector AI adoption trends.
Augmentation potentialclaude-haiku-4-5-202510013/5Automated washers and sterilizers assist technologists by handling bulk processing, reducing manual handling time, and providing consistent results. However, the assistance is limited to specific steps; technologists must still manually prepare, load, unload, inspect, and document sterilization outcomes.
Augmentation potentialclaude-sonnet-51/5AI offers essentially no meaningful assistance for the physical cleaning and sterilization steps themselves, though software may track compliance logs separately from the hands-on task.
Task automatabilityclaude-haiku-4-5-202510012/5Cleaning and sterilizing surgical instruments involves specialized handling, validation of sterilization protocols, and quality verification that requires human judgment and physical dexterity. While some robotic systems exist for instrument handling in specialized settings, they require significant setup and cannot reliably achieve 50% time savings at equal quality across typical ophthalmic practices today.
Task automatabilityclaude-sonnet-51/5This is a physical manual task requiring dexterity to clean and sterilize delicate instruments; current AI systems (software/LLMs) cannot perform physical cleaning or sterilization.It requires robotic hardware not commonly deployed for this purpose.
Adoption barriersclaude-haiku-4-5-202510014/5Sterilization in surgical settings is heavily regulated by FDA, CDC, and state health authorities, with strict validation and documentation requirements that mandate human oversight and sign-off. Liability for instrument sterility is legally tied to responsible personnel, creating strong regulatory barriers to full automation.
Adoption barriersclaude-sonnet-54/5Infection control regulations, sterilization protocols, and liability for improperly sterilized surgical instruments impose strict procedural and often certification-based requirements on who can perform and verify this task.
Cost vs. human wageclaude-haiku-4-5-202510012/5Automated sterilization equipment has high upfront capital costs and ongoing maintenance expenses. For small to mid-size ophthalmic practices, the total cost of ownership often exceeds the loaded labor cost of a technologist performing manual cleaning and sterilization.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI-based substitute performing this physical task, so cost comparison favors the human worker by default; specialized sterilization equipment has cost but isn't AI-driven automation of the judgment/handling involved.
Technical feasibility todayclaude-haiku-4-5-202510012/5Automated instrument washers and sterilizers exist and are used in many facilities, but they are semi-automated tools requiring human loading, unloading, and verification rather than fully autonomous systems. No deployed AI product performs end-to-end cleaning and sterilization validation at production scale in ophthalmic settings.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs autonomous cleaning/sterilization of ophthalmic surgical instruments in clinical settings; automated sterilizers exist but are equipment, not AI-driven task replacement of the technologist's role.

Collect ophthalmic measurements or other diagnostic information, using ultrasound equipment, such as A-scan ultrasound biometry or B-scan ultrasonography equipment.

13

CI 5–21 · exposure 13 · augmentation 38 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare imaging remains a laggard sector for autonomous automation due to regulatory, liability, and patient-safety requirements. Ophthalmic practices are typically small and conservative, with strong institutional reliance on certified human technologists rather than deployed automated systems.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic imaging sectors adopt AI slowly for hands-on patient procedures, with most AI advances limited to image analysis rather than physical data acquisition.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-assisted image analysis (automated cataract lens power calculation, IOL recommendations from B-scan images, measurement quality flags) can assist technologists in interpretation and documentation, but the actual acquisition process remains human-driven and benefits less from augmentation.
Augmentation potentialclaude-sonnet-52/5AI can assist with post-scan image analysis or interpretation support, but offers little assistance in the physical act of collecting the ultrasound measurements themselves.
Task automatabilityclaude-haiku-4-5-202510012/5While ultrasound image capture itself is partially automatable, the task requires real-time patient positioning, probe manipulation, and clinical judgment to obtain diagnostic-quality measurements. Current AI cannot reliably execute the physical probe placement, tissue contact optimization, and on-the-fly measurement decisions that define competent ophthalmic ultrasound work.
Task automatabilityclaude-sonnet-51/5This requires hands-on manipulation of ultrasound probes on a patient's eye, real-time interpretation, and physical positioning skills that no current AI system can perform autonomously.
Adoption barriersclaude-haiku-4-5-202510014/5Ophthalmic ultrasound measurement has regulatory oversight (requires trained operators and physician supervision), patient contact is mandatory, and clinical liability for diagnostic error is high. Medical device automation of this task would face FDA scrutiny and institutional credentialing requirements that create substantial adoption friction.
Adoption barriersclaude-sonnet-54/5Direct patient contact, use of medical devices, and clinical certification requirements create strong barriers, though not all jurisdictions require full licensure for this specific task.
Cost vs. human wageclaude-haiku-4-5-202510011/5The equipment, specialized training, and liability insurance required to automate ultrasound measurement would exceed the cost of employing a technologist for years. Additionally, oversight and validation of measurements would still require a human expert, negating cost savings.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical data-collection task, so the human technologist remains the only cost-effective option today.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system can independently perform ophthalmic ultrasound measurement end-to-end. Image analysis and measurement extraction from ultrasound exist in research and narrow clinical settings, but the embodied task of patient interaction, probe handling, and real-time quality assessment remains purely human-operated in production.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously collects A-scan or B-scan ultrasound measurements on patients; this remains a physical, hands-on clinical procedure performed by trained technologists.

Assess refractive condition of eyes, using retinoscope.

13

CI 5–21 · exposure 13 · augmentation 38 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Ophthalmology remains a human-contact, regulation-bound clinical specialty with slow digitization of core diagnostic procedures. Retinoscopy adoption of AI alternatives is virtually nonexistent, and the specialty has not pursued automation of this foundational refractive assessment technique.
Sector adoption velocityclaude-sonnet-52/5Eye care/ophthalmology is a physical, hands-on clinical field with slower AI adoption for direct diagnostic manipulation tasks, though autorefraction technology has some presence as an adjunct tool.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-assisted image analysis of fundus photos or automated refraction algorithms could help technologists interpret and cross-check findings, but current tools offer modest support. Augmentation potential exists through decision-support systems, though retinoscopy itself demands human manual skill.
Augmentation potentialclaude-sonnet-52/5AI-enabled autorefractors and diagnostic software can supplement retinoscopy findings, but they don't meaningfully augment the technologist's actual retinoscope-based assessment technique itself.
Task automatabilityclaude-haiku-4-5-202510012/5Retinoscopy is a hands-on clinical technique requiring precise physical manipulation of a retinoscope and real-time interpretation of light reflexes on the retina. While image analysis of retinal photos is automatable, the dynamic observation and physical instrument control intrinsic to retinoscopy cannot be performed end-to-end by current AI without human intervention.
Task automatabilityclaude-sonnet-51/5Retinoscopy requires manual manipulation of a handheld instrument, real-time observation of light reflex movement, and physical positioning relative to the patient's eye, none of which current AI systems can perform end-to-end without robotic hardware that doesn't exist in deployed form.
Adoption barriersclaude-haiku-4-5-202510014/5Retinoscopy results inform eyeglass/contact lens prescriptions that directly affect patient safety and vision; a licensed healthcare provider must clinically validate findings. Regulatory and liability frameworks require human professional oversight, and patient contact is intrinsic to the task.
Adoption barriersclaude-sonnet-54/5Refractive assessment is typically performed by licensed/certified ophthalmic personnel under clinical protocols, and diagnostic error has direct patient-care consequences, creating substantial regulatory and liability barriers to substitution.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of developing and deploying a robotic arm with AI vision to perform retinoscopy would far exceed the wages of a trained ophthalmic technologist for years. Human-performed retinoscopy remains substantially cheaper than any viable automated alternative.
Cost vs. human wageclaude-sonnet-51/5Since no viable AI substitute performs this physical diagnostic task, there is no meaningful AI cost basis to compare; the human technologist remains the only cost-effective option.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system can independently perform retinoscopy. Research prototypes may analyze retinal images, but actual retinoscopy—the iterative, tactile adjustment of lenses while observing reflex position—requires physical presence and human clinical judgment currently unavailable in production systems.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs manual retinoscopy; autorefractors exist as separate devices but do not replicate the technologist's handheld retinoscope assessment, and no AI-driven robotic system does this in production clinics.

Maintain ophthalmic instruments or equipment.

12

CI 5–19 · exposure 8 · augmentation 25 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Ophthalmic practices and medical facilities are subject to strict regulatory oversight and tend toward conservative adoption of automation in clinical support functions. No evidence of meaningful AI adoption for equipment maintenance in this sector.
Sector adoption velocityclaude-sonnet-51/5Healthcare equipment maintenance is a physical, low-digitization task with minimal AI/robotics adoption in production settings.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could provide minor assistance through automated tracking of maintenance schedules, equipment inventory, and diagnostic checklists, but the core task of physical inspection and repair remains manual. Augmentation is limited to administrative and organizational support.
Augmentation potentialclaude-sonnet-52/5AI could assist with maintenance scheduling, diagnostic alerts, or documentation, but offers little help with the actual physical upkeep of the instruments.
Task automatabilityclaude-haiku-4-5-202510012/5Maintaining ophthalmic instruments involves physical inspection, calibration, and repair work that requires human dexterity and visual judgment. Current AI systems cannot reliably perform hands-on maintenance, cleaning, or component replacement, though they could assist with documentation and scheduling.
Task automatabilityclaude-sonnet-51/5Physically maintaining, cleaning, calibrating, and troubleshooting ophthalmic instruments requires hands-on manipulation of specialized hardware that current AI systems cannot perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510014/5Ophthalmic equipment maintenance often requires technician certification and manufacturer authorization. Liability exposure for equipment failure affecting patient safety, combined with regulatory requirements for device upkeep, creates substantial legal and compliance barriers to automation.
Adoption barriersclaude-sonnet-53/5While not formally licensed maintenance work, it requires hands-on technical skill and often manufacturer training or certification for calibration of medical devices, creating moderate friction against automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot yet perform this task competently, so cost comparison is moot. Physical maintenance robots capable of working on precision ophthalmic equipment are not commercially available or cost-effective.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute for the physical labor involved, so AI cost comparison is not applicable and the human remains the only cost-effective option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system performs end-to-end ophthalmic instrument maintenance in production. While computer vision can inspect surfaces, the integration of diagnosis, logistics, and hands-on repair remains firmly in human domain.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs physical equipment maintenance for ophthalmic instruments; this remains a manual technician task requiring physical dexterity and equipment-specific knowledge.

Conduct low vision blindness tests.

8

CI 0–16 · exposure 8 · augmentation 38 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare adoption of AI in clinical testing is cautious and heavily regulated; low vision assessment remains a specialized, in-person diagnostic service with minimal sector-wide automation despite general healthcare digitization.
Sector adoption velocityclaude-sonnet-52/5Healthcare diagnostic testing sectors adopt AI slowly for hands-on patient procedures, with most AI use confined to image analysis rather than test administration.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could potentially assist with data logging, statistical analysis of test results, or guideline lookup, but the core task of conducting the test itself—patient positioning, equipment operation, response interpretation—offers limited augmentation opportunities while the technologist remains the primary agent.
Augmentation potentialclaude-sonnet-53/5AI can assist with data recording, analysis of test results, and flagging abnormalities, but the physical test administration itself sees limited AI augmentation.
Task automatabilityclaude-haiku-4-5-202510011/5Low vision blindness tests require direct patient interaction, precise manual instrument calibration, and real-time adaptation to individual patient responses and visual capabilities. Current AI systems cannot reliably perform the hands-on assessment and equipment manipulation needed for valid testing outcomes.
Task automatabilityclaude-sonnet-52/5Administering low vision tests requires hands-on patient interaction, equipment operation, and real-time judgment adjustments that current AI cannot perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Medical testing on human subjects falls under healthcare regulation; technologists must be licensed and credentialed, patients require informed consent, and liability for incorrect vision assessment falls on the credentialed professional—creating hard legal and regulatory barriers to full automation.
Adoption barriersclaude-sonnet-54/5Direct patient contact, clinical judgment, and often certification/licensure requirements for administering vision tests create strong barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The task involves specialized clinical equipment, patient presence, and legal/medical accountability; AI systems would require expensive integration with ophthalmic devices and cannot substitute for the technologist's physical and clinical role, making automation more costly than human performance.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical, patient-facing test, so cost comparison favors the human entirely.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs comprehensive low vision blindness testing independently; the task fundamentally requires in-person clinical interaction with specialized ophthalmic equipment that only trained technologists can operate and interpret in real time.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously conducts low vision blindness testing on patients; this remains a hands-on clinical task performed by trained technologists.

Supervise or instruct ophthalmic staff.

5

CI 5–5 · exposure 0 · augmentation 38 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare organizations remain conservative on autonomous decision-making and maintain human supervision structures as a fundamental organizational requirement, with minimal AI displacement of supervisory roles.
Sector adoption velocityclaude-sonnet-51/5Healthcare management and clinical supervision roles show minimal AI-driven displacement or adoption in practice today.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist supervisors with scheduling optimization or documentation drafting, but these are peripheral; the core supervisory task—judgment, feedback, accountability—remains primarily human-driven with limited augmentation potential.
Augmentation potentialclaude-sonnet-53/5AI can help with training materials, scheduling, performance tracking, and knowledge checks, moderately aiding supervisors without replacing their judgment.
Task automatabilityclaude-haiku-4-5-202510011/5Supervision and instruction of staff requires real-time judgment, interpersonal presence, conflict resolution, and performance coaching—capabilities that current AI systems cannot deliver end-to-end. AI cannot serve as an ongoing supervisor or meaningful instructor in a medical setting.
Task automatabilityclaude-sonnet-51/5Supervising and instructing staff requires interpersonal leadership, real-time judgment about individual competencies, and accountability that current AI cannot replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Medical settings require licensed, accountable human supervisors; regulatory frameworks implicitly assume human oversight of staff. Liability and patient safety standards create strong barriers to replacing supervisory functions with AI.
Adoption barriersclaude-sonnet-54/5Supervisory responsibility often carries organizational and sometimes regulatory accountability (e.g., certification oversight, liability for clinical errors), making a human overseer necessary in practice.
Cost vs. human wageclaude-haiku-4-5-202510011/5Human supervisors command significant loaded wages; there is no current AI system that performs supervisory functions at any fraction of that cost, since the task is not meaningfully automated.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute providing supervisory output, so cost comparison favors the human role entirely.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably supervises or instructs medical staff in production environments. While AI can assist with scheduling or generate training materials, it cannot replace the supervisory role that requires accountability, decision authority, and human judgment.
Technical feasibility todayclaude-sonnet-51/5No deployed products perform clinical staff supervision or in-person training of ophthalmic technicians; this remains a human management function.

Perform fluorescein angiography of the eye.

3

CI 0–5 · exposure 5 · augmentation 50 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Ophthalmic practices remain traditional in procedural delivery; automation of hands-on clinical procedures is not occurring in the field, with adoption limited to image analysis tools rather than task displacement of technologists.
Sector adoption velocityclaude-sonnet-51/5Ophthalmic imaging procedures are physically performed in clinical settings with minimal AI displacement of the hands-on task itself, though AI is used for downstream image analysis.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by automating post-procedure image analysis, quality assessment, and flagging abnormalities for technologist review, improving diagnostic workflow without replacing the technologist's procedural role.
Augmentation potentialclaude-sonnet-53/5AI can assist by analyzing captured angiography images for diagnostic patterns (e.g., leakage detection), but it does not assist with the actual performance of injecting dye and capturing images.
Task automatabilityclaude-haiku-4-5-202510011/5Fluorescein angiography requires direct physical handling of patients (dilating eyes, injecting dye intravenously, positioning at imaging equipment) and real-time clinical judgment to manage complications or patient discomfort—tasks fundamentally beyond current AI automation without human presence.
Task automatabilityclaude-sonnet-51/5This is a hands-on clinical procedure requiring IV dye injection, precise camera operation, and real-time patient monitoring that current AI cannot physically perform.
Adoption barriersclaude-haiku-4-5-202510015/5This task involves invasive procedures (intravenous dye injection), direct patient contact, and medical licensing requirements; only qualified ophthalmic technologists are legally authorized to perform fluorescein angiography, creating hard regulatory and liability barriers to automation.
Adoption barriersclaude-sonnet-55/5Injecting dye and operating diagnostic imaging equipment on patients requires licensed medical personnel, with strict liability, safety, and regulatory requirements around invasive procedures.
Cost vs. human wageclaude-haiku-4-5-202510011/5The equipment cost, regulatory compliance, and need for licensed technician oversight make this procedure expensive; AI image analysis is a small fraction of total cost, and no substitution would meaningfully reduce the per-procedure expense versus a trained technologist's wage.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this procedure, so no cost comparison favors AI; the human technologist is the only option for delivery.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with image analysis post-acquisition, no deployed system performs the full procedure autonomously; the image capture itself requires skilled human technician operation of specialized equipment and direct patient interaction, which remains at the research stage for full automation.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs the physical administration and imaging capture of fluorescein angiography; this remains entirely a human clinical procedure.

Assist physicians in performing ophthalmic procedures, including surgery.

1

CI 0–3 · exposure 0 · augmentation 38 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of AI in ophthalmic surgery remains at the research and pilot stage, focused on diagnostic imaging or surgical planning rather than real-time procedure assistance. Hospitals and surgical centers continue to rely on human technologists and show no material shift toward automated assistants.
Sector adoption velocityclaude-sonnet-52/5Healthcare surgical settings adopt AI slowly for hands-on tasks due to safety, liability, and regulatory constraints, though imaging/diagnostic AI adoption is increasing elsewhere in ophthalmology.
Augmentation potentialclaude-haiku-4-5-202510012/5AI has minimal current role in augmenting a technologist's performance during surgery; preoperative imaging analysis and postoperative documentation tools offer modest assistance, but the core surgical-assistance task itself is not materially augmented by available AI systems today.
Augmentation potentialclaude-sonnet-53/5AI-assisted imaging, surgical planning, and robotic-assisted tools can enhance precision and workflow support around the procedure, even though the hands-on assisting role itself remains human.
Task automatabilityclaude-haiku-4-5-202510011/5Ophthalmic surgical assistance is a highly dynamic, real-time task requiring physical dexterity, sterile field management, anticipatory judgment, and direct interaction with a surgeon and patient. Current AI cannot physically position instruments, manage the surgical environment, or adapt assistive actions to evolving intraoperative conditions.
Task automatabilityclaude-sonnet-51/5Physically assisting in surgical procedures requires real-time manual dexterity, sterile technique, and physical presence that current AI systems cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Surgical assistance is legally and clinically embedded within the physician's responsibility. Malpractice, patient consent, sterile protocol, and medical licensure frameworks require a qualified human assistant to be present and accountable; regulatory bodies and hospital credentialing do not permit automated surgical assistance as a substitute.
Adoption barriersclaude-sonnet-55/5Surgical assistance is tightly regulated, requires certified/licensed personnel, involves direct patient contact and liability, and mandates human presence in the OR.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of an AI system capable of performing surgical assistance (hardware, integration, real-time computer vision, safety certification) would far exceed the fully-loaded wage of a trained ophthalmic technologist, especially when factoring in the redundancy and oversight required in a surgical context.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical assistive role, so cost comparison favors humans entirely; any AI-enabled equipment adds cost rather than replacing labor.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system today reliably performs surgical assistance as a standalone agent. Research prototypes exist for robotic guidance in controlled settings, but production systems require a human technologist in the operating room for safety, legal, and quality-assurance reasons.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously assists in ophthalmic surgery in the hands-on, physical-support role this task describes; robotic surgical aids exist but require human technologists, not AI substitution.

Administer topical ophthalmic or oral medications.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare settings, particularly ophthalmology clinics, have slow adoption of automation for direct patient care tasks due to regulatory constraints, liability concerns, and patient preference for human contact during medication administration.
Sector adoption velocityclaude-sonnet-51/5Healthcare direct patient care tasks involving medication administration show minimal automation adoption due to safety-critical, hands-on nature.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist by verifying correct medication identification or flagging drug interactions before administration, but the core task of physically administering the medication remains human-performed with minimal AI augmentation potential.
Augmentation potentialclaude-sonnet-52/5AI can support documentation, dosage reminders, or scheduling around medication administration, but offers minimal direct assistance to the physical act itself.
Task automatabilityclaude-haiku-4-5-202510011/5Administering medications to patients requires physical handling (instilling eye drops, ensuring proper dosage) and direct patient contact that current AI systems cannot perform. While AI could support medication verification or scheduling, the core task of safe medication administration remains exclusively human.
Task automatabilityclaude-sonnet-51/5Physically administering eye drops or oral medications to a patient requires hands-on interaction, patient identification, and clinical judgment that current AI systems cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Medication administration is heavily regulated; only licensed healthcare personnel can legally administer medications in most jurisdictions. Liability, patient safety requirements, and strict regulatory oversight create near-absolute barriers to automation.
Adoption barriersclaude-sonnet-55/5Administering medications is a licensed clinical activity with strict regulatory, liability, and safety requirements mandating human execution.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of developing, maintaining, and deploying robotic systems capable of safe medication administration would far exceed the loaded wage of an ophthalmic medical technologist performing this routine clinical task.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical task, so cost comparison favors the human by default since AI cannot deliver the output at all.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product can physically administer topical or oral medications to patients. This task involves robot manipulation in clinical settings, which remains in research phases and is not deployed in ophthalmic practices at scale.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product administers topical or oral medications to patients; this remains a purely human clinical action requiring physical presence.

Perform advanced ophthalmic procedures, including electrophysiological, electrophysical, or microbial procedures.

0

CI 0–0 · exposure 0 · augmentation 38 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare, especially specialized ophthalmology, adopts automation slowly. These clinical procedures remain largely manual and technician-dependent; no broad adoption of autonomous AI systems for advanced ophthalmic testing is evident in the field.
Sector adoption velocityclaude-sonnet-51/5Direct patient-contact clinical procedures in ophthalmology show minimal AI adoption for the physical task itself, despite AI growth in image interpretation elsewhere in the field.
Augmentation potentialclaude-haiku-4-5-202510012/5Some software assists with data interpretation and recording (e.g., analysis of electrophysiology waveforms), but current AI provides minimal active assistance during the procedures themselves. The task is primarily hands-on execution, where augmentation opportunities are limited.
Augmentation potentialclaude-sonnet-53/5AI can assist with interpreting electrophysiological waveform data or flagging abnormal microbial culture results, aiding the technologist's analysis even though it cannot perform the procedure itself.
Task automatabilityclaude-haiku-4-5-202510011/5These procedures require hands-on manipulation of specialized equipment, precise physical positioning of patients, and real-time clinical judgment that current AI cannot execute. The electrophysiological and microbial procedures involve sterile technique, patient interaction, and parameter adjustment based on live biometric feedback.
Task automatabilityclaude-sonnet-51/5This is a hands-on clinical procedure involving direct patient contact with specialized diagnostic equipment (e.g., ERG, EOG, microbial cultures); no current AI system can physically perform these procedures.
Adoption barriersclaude-haiku-4-5-202510015/5Ophthalmic procedures are tightly regulated; performing advanced electrophysiological and microbial tests requires state licensure or certification and direct patient contact, and liability for error (e.g., contamination, misdiagnosis) falls on the credentialed technologist, creating hard legal barriers to automation.
Adoption barriersclaude-sonnet-55/5These are licensed, hands-on clinical procedures requiring certified technologists working under physician supervision, with strict regulatory and patient-safety requirements precluding automation of the physical task.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot yet perform these procedures independently, so there is no meaningful cost comparison. The equipment, calibration, and oversight required would exceed the cost of a trained ophthalmic technologist.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical procedure at all, so any AI cost is essentially wasted compared to the human technologist who is required to perform it.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system today can autonomously perform electrophysiological testing, electrophysical procedures, or microbial cultures in ophthalmic settings. These remain technician-performed tasks requiring certified personnel and cannot be substituted by current AI or robotic agents at production scale.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs the physical execution of electrophysiological or microbial ophthalmic testing; this remains entirely research-stage or nonexistent for the procedural component.

Related occupations — Healthcare Practitioners & Technical

How to read this

A high substitution score does not mean this job disappears — it means a large share of its current tasks face replacement pressure, so the mix of tasks is likely to change. High augmentation alongside substitution typically means the occupation reorganizes around the protected tasks. Wide confidence intervals mean the rater panel disagreed: treat those scores as open questions, not verdicts.

What would change this score

New model capabilities (automatability, feasibility), falling inference costs (cost ratio), regulation and licensing shifts (barriers), and measured sector adoption (velocity) all re-enter at every index release. Each release is recomputed, versioned and kept queryable — scores are claims with a date on them, not permanent labels.