Neurologists

29-1217.00
Median wage $248,560/yr10,590 employed (US)Rank #734 of 923 scored · top 80% by substitution

Diagnose, manage, and treat disorders and diseases of the brain, spinal cord, and peripheral nerves, with a primarily nonsurgical focus.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution18
Exposure19
Augmentation62

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

24 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

0%

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%17

panel mean rating 1.7/5 → substitution pressure 17/100

Technical feasibility todayw 20%21

panel mean rating 1.9/5 → substitution pressure 21/100

Cost vs. human wagew 15%20

panel mean rating 1.8/5 → substitution pressure 20/100

Adoption barriersw 20%inverted — strong barriers lower the score9

panel mean rating 4.6/5 (barrier strength) → substitution pressure 9/100

Sector adoption velocityw 10%25

panel mean rating 2.0/5 → substitution pressure 25/100

Task breakdown (24 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.

Order or interpret results of laboratory analyses of patients' blood or cerebrospinal fluid.

45

CI 20–70 · exposure 58 · augmentation 88 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare is adopting clinical AI slowly relative to tech/finance sectors; pilot programs are common but production deployment of autonomous lab interpretation remains limited. Legacy EHR integration, liability concerns, and physician gatekeeping slow velocity even where technical feasibility is high.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialized clinical diagnostics like neurology, adopts AI cautiously due to regulatory, liability, and workflow integration challenges, with pilots more common than production use.
Augmentation potentialclaude-haiku-4-5-202510015/5AI augmentation is already transforming this task: neurologists use AI-assisted interpretation tools to flag abnormalities, suggest differential diagnoses, and prioritize critical results, substantially raising diagnostic speed and confidence while the physician retains final authority.
Augmentation potentialclaude-sonnet-54/5AI-assisted lab analysis and pattern recognition (e.g., flagging abnormal CSF markers, correlating with differential diagnoses) can meaningfully speed up a neurologist's interpretive workflow while the physician retains final judgment.
Task automatabilityclaude-haiku-4-5-202510015/5AI systems can now interpret laboratory results (blood work, CSF panels) with performance matching or exceeding human radiologists and pathologists in many domains, using standard ML classifiers and large language models. Structured lab data is highly amenable to automation with ≥50% time savings at equal quality for interpretation tasks.
Task automatabilityclaude-sonnet-52/5AI can flag abnormal values and suggest differentials, but integrating lab results with clinical context to guide diagnosis/treatment for neurological conditions still requires physician judgment and legal accountability, so full end-to-end automation isn't yet achievable at equal quality.
Adoption barriersclaude-haiku-4-5-202510015/5Legal and regulatory barriers are substantial: physicians must order tests and remain liable for clinical decisions; lab results require credentialed interpretation in most jurisdictions, and malpractice risk is high. Licensing requirements and human sign-off obligations create hard barriers to autonomous deployment.
Adoption barriersclaude-sonnet-55/5Ordering and interpreting diagnostic labs for treatment decisions is a licensed medical act requiring physician sign-off, with high liability exposure for missed or misinterpreted results.
Cost vs. human wageclaude-haiku-4-5-202510015/5AI inference on structured lab results is extremely cheap (<$0.01 per interpretation) compared to neurologist time ($50–$150+ per task), yielding a >100× cost advantage even accounting for integration and oversight.
Cost vs. human wageclaude-sonnet-52/5AI tools can cheaply flag abnormal lab values, but the physician's interpretive and diagnostic work still requires costly oversight, keeping overall cost savings modest compared to full physician cost.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed AI products (e.g., clinical decision support systems, LLM-based lab interpretation tools) exist in production at major healthcare systems and commercial EHR platforms, though adoption remains uneven and many still require physician review before action. Material error rates on edge cases and liability concerns keep the rating from 5.
Technical feasibility todayclaude-sonnet-52/5Some clinical decision-support and lab-interpretation tools exist, but no deployed product independently and reliably interprets CSF/blood analyses for neurology-specific diagnosis at scale in production.

Prepare, maintain, or review records that include patients' histories, neurological examination findings, treatment plans, or outcomes.

39

CI 32–45 · exposure 42 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Hospitals and large practices are piloting AI-assisted documentation (ambient scribing), but adoption remains uneven; many neurologists still manually dictate or type notes, and regulatory/liability concerns slow broad deployment.
Sector adoption velocityclaude-sonnet-53/5Healthcare adoption of AI documentation tools is growing quickly in some systems but overall industry adoption remains uneven and cautious due to compliance and EHR integration challenges.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist by transcribing spoken notes, suggesting structured formats, auto-populating standard fields, and flagging missing data, materially reducing documentation burden while the neurologist retains clinical authority over the record.
Augmentation potentialclaude-sonnet-54/5AI scribes and summarization tools meaningfully reduce documentation burden and speed up note-taking and record review while the neurologist retains final oversight.
Task automatabilityclaude-haiku-4-5-202510012/5Current AI can auto-populate templated sections and extract data from images/reports, but cannot independently synthesize clinical histories, examination findings, and treatment plans at the quality and legal standard required. A human neurologist must still review, validate, and sign the record.
Task automatabilityclaude-sonnet-53/5AI can draft clinical documentation from dictation/transcripts and summarize histories, but review/finalization of records requires physician verification for accuracy and liability, so only partial automation meets the 50% bar.
Adoption barriersclaude-haiku-4-5-202510014/5Medical records are legally binding and subject to regulatory requirements (HIPAA, CMS, state medical board rules); the neurologist has fiduciary and legal responsibility for accuracy and completeness, creating a hard barrier to full automation without physician sign-off.
Adoption barriersclaude-sonnet-54/5Medical records require physician attestation and are subject to legal/regulatory documentation standards (e.g., for billing, malpractice, HIPAA), so a licensed neurologist must ultimately review and sign off.
Cost vs. human wageclaude-haiku-4-5-202510012/5While ambient scribing reduces transcription time, the neurologist must still spend significant time reviewing, editing, and ensuring accuracy; total cost savings are modest compared to the neurologist's wage, and oversight overhead can be substantial.
Cost vs. human wageclaude-sonnet-53/5AI scribe/documentation tools are cheaper than transcription staff but still require licensing fees plus physician review time, making net savings moderate rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510013/5EHR systems with AI-assisted documentation (e.g., ambient scribing, note templates) exist in production, but they require substantial human review and correction; error rates in capturing nuanced neurological findings remain material.
Technical feasibility todayclaude-sonnet-53/5Ambient AI scribes and clinical documentation tools are deployed in production at hospitals and clinics, but accuracy issues and need for physician review remain, especially for specialized neurological findings.

Interpret the results of neuroimaging studies, such as Magnetic Resonance Imaging (MRI), Single Photon Emission Computed Tomography (SPECT), and Positron Emission Tomography (PET) scans.

33

CI 20–46 · exposure 38 · augmentation 75 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Despite technical maturity, adoption in clinical neurology remains slow and primarily pilot-stage. Many institutions lack integration infrastructure, radiologists and neurologists show variable comfort with AI, and reimbursement incentives do not yet strongly favor automation, keeping this in the laggard-to-middling adoption range.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall is a laggard sector for full AI decision autonomy due to regulation and liability, though pilot programs for imaging AI in radiology/neurology are growing; production-level autonomous interpretation is still rare.
Augmentation potentialclaude-haiku-4-5-202510014/5AI demonstrates strong augmentation potential: it can highlight regions of interest, flag likely abnormalities, and accelerate triage, allowing neurologists to focus on complex cases and clinical correlation. Studies show AI assistance speeds interpretation and reduces false negatives, making it a productivity multiplier when well-integrated into workflow.
Augmentation potentialclaude-sonnet-54/5AI-based image analysis tools meaningfully help detect and quantify lesions, atrophy patterns, or perfusion abnormalities, speeding preliminary assessment while the neurologist retains final interpretive responsibility.
Task automatabilityclaude-haiku-4-5-202510013/5AI can detect certain abnormalities in neuroimaging with performance approaching human radiologists on specific benchmarks (e.g., stroke, tumor detection), but clinical interpretation requires synthesis with patient history, symptom severity, and differential diagnosis—tasks that typically demand human judgment. Current systems handle pattern recognition well but fall short on end-to-end clinical decision-making, likely saving 30–50% of time with oversight.
Task automatabilityclaude-sonnet-52/5AI can flag anomalies and assist quantification in neuroimaging, but full end-to-end clinical interpretation integrating patient history, differential diagnosis, and treatment implications still requires physician judgment, so the 50% time-saving-at-equal-quality bar is not met for the whole task.
Adoption barriersclaude-haiku-4-5-202510014/5Neuroimaging interpretation carries significant liability and error-cost asymmetry: misdiagnosis can lead to serious patient harm, creating strong organizational and regulatory pressure to retain physician sign-off. Malpractice liability and regulatory expectations (FDA 510(k) clearance for diagnostic claims) impose meaningful friction on fully autonomous deployment.
Adoption barriersclaude-sonnet-55/5Interpreting diagnostic neuroimaging for clinical decision-making legally requires physician sign-off, with high liability exposure for misdiagnosis, and regulatory frameworks (FDA, medical board licensing) strictly govern this activity.
Cost vs. human wageclaude-haiku-4-5-202510014/5Inference cost for a single neuroimaging AI run is typically low ($1–10), and integration into PACS systems is mature. Accounting for oversight and validation, the per-study cost is substantially below the loaded wage of a neurologist ($50–150/hour), especially in high-volume settings.
Cost vs. human wageclaude-sonnet-52/5Specialized imaging AI tools require licensing, integration with PACS, and ongoing oversight, so while inference cost is low, the total cost of validated, compliant deployment is not dramatically cheaper than incremental physician time for this narrow task.
Technical feasibility todayclaude-haiku-4-5-202510013/5FDA-cleared AI products for neuroimaging analysis exist (e.g., stroke detection, lesion segmentation) and are deployed in some healthcare systems, but adoption remains narrow and error rates on edge cases remain material. Most neurologists still perform primary interpretation themselves, treating AI as a second reader rather than autonomous system.
Technical feasibility todayclaude-sonnet-52/5Some FDA-cleared tools exist for stroke triage or volumetric analysis (e.g., lesion detection, brain volume quantification), but these are narrow adjuncts used alongside radiologist/neurologist review, not standalone interpretation systems deployed broadly for neurologist-level reads.

Interview patients to obtain information, such as complaints, symptoms, medical histories, and family histories.

25

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

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of autonomous patient interview AI remains slow in neurology practices; most deployments are narrow intake forms or symptom screeners under human review rather than AI-driven interviews replacing clinician judgment.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialized fields like neurology, has been slower to adopt AI-driven patient interviewing compared to information/finance sectors, with pilots more common than production use.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by suggesting follow-up questions, flagging red-flag symptoms from transcribed interviews, and organizing patient-reported history into structured summaries, improving the neurologist's efficiency and recall without replacing their clinical judgment.
Augmentation potentialclaude-sonnet-54/5AI-powered intake forms, ambient scribes, and pre-visit symptom questionnaires can meaningfully speed up and structure history-gathering, letting neurologists focus on interpretation and follow-up during the actual encounter.
Task automatabilityclaude-haiku-4-5-202510012/5AI can conduct structured symptom collection and extract key medical history details, but neurological interviews require nuanced judgment about symptom severity, temporal patterns, and family risk factors that demand human interpretation and follow-up probing that current systems cannot reliably perform end-to-end.
Task automatabilityclaude-sonnet-52/5While AI can conduct structured symptom intake via chatbots, the nuanced, adaptive clinical interviewing that neurologists perform—probing subtle neurological symptoms, reading non-verbal cues, and building rapport—still requires substantial human judgment, so full end-to-end automation with equal quality is not yet achievable.
Adoption barriersclaude-haiku-4-5-202510014/5Strong regulatory and liability barriers exist: clinical interviews are part of the standard-of-care diagnostic process, physicians face malpractice exposure if AI misses critical history, and medical licensing requirements typically mandate that the clinician personally conduct or closely supervise patient assessment.
Adoption barriersclaude-sonnet-54/5Medical licensing, liability for missed diagnoses, and the need for clinical judgment in interpreting neurological symptoms create strong barriers requiring physician involvement and sign-off.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted intake screening may reduce some clerical time, but the neurologist's involvement in the substantive interview is essential; the all-in cost of deploying AI oversight for interview automation often exceeds the direct savings from partial automation.
Cost vs. human wageclaude-sonnet-52/5AI-assisted intake forms are cheap to run, but because a licensed physician must still verify and expand on the history for diagnostic purposes, the net cost saving versus the physician's time is modest rather than order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbots and structured intake forms exist for preliminary patient data collection, but deployed medical systems still require human clinicians to validate responses, probe inconsistencies, and make diagnostic sense of patient narratives—no production system performs this independently at clinical reliability.
Technical feasibility todayclaude-sonnet-52/5Some AI intake tools and symptom checkers exist in production for general triage, but reliable, production-grade AI systems that fully replace neurologist-led history taking are not deployed at scale in real clinical settings.

Provide training to medical students or staff members.

25

CI 25–25 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Medical education adoption of AI is still largely pilot-stage (supplementary tools for preparation, note-taking), with little displacement of human instruction. Medical schools remain conservative about delegating core training to AI, prioritizing human mentorship and accreditation compliance.
Sector adoption velocityclaude-sonnet-52/5Academic medicine adopts AI tools cautiously; pilots for AI-assisted teaching materials exist but broad production deployment for training staff is still limited.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can augment training by generating illustrative cases, creating practice problems, summarizing literature, and personalizing review materials—genuinely useful for preparing sessions and student self-study. However, it cannot replace the Socratic dialogue, clinical reasoning modeling, and real-time assessment that neurologists provide during live instruction.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by generating case studies, quizzes, explanations, and summarizing literature, boosting efficiency of the training process while the physician remains the teacher.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with preparing educational materials, generating practice questions, and drafting lecture content, but cannot replicate the interactive, adaptive instruction and real-time feedback adjustment that characterizes effective medical training. The task requires modeling student understanding and adjusting explanations, which remains beyond current AI.
Task automatabilityclaude-sonnet-52/5Training involves synthesizing clinical knowledge, modeling bedside reasoning, mentorship, and hands-on demonstration that current AI cannot deliver end-to-end, though it can help prepare materials.'
Adoption barriersclaude-haiku-4-5-202510014/5Medical education is heavily regulated (accreditation standards, licensing boards, curriculum requirements), and there is a strong institutional and professional norm that expert clinicians must directly mentor trainees for credibility and accountability. Regulatory bodies often mandate human instructor involvement in medical training programs.
Adoption barriersclaude-sonnet-54/5Medical education and supervision typically require licensed physician oversight and institutional accreditation standards, creating strong structural barriers to full AI substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5Developing, customizing, and overseeing AI-based medical education platforms still involves substantial setup and human review costs. A neurologist's salary amortized across training hours remains competitive with the total cost of AI infrastructure, content curation, and human oversight required.
Cost vs. human wageclaude-sonnet-52/5Human clinical educators command high wages, but AI cannot yet fully replace supervised teaching, so cost comparisons only apply to partial content-generation tasks, not the full training role.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI tutoring systems and educational content generators exist, they are not yet widely deployed as replacements for medical training delivery by neurologists. Products like ChatGPT can generate study materials, but no mature system reliably delivers comprehensive medical education with the clinical judgment and personalization that human neurologists provide.
Technical feasibility todayclaude-sonnet-52/5AI tutoring products and case-based learning tools exist but are not deployed as substitutes for attending-led clinical teaching in production settings.

Perform or interpret the outcomes of procedures or diagnostic tests, such as lumbar punctures, electroencephalography, electromyography, and nerve conduction velocity tests.

23

CI 20–25 · exposure 25 · augmentation 75 · 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 limited and largely experimental; most neurology practices are still in pilot or selective-adoption phases for AI-assisted interpretation, with slow uptake driven by liability concerns, regulatory uncertainty, and the need for validation in diverse clinical populations.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialized neurology diagnostics, adopts AI tools slowly due to regulatory approval processes, liability concerns, and clinical validation requirements.
Augmentation potentialclaude-haiku-4-5-202510014/5AI demonstrates strong augmentation potential: machine learning can rapidly flag abnormalities in EEG, EMG, and NCS data, surface patterns a human might miss, and reduce cognitive load, enabling neurologists to focus on clinical integration and patient communication while AI handles signal-level pattern detection.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by pre-screening EEG patterns, flagging abnormalities in nerve conduction data, and speeding up initial data review, improving physician efficiency.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in analyzing some diagnostic test outputs (e.g., EEG pattern recognition, EMG signal processing), the task requires integrating multiple complex test modalities, clinical context, and patient-specific factors into a coherent diagnostic interpretation. Current AI systems lack the end-to-end capability to replace the neurologist's synthesis across diverse procedures and clinical judgment needed to meet the 50% time-saving threshold consistently.
Task automatabilityclaude-sonnet-52/5Physical procedures like lumbar punctures require hands-on skill AI cannot perform; AI can assist interpretation of EEG/EMG/NCV signals but final clinical interpretation and integration with patient context still requires physician judgment.'
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and regulatory barriers exist: neurologists are licensed medical professionals whose diagnostic interpretation is legally required and carries liability; malpractice risk and clinical responsibility cannot be fully delegated, and FDA/medical device oversight restricts autonomous diagnostic automation in clinical practice.
Adoption barriersclaude-sonnet-55/5Performing invasive procedures and legally interpreting diagnostic neurological tests requires a licensed physician, with high liability and regulatory requirements for sign-off.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI diagnostic assistance (software licensing, infrastructure, quality assurance) remains costly relative to the time neurologists save on routine interpretation, and oversight labor must still be factored in; AI has not yet achieved cost parity with human interpretation, let alone a significant advantage.
Cost vs. human wageclaude-sonnet-52/5AI diagnostic software has licensing and integration costs plus mandatory physician oversight, so total cost is not dramatically cheaper than a neurologist's time for these specialized tests.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed AI products exist for narrow modalities (e.g., EEG analysis, NCS data interpretation), but they are typically narrow-scope and require specialist oversight; no mature, production-grade system reliably performs the full range of procedure interpretation across all test types at scale in real clinical settings without significant human review.
Technical feasibility todayclaude-sonnet-52/5Some AI-assisted EEG seizure detection and signal analysis tools exist in production, but comprehensive interpretation of EMG/NCV and procedural performance remain physician-driven with no mature end-to-end products.

Develop treatment plans based on diagnoses and on evaluation of factors, such as age and general health, or procedural risks and costs.

23

CI 20–25 · exposure 25 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and neurology practices remain conservative in automation of clinical decision-making; most adoption is in narrow decision-support roles (e.g., imaging interpretation aids), not autonomous treatment planning. Adoption remains pilot-stage in most institutions.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialty medicine like neurology, has historically slow AI adoption for core clinical decision-making due to regulatory caution and liability concerns, despite growing pilots.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully augment treatment planning by rapidly retrieving relevant clinical guidelines, summarizing patient data, flagging drug interactions, and suggesting options—keeping the neurologist in a supervisory role and substantially accelerating literature and data synthesis that would otherwise consume significant time.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist neurologists by synthesizing research, flagging drug interactions, summarizing patient history, and suggesting evidence-based options, improving efficiency while the physician retains final decision authority.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in aggregating diagnostic data and suggesting evidence-based treatment options, the task requires synthesizing complex medical history, patient-specific factors (age, comorbidities, preferences), and balancing risk-benefit tradeoffs—judgment calls that current AI systems cannot reliably perform end-to-end at the quality and accountability expected for neurological treatment planning.
Task automatabilityclaude-sonnet-52/5Treatment planning requires integrating nuanced clinical judgment, patient values, comorbidities, and legal accountability that current AI cannot fully replicate end-to-end, though it can assist with parts like literature review or drug interaction checks.
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and regulatory barriers: neurologists are licensed physicians who must take responsibility for treatment decisions; malpractice liability, standard of care expectations, and medical licensing law require a physician to author and sign off on treatment plans, not an automated system.
Adoption barriersclaude-sonnet-55/5Treatment plans must be authored and legally signed off by a licensed physician; malpractice liability and regulatory requirements make this a hard barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5Neurologist-authored treatment plans reflect their expertise and liability; AI tools may reduce documentation overhead modestly, but the incremental cost savings do not yet approach an order of magnitude given the need for physician review and modification of AI suggestions.
Cost vs. human wageclaude-sonnet-52/5AI tools can cheaply generate draft suggestions, but the need for physician review, liability coverage, and integration into EHR workflows keeps the effective cost comparable to or only modestly below physician time for this specific judgment task.
Technical feasibility todayclaude-haiku-4-5-202510012/5Clinical decision support tools exist and can flag relevant guidelines and literature, but no deployed product independently generates and justifies complete treatment plans that neurologists rely on without substantial review. AI-generated plans lack the legal and clinical standing to replace physician authorship.
Technical feasibility todayclaude-sonnet-52/5Clinical decision-support tools exist and are used for suggestions, but no deployed product autonomously creates neurological treatment plans reliably at scale without physician authorship and oversight.

Counsel patients or others on the background of neurological disorders including risk factors, or genetic or environmental concerns.

23

CI 20–25 · exposure 25 · augmentation 63 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare, while increasingly digital, remains highly regulated and conservative on patient-facing decision-making. Neurology practices have not meaningfully adopted autonomous AI counseling systems; adoption is limited to decision-support and scheduling tools, not patient communication automation.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially direct patient counseling, remains a slow-adopting sector for AI due to regulatory, liability, and trust concerns despite growing pilot programs.
Augmentation potentialclaude-haiku-4-5-202510013/5AI tools can assist neurologists by pre-drafting patient education materials, summarizing family history and genetic risk data, and organizing evidence on environmental factors—raising their preparation efficiency. However, the live counseling interaction itself resists augmentation unless the AI is purely a passive information-lookup tool, limiting transformative productivity gain.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist neurologists by summarizing genetic/environmental risk literature, drafting patient education materials, and preparing talking points, improving counseling efficiency while the physician remains the decision-maker.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can retrieve and summarize information about neurological disorders, risk factors, and genetic/environmental concerns, counseling is inherently interactive and requires reading patient affect, adjusting explanations based on comprehension, and addressing emotional concerns—capabilities that current AI systems lack reliably. A minority of the task (information provision) can be automated, but the core counseling function cannot be displaced without human judgment.
Task automatabilityclaude-sonnet-52/5AI can generate accurate general information about neurological disorder risk factors and genetics, but personalized counseling requires integrating patient history, emotional context, and nuanced judgment that current systems cannot reliably replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Professional liability, regulatory oversight of medical advice, and the medico-legal requirement that information be delivered by or under direct supervision of a licensed physician create hard barriers. Medical licensing boards and malpractice law strongly disfavor autonomous AI delivery of patient counseling without physician sign-off.
Adoption barriersclaude-sonnet-55/5Counseling patients on medical risk and genetic concerns is a licensed medical activity with significant liability exposure, requiring a physician's judgment and legal accountability.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI system deployment (fine-tuned models, integration, compliance oversight, liability management) remains expensive relative to lower-wage staff roles (nurses, counselors) who might assist, and far more expensive than the neurologist's time it would need to replace or supervise. Integration costs and error-checking overhead favor human counseling on this task.
Cost vs. human wageclaude-sonnet-52/5While generating informational text is cheap, the physician oversight, liability review, and personalization required keep overall cost comparable to or only modestly cheaper than physician time.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs patient counseling on complex neurological topics end-to-end; systems exist for information retrieval and decision support but not autonomous patient counseling. Pilots and chatbots exist but show material error rates and cannot handle the conversational complexity, emotional nuance, and medico-legal responsibility required.
Technical feasibility todayclaude-sonnet-52/5Chatbots and clinical decision support tools can provide informational content, but no deployed product independently conducts patient counseling on neurological risk factors in production clinical settings.

Order supportive care services, such as physical therapy, specialized nursing care, and social services.

23

CI 20–25 · exposure 25 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Neurologists and hospital systems are in the early/pilot phase of AI-assisted ordering workflows; most still rely on manual order entry or basic EHR templates rather than AI agents, reflecting slow adoption of automation for high-liability clinical decisions.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially physician clinical ordering workflows, adopts AI decision support slowly due to regulatory, liability, and EHR integration constraints, with most use still in pilot or advisory phases.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by suggesting appropriate services based on diagnosis or guidelines, surfacing available providers, or auto-populating order fields—productivity gains are real but moderate, as the neurologist retains full clinical judgment and the ordering task is already relatively quick.
Augmentation potentialclaude-sonnet-54/5AI-driven clinical decision support and EHR order-set suggestions can meaningfully speed up identification and drafting of supportive care referrals, improving physician efficiency while they retain final authority.
Task automatabilityclaude-haiku-4-5-202510012/5Ordering supportive care requires clinical judgment about patient needs, coordination with multiple services, and documentation in medical systems—tasks that demand human decision-making. While AI could assist in drafting orders or retrieving guidelines, the clinician must evaluate the individual patient, determine necessity, and take legal responsibility for the order, preventing full end-to-end automation.
Task automatabilityclaude-sonnet-52/5AI can help identify appropriate referrals and draft orders, but the physician must review the patient's clinical status and formally authorize the order, limiting full end-to-end automation.dit rate as low-moderate.
Adoption barriersclaude-haiku-4-5-202510014/5Legal and regulatory barriers are substantial: the neurologist must personally authorize the order and is liable for its appropriateness; medical licensing and standard of care requirements mean a licensed physician must make and sign the final decision, creating a hard barrier to full automation.
Adoption barriersclaude-sonnet-55/5Ordering medical services requires a licensed physician's authorization; this is a legally and clinically mandated human sign-off task with high liability exposure.
Cost vs. human wageclaude-haiku-4-5-202510012/5The neurologist's loaded cost (salary, benefits, overhead) remains moderate to high for this relatively straightforward coordination task, while AI tooling for order drafting still requires human oversight and integration costs that limit overall savings; the ratio is not yet favorable to automation.
Cost vs. human wageclaude-sonnet-52/5While drafting suggestions is cheap, the requirement for physician review and liability oversight keeps the effective cost comparable to human-driven ordering rather than dramatically cheaper.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed clinical AI systems autonomously order supportive care services reliably in production; the task requires integration with EHRs, knowledge of insurance/availability, and clinical reasoning that current tools do not handle at scale. AI-assisted draft generation exists in research, but not production ordering systems.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools can suggest care plans and order sets, but no deployed product independently orders supportive care services without physician sign-off in production settings.

Participate in continuing education activities to maintain and expand competence.

22

CI 16–28 · exposure 17 · augmentation 63 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While healthcare sectors are adopting AI, continuing education itself remains a regulatory and professional requirement that must be fulfilled by the licensed individual, limiting substitution velocity.
Sector adoption velocityclaude-sonnet-53/5Healthcare and medical education are moderately digitized with growing use of AI-assisted learning tools and decision support, but adoption of AI specifically for CME activities is still nascent.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by summarizing literature, identifying relevant educational content, organizing learning materials, and generating study aids, thereby accelerating the neurologist's learning process.
Augmentation potentialclaude-sonnet-54/5AI can significantly enhance CME by summarizing new research, generating personalized quizzes, and flagging relevant case studies, substantially boosting learning efficiency while the physician remains the learner.
Task automatabilityclaude-haiku-4-5-202510011/5Continuing education inherently requires human judgment, reflection, and integration of new knowledge with existing expertise. AI cannot autonomously decide what competencies a neurologist needs to develop or meaningfully participate in reflective learning activities.
Task automatabilityclaude-sonnet-52/5AI can help curate and summarize medical literature and CME content, but the actual learning, certification, and competence-building process requires human cognitive engagement and cannot be fully offloaded to AI.'
Adoption barriersclaude-haiku-4-5-202510014/5Most licensing boards and professional bodies explicitly require physicians to personally participate in and log continuing medical education hours; delegation or substitution with AI is not permitted for compliance with licensure maintenance.
Adoption barriersclaude-sonnet-54/5Board certification and licensing bodies mandate that the licensed physician personally complete and attest to continuing education, creating a strong regulatory/professional barrier to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for content generation or curation are inexpensive, but they represent only a small fraction of the total cost of continuing education, which is dominated by the neurologist's time investment and course/license fees.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate summaries or practice questions, but the human must still spend time actually learning and completing accredited CME, so cost savings are partial at best.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can help generate summaries of medical literature or identify relevant courses, but cannot independently participate in or complete continuing education activities that require human engagement, assessment, and certification.
Technical feasibility todayclaude-sonnet-52/5AI tools like literature summarizers and clinical decision support exist and are used to support learning, but no deployed product independently completes continuing education requirements for a physician.

Participate in neuroscience research activities.

22

CI 11–32 · exposure 13 · augmentation 63 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While AI adoption is growing in academic and clinical neuroscience, actual displacement of research participation remains limited. Most adoption is experimental and targeted at narrow subtasks; full research workflow automation is not a widespread pattern even in well-resourced institutions.
Sector adoption velocityclaude-sonnet-53/5Academic medicine and research institutions are adopting AI tools for literature synthesis and data analysis, but adoption of AI as an active research participant remains limited and exploratory.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist neurologists in research by accelerating literature synthesis, automating routine data processing, and suggesting statistical approaches, but the human researcher remains essential for design, interpretation, and judgment.
Augmentation potentialclaude-sonnet-54/5AI substantially assists with literature reviews, statistical analysis, data visualization, and hypothesis generation, meaningfully boosting researcher productivity while humans retain control of design and interpretation.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with literature review, data analysis, and hypothesis generation, neurological research involves experimental design, complex troubleshooting, and creative interpretation that require human expertise. Current AI falls far short of end-to-end research participation with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-51/5Neuroscience research involves hypothesis generation, experimental design, hands-on data collection, and creative interpretation that current AI cannot autonomously perform end-to-end at equal quality.
Adoption barriersclaude-haiku-4-5-202510014/5Research participation requires accountability, institutional review, publication credit, and human judgment on methodology and interpretation. IRB oversight, authorship norms, and institutional liability mean that a licensed neurologist must ultimately direct and sign off on research activities.
Adoption barriersclaude-sonnet-53/5No strict licensing requirement to 'do research,' but institutional review boards, funding bodies, and publication standards require credentialed human oversight and accountability.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools (subscriptions, APIs, computational resources) are relatively inexpensive in isolation, but the total integration cost into a research workflow, combined with required human oversight and validation, approaches or exceeds the cost of direct researcher time for this complex activity.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply assist with literature review or data analysis, but the overall research process still requires expensive expert oversight, lab work, and judgment, keeping costs comparable to human-led research.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed AI tools can help with specific research subtasks (literature mining, statistical analysis, data visualization) but no integrated product reliably performs 'participation in neuroscience research' as a whole. Research-stage systems may show promise but are not production-ready in neuroscience labs.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently conducts neuroscience research; existing tools are research-stage aids used by human scientists, not autonomous researchers.

Diagnose neurological conditions based on interpretation of examination findings, histories, or test results.

21

CI 16–25 · exposure 30 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Neurologists and hospital systems adopt AI as a supportive tool slowly; most deployments remain in pilot or niche imaging contexts. The specialty is small relative to primary care, clinical adoption lags, and physicians exhibit high skepticism toward delegating diagnostic reasoning to black-box systems.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialty diagnostic fields like neurology, has historically slow, cautious AI adoption due to regulatory, liability, and workflow integration hurdles.
Augmentation potentialclaude-haiku-4-5-202510014/5AI significantly augments neurologist productivity by rapidly flagging abnormalities on imaging, summarizing literature on rare conditions, and organizing test results—enabling faster, more confident diagnosis without replacing the physician's role. Large language models and structured analytics measurably improve workflow efficiency and decision quality.
Augmentation potentialclaude-sonnet-54/5AI decision-support and imaging analysis tools meaningfully help neurologists narrow differentials and flag findings faster, enhancing productivity while the physician retains diagnostic authority.
Task automatabilityclaude-haiku-4-5-202510012/5While AI systems can assist in pattern recognition from imaging and lab results, neurological diagnosis fundamentally requires synthesis of complex patient histories, physical examination findings, and contextual factors that current AI struggles to integrate end-to-end. The task demands real-time clinical judgment about ambiguous presentations, ruling out mimics, and integrating non-standardized examination data—well below the 50% time-saving threshold for autonomous performance.
Task automatabilityclaude-sonnet-52/5Diagnosis requires integrating nuanced clinical exam findings, patient history, and judgment about differential diagnosis; AI can assist with pattern recognition (e.g., imaging) but cannot reliably perform the full diagnostic synthesis end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5Legal and regulatory barriers are high: diagnosis is a licensed medical act that typically requires a physician to examine the patient, sign the diagnostic impression, and accept liability. Malpractice exposure, FDA oversight of clinical decision-support tools, and medical licensing laws create hard barriers to fully automated deployment.
Adoption barriersclaude-sonnet-55/5Diagnosis is a licensed medical act requiring a physician's legal responsibility and accountability, making this a hard regulatory and liability barrier.
Cost vs. human wageclaude-haiku-4-5-202510011/5Current neurologist salaries are high (~$250k+ loaded), and comprehensive diagnostic AI systems require expensive imaging infrastructure, integration with EMRs, and ongoing validation. The total cost of AI-assisted diagnosis—including oversight and liability—remains substantially higher than the alternative of brief human consultation for most cases.
Cost vs. human wageclaude-sonnet-52/5AI diagnostic support tools carry integration, validation, and liability review costs that remain high relative to the marginal value they add versus physician time already required for legal sign-off.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI products exist for narrow domains (e.g., EEG interpretation, MS lesion detection on MRI) with demonstrated performance in controlled settings, but no deployed system reliably performs the full diagnostic task across the range of neurological conditions clinicians encounter. Error rates on complex cases and inability to handle novel presentations limit production deployment to decision-support roles.
Technical feasibility todayclaude-sonnet-52/5Some deployed tools aid in radiology/EEG interpretation, but no product independently and reliably diagnoses neurological conditions from full clinical context in production settings.

Advise other physicians on the treatment of neurological problems.

16

CI 6–25 · exposure 13 · augmentation 63 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Despite digital health advancement, neurological consultation adoption remains cautious; AI is used in research and support roles rather than autonomous advice-giving, with slow production deployment in actual clinical practice.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialist consultation, adopts AI slowly due to regulatory, liability, and trust constraints despite some diagnostic-support pilots.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist neurologists by rapidly synthesizing literature, flagging differential diagnoses, and organizing clinical data, thereby improving productivity on diagnostic reasoning while the physician retains decision authority.
Augmentation potentialclaude-sonnet-54/5AI tools (e.g., clinical decision support, literature synthesis, imaging analysis) can meaningfully help neurologists formulate advice faster and more thoroughly, though the human remains central.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with literature synthesis and differential diagnosis suggestion, but neurological consultation requires nuanced clinical judgment, integration of patient context, and accountability for treatment recommendations that current AI systems cannot reliably provide end-to-end.
Task automatabilityclaude-sonnet-51/5This requires clinical judgment integrating patient-specific context, examination findings, and specialist expertise built over years of training; current AI cannot reliably substitute for peer consultative advice at equal quality end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Medical licensing and malpractice liability create substantial barriers; neurological treatment advice must be signed by a licensed physician, and liability for adverse outcomes falls on the physician, not the AI system.
Adoption barriersclaude-sonnet-55/5Only a licensed neurologist can legally provide medical consultation and bear liability for advice given to another physician regarding patient treatment.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI consultation tools have modest inference costs but require significant human neurologist oversight, integration, and verification, making the all-in cost comparable to or higher than direct consultation.
Cost vs. human wageclaude-sonnet-52/5While AI query costs are cheap, the liability and oversight needed to trust AI-generated specialist advice erase most savings versus a neurologist's consult fee.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI systems can support diagnosis through literature retrieval and pattern matching, no deployed product reliably performs neurological consultation independently; clinical oversight and human neurologist review remain essential in all production settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently advises physicians on neurological case management in production; AI diagnostic aids exist but are not substitutes for specialist consults.

Identify and treat major neurological system diseases and disorders, such as central nervous system infection, cranio spinal trauma, dementia, and stroke.

15

CI 3–28 · exposure 17 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare AI adoption is accelerating in imaging and EHR analysis, but neurological practice remains conservative. Pilots are common (stroke networks, dementia screening), but production deployment as a replacement for physician evaluation is still rare and heavily regulated.
Sector adoption velocityclaude-sonnet-52/5Healthcare adoption of AI for diagnosis support is growing but remains cautious and heavily regulated, with treatment decisions still fully human-driven.
Augmentation potentialclaude-haiku-4-5-202510014/5AI substantially assists neurologists today: imaging analysis (MRI/CT interpretation support), EHR pattern recognition, literature synthesis, and patient risk stratification all improve diagnostic speed and confidence. These tools demonstrably raise neurologist productivity while the physician remains in full control.
Augmentation potentialclaude-sonnet-54/5AI tools like imaging analysis, clinical decision support, and literature synthesis meaningfully assist neurologists in diagnosis and treatment planning, improving speed and accuracy.
Task automatabilityclaude-haiku-4-5-202510012/5While AI excels at image analysis (CT/MRI) and pattern recognition in EHR data, neurological diagnosis requires integrating patient history, examination findings, imaging, labs, and clinical judgment. Current AI cannot reliably perform the full diagnostic and treatment-planning workflow end-to-end with 50% time savings at equal quality; human neurologists remain essential for differential diagnosis, treatment decisions, and patient communication.
Task automatabilityclaude-sonnet-51/5Diagnosing and treating major neurological diseases requires physical examination, patient interaction, invasive procedures, and clinical judgment that current AI cannot execute end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Neurologists are licensed physicians; diagnosis and treatment of serious CNS disease is legally restricted to qualified medical professionals. Liability, regulatory requirements (FDA oversight of diagnostic devices), informed consent, and the need for clinical judgment and accountability create hard barriers to autonomous automation.
Adoption barriersclaude-sonnet-55/5Medical licensure, malpractice liability, and legal requirements mandate that only credentialed physicians diagnose and treat these conditions.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for imaging or screening is cheap, but integration into clinical workflows, model maintenance, regulatory compliance, and required physician oversight add substantial cost. The all-in cost remains high relative to the neurologist's already-specialized labor, without eliminating the need for human expertise.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physician's role in this task, so there is no viable cost comparison—human neurologists remain necessary for treatment delivery.
Technical feasibility todayclaude-haiku-4-5-202510013/5AI diagnostic aids exist in production (e.g., stroke detection algorithms, dementia risk models) but are narrow in scope and deployed only as assistive tools, not autonomous decision-makers. No system reliably handles the full spectrum of major CNS disorders (infection, trauma, dementia, stroke) without neurologist review and override.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently diagnoses and treats neurological disorders; AI tools exist only as decision-support aids used by physicians, not as autonomous practitioners.

Communicate with other health care professionals regarding patients' conditions and care.

14

CI 3–25 · exposure 13 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While health systems are piloting AI documentation aids, actual adoption of AI for independent professional communication remains limited due to liability concerns, regulatory caution, and strong organizational commitment to human accountability in care coordination.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI unevenly and cautiously for clinical communication tasks, with slow integration into interprofessional workflows despite documentation-support pilots.
Augmentation potentialclaude-haiku-4-5-202510014/5Current AI tools meaningfully assist neurologists by drafting summaries, organizing patient data, and suggesting communication points, allowing physicians to communicate more efficiently while maintaining full clinical responsibility and decision authority.
Augmentation potentialclaude-sonnet-54/5AI tools (e.g., ambient scribes, summarization of records, drafting referral notes) can meaningfully speed up preparation and documentation supporting these communications, even though the human remains central.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can draft clinical notes and summarize patient data, the task requires nuanced communication addressing complex clinical judgment, accountability, and interprofessional collaboration that demands human oversight and decision-making authority.
Task automatabilityclaude-sonnet-51/5Interprofessional communication about complex, evolving patient conditions requires real-time clinical judgment, nuanced synthesis, and accountability that current AI cannot autonomously perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Legal liability, regulatory requirements (e.g., HIPAA, medical board standards), and the requirement that a licensed physician be responsible for clinical communication create substantial barriers to full automation of this task.
Adoption barriersclaude-sonnet-55/5Licensure, medical liability, and legal requirements for physician-to-physician communication about diagnosis and treatment make this a hard, protected barrier against automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted communication tools reduce drafting time but require significant neurologist review and validation, keeping total cost per communication interaction comparable to or only modestly below unassisted professional communication.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot substitute for the physician in this role, there is no viable AI-only cost comparison; any AI use is supplementary, not a replacement of the labor cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs end-to-end professional health care communication independently; existing systems assist with documentation and summaries but do not independently handle the clinical accountability and dynamic consultation exchanges required.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently communicates with other clinicians on behalf of a neurologist regarding patient care; this remains a human-led interaction in practice.

Inform patients or families of neurological diagnoses and prognoses, or benefits, risks and costs of various treatment plans.

12

CI 4–20 · exposure 13 · augmentation 63 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare sectors have been slow to adopt AI for direct patient communication due to regulatory caution, liability concerns, and professional norms; no measurable production displacement of neurologist-patient disclosure conversations exists today.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall has slow, cautious AI adoption for direct patient communication, especially for high-stakes diagnostic and prognostic conversations, despite faster uptake in documentation support.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by preparing summaries, organizing treatment comparisons, and drafting talking points, moderately improving efficiency; however, the neurologist must deliver and adapt the message in real-time, limiting the transformative potential.
Augmentation potentialclaude-sonnet-54/5AI can help neurologists prepare clear explanations, summarize research on treatment options, and draft patient education materials, meaningfully aiding preparation for these conversations.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can draft diagnostic summaries and treatment information, the task requires genuine informed consent dialogue, emotional intelligence, and adaptation to individual patient/family context and comprehension levels—elements current AI systems cannot reliably execute end-to-end with equal quality to a skilled neurologist.
Task automatabilityclaude-sonnet-51/5This task requires empathetic, personalized real-time communication of complex medical information and emotional support, which current AI cannot perform end-to-end without a physician present.
Adoption barriersclaude-haiku-4-5-202510015/5Regulatory and legal requirements in most jurisdictions mandate that a licensed physician personally conduct informed consent discussions and sign documentation; malpractice and liability frameworks explicitly place responsibility on the clinician, not an AI system.
Adoption barriersclaude-sonnet-55/5Informed consent and diagnostic disclosure are legally and ethically mandated to be performed by a licensed physician, with significant liability exposure for errors or omissions.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI could reduce time spent drafting educational materials, but the neurologist's presence and judgment are irreplaceable; any system would supplement rather than replace, and integration/oversight costs remain high relative to the modest labor savings.
Cost vs. human wageclaude-sonnet-52/5While AI could draft explanatory content cheaply, the actual delivery still requires a licensed neurologist's time, so overall cost savings versus the human-delivered task are limited.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI product reliably performs this task independently in clinical practice; some chatbots exist for patient education, but they lack the medical authority, individualization, and accountability a neurologist must provide for disclosure conversations that have legal and therapeutic weight.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently informs patients/families of neurological diagnoses and treatment tradeoffs in clinical practice; this remains a physician-delivered function.

Examine patients to obtain information about functional status of areas, such as vision, physical strength, coordination, reflexes, sensations, language skills, cognitive abilities, and mental status.

8

CI 0–16 · exposure 13 · augmentation 50 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI in direct clinical examination is cautious and slow, with pilots confined to narrow decision-support roles. Neurological examination remains a high-touch, high-stakes interaction with limited automation adoption in production clinical settings.
Sector adoption velocityclaude-sonnet-51/5Hands-on clinical examination remains almost entirely human-performed with negligible AI displacement in this specific physical task despite AI adoption elsewhere in healthcare documentation and diagnostics support.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can provide useful assistance through cognitive screening tools, automated reflex analysis from video, language analysis software, and structured documentation support, improving efficiency and reducing clerical burden. However, the augmentation is limited to peripheral aspects rather than transforming the core examination process.
Augmentation potentialclaude-sonnet-53/5AI can assist with structured symptom intake, cognitive test scoring, documentation, and flagging abnormal patterns from exam data, improving efficiency without performing the physical exam itself.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with interpretation of test results and cognitive screening, the physical examination components (reflex testing, strength assessment, coordination checks) and real-time patient interaction required for obtaining functional status information cannot be reliably automated end-to-end. Current AI cannot perform the hands-on clinical examination or achieve 50% time savings at equal quality on the full task.
Task automatabilityclaude-sonnet-51/5This requires hands-on physical examination (reflexes, strength testing, coordination maneuvers, sensory testing) that AI cannot physically perform; it is inherently a manual, patient-contact task.
Adoption barriersclaude-haiku-4-5-202510015/5A licensed neurologist must legally perform the patient examination and interpret findings; this is a core credentialing requirement for neurological practice. Direct patient contact, clinical responsibility, and liability mean the task is protected by licensing, clinical standard-of-care requirements, and organizational/legal constraints on task delegation.
Adoption barriersclaude-sonnet-55/5Physical examination of patients requires a licensed physician, involves direct liability, and is legally and clinically mandated to be performed by a qualified human.
Cost vs. human wageclaude-haiku-4-5-202510011/5Current AI systems that could assist with parts of this task (specialized software, EHR integration, image analysis) add to clinical workflow costs rather than replacing them. The human neurologist remains essential, so the all-in cost of AI-assisted examination exceeds the cost of human examination alone.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing the physical exam, so cost comparison favors the human by default since AI cannot deliver the output at all.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI tools exist for some narrow components (e.g., language analysis, cognitive screening software, image interpretation), but no deployed product reliably performs the complete functional neurological examination. Production systems lack the ability to conduct multi-domain physical and cognitive assessments with the consistency required for clinical decision-making.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs physical neurological exams; AI tools at best support documentation or interpret pre-collected data, not the hands-on exam itself.

Refer patients to other health care practitioners as necessary.

8

CI 5–11 · exposure 9 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare remains highly regulated and slow to adopt autonomous AI systems in clinical decision-making; referral generation is not a sector priority for automation, and physician oversight requirements are strict and uniform.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially clinical decision-making in specialty medicine, adopts AI slowly due to regulatory, liability, and workflow integration challenges despite growing EHR-based decision support.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by suggesting relevant specialists based on clinical data or highlighting appropriate referral pathways, reducing search time and potentially improving guideline adherence, but the physician retains full decision-making authority and must validate each referral.
Augmentation potentialclaude-sonnet-53/5AI can help flag red-flag symptoms, suggest relevant specialists, and streamline documentation, aiding physicians in efficiently determining and processing referrals.
Task automatabilityclaude-haiku-4-5-202510011/5Referrals require complex clinical judgment, understanding of patient context, and knowledge of appropriate specialists—tasks that demand human expertise. AI cannot autonomously determine medical necessity or make legally binding referral decisions today.
Task automatabilityclaude-sonnet-51/5Referring patients requires clinical judgment, medical licensure, and legal responsibility that current AI cannot assume end-to-end; AI cannot autonomously initiate or execute a referral without physician sign-off.
Adoption barriersclaude-haiku-4-5-202510015/5Regulatory and liability barriers are near-absolute: only licensed physicians can legally authorize and sign referrals, and malpractice liability attaches to the physician, not an algorithm, creating a hard requirement for human professional judgment and accountability.
Adoption barriersclaude-sonnet-55/5Referrals are a licensed medical act tied to physician liability and scope-of-practice regulation; a physician must legally authorize and be accountable for referrals.
Cost vs. human wageclaude-haiku-4-5-202510011/5The neurologist's time remains necessary for clinical judgment; any AI assistance only reduces marginal time on documentation or research, not the core decision-making labor, making total substitution cost unlikely to undercut human wages.
Cost vs. human wageclaude-sonnet-52/5AI-assisted triage tools are cheap to run, but the physician's judgment, liability, and care coordination still dominate cost, so overall cost savings versus the physician's time are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed clinical AI system independently generates referrals in production; while AI can suggest specialists based on diagnosis codes, the legal and clinical responsibility remains with the physician who must review, approve, and sign the referral.
Technical feasibility todayclaude-sonnet-52/5Some clinical decision-support tools can suggest specialist referral based on symptoms, but no deployed product autonomously manages referral decisions and execution in production neurology practice.

Coordinate neurological services with other health care team activities.

7

CI 3–11 · exposure 5 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare sectors adopt AI for narrow, well-defined tasks (imaging analysis, data entry) rather than coordination roles; adoption of AI for care coordination remains in pilot phases with limited production deployment due to regulatory and liability constraints.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially clinical coordination workflows, adopts AI slowly due to regulatory, liability, and interoperability constraints.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist by aggregating team availability, flagging scheduling conflicts, or summarizing patient status across departments, but the neurologist must interpret, decide, and communicate—making assistance meaningful but limited to workflow optimization rather than transformative productivity gain.
Augmentation potentialclaude-sonnet-53/5AI can help by summarizing patient records, flagging care gaps, or drafting communications between specialists, aiding but not replacing the coordination role.
Task automatabilityclaude-haiku-4-5-202510011/5Coordinating services requires real-time decision-making about patient care priorities, understanding complex team dynamics, and navigating competing medical needs—tasks that demand human judgment and accountability that current AI cannot replicate end-to-end.
Task automatabilityclaude-sonnet-51/5Coordinating care requires real-time judgment, negotiation with other clinicians, and integration of clinical context that current AI cannot autonomously perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Healthcare coordination and care team leadership are legally and professionally bound to licensed practitioners; regulatory frameworks (FDA, CMS, state medical boards) and liability requirements mandate human responsibility and sign-off for care pathway decisions.
Adoption barriersclaude-sonnet-55/5Care coordination decisions carry direct liability and require a licensed physician's judgment and legal accountability, creating a hard barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The overhead of integrating AI into healthcare coordination, combined with the need for human oversight and error correction, exceeds the cost of a neurologist or care coordinator performing this task directly.
Cost vs. human wageclaude-sonnet-52/5Any AI assistance would still require significant human oversight and integration costs, making all-in cost savings modest at best compared to physician time.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can help schedule appointments or flag scheduling conflicts, no deployed product reliably performs the full coordination function (prioritization, conflict resolution, stakeholder communication) that a neurologist must execute with accountability.
Technical feasibility todayclaude-sonnet-51/5No deployed product manages cross-team clinical coordination for neurology patients; existing tools only support scheduling or documentation fragments.

Prescribe or administer medications, such as anti-epileptic drugs, and monitor patients for behavioral and cognitive side effects.

5

CI 3–7 · exposure 5 · augmentation 50 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of AI in clinical medication management remains cautious and narrow (mainly decision-support dashboards), reflecting the high stakes of prescribing errors and strong regulatory constraints on autonomous clinical decisions.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialty clinical care like neurology, has seen slow deep-adoption of autonomous AI decision-making due to regulatory and safety constraints, though AI-assisted documentation is spreading.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by flagging drug interactions, summarizing side-effect profiles, and organizing patient monitoring data, helping the neurologist work more efficiently while maintaining full clinical control and responsibility.
Augmentation potentialclaude-sonnet-53/5AI can assist with drug interaction checks, dosing calculators, and flagging documented side effects in records, improving efficiency without replacing clinical judgment.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires clinical judgment about individual patient pharmacology, real-time assessment of cognitive and behavioral changes, and legal authority to prescribe. Current AI cannot legally prescribe medications or conduct the nuanced in-person monitoring needed to detect side effects in real time.
Task automatabilityclaude-sonnet-51/5Prescribing medication and monitoring patients for behavioral/cognitive side effects requires longitudinal clinical judgment, physical exam, and patient interaction that current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Strong legal and regulatory barriers mandate that a licensed neurologist must prescribe medications and bear liability for adverse outcomes. Medical licensing, DEA authority (for controlled substances), and patient safety requirements create hard barriers to full automation.
Adoption barriersclaude-sonnet-55/5Prescribing controlled and specialized medications like anti-epileptics legally requires a licensed physician, and monitoring for adverse effects carries high liability and clinical accountability.
Cost vs. human wageclaude-haiku-4-5-202510011/5The liability, regulatory compliance, and required human oversight costs substantially exceed any computational savings. A neurologist's clinical judgment and legal responsibility cannot be substituted; AI reduces costs only marginally as a support tool.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the licensed prescriber and monitoring role, so there is no viable AI-only cost comparison; a physician's involvement remains mandatory.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with drug interaction checking and side-effect literature review, no deployed product reliably performs the full clinical decision-making and monitoring workflow independently. Clinical oversight and human prescription authority remain mandatory.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently prescribes or monitors neurological patients on medications; clinical decision support tools exist but require physician oversight and action.

Determine brain death using accepted tests and procedures.

3

CI 0–5 · exposure 5 · augmentation 25 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5This task is embedded in highly regulated, conservative medical practice with strong institutional and legal requirements for physician oversight. Adoption of autonomous or near-autonomous AI for brain death determination is minimal; the sector remains centered on physician-led protocols with no meaningful displacement occurring.
Sector adoption velocityclaude-sonnet-51/5Critical care neurology and end-of-life determination are extremely conservative, high-stakes clinical domains with essentially no AI adoption for this specific task.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can provide some assistance by automating analysis of individual test results (EEG, imaging), but the high stakes and need for integrated clinical judgment mean the augmentation is limited to supporting specific sub-tasks rather than meaningfully transforming overall physician productivity on the core determination task.
Augmentation potentialclaude-sonnet-52/5AI could assist with documentation, checklist compliance, or EEG/imaging interpretation support, but offers minimal help with the core physical examination and clinical determination itself.
Task automatabilityclaude-haiku-4-5-202510011/5Brain death determination is a complex clinical task requiring integration of multiple test modalities (neurological exams, imaging, EEG, confirmatory tests), interpretation of nuanced findings, and ultimately a legal/medical judgment that no current AI system performs end-to-end autonomously. The task critically depends on real-time patient examination and clinical correlation that AI cannot replicate.
Task automatabilityclaude-sonnet-51/5Determining brain death requires hands-on clinical examination (pupillary response, corneal reflex, apnea testing) and legal/medical judgment that cannot be performed by AI systems today; no end-to-end automation is feasible or appropriate.
Adoption barriersclaude-haiku-4-5-202510015/5Brain death determination is legally mandated to be performed and signed by licensed physicians (often multiple neurologists under statutory protocols). Regulatory requirements, legal liability for errors, and the irreversible consequences of the determination create hard barriers that prevent substitution of human physician judgment.
Adoption barriersclaude-sonnet-55/5Brain death determination is legally mandated to be performed and certified by licensed physicians following strict statutory protocols, representing one of the strongest possible regulatory and liability barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems for supporting neurological diagnosis remain expensive to implement, validate, and maintain in clinical settings, while the time and expertise of a neurologist performing this critical task are relatively inexpensive compared to the liability and validation costs of automation. The all-in cost of AI oversight would exceed the direct cost of physician evaluation.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical exam and legal certification, so no meaningful cost comparison exists—the human physician is the only viable performer.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist with analyzing individual components (e.g., interpreting EEG patterns or imaging), no deployed clinical product reliably performs the full brain death determination process. AI tools exist for narrow sub-tasks (image analysis, EEG interpretation) but not for the integrated diagnostic workflow that this task demands.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs brain death determination; this remains a purely human, bedside clinical procedure with legal accountability attached.

Perform specialized treatments in areas such as sleep disorders, neuroimmunology, neuro-oncology, behavioral neurology, and neurogenetics.

3

CI 3–3 · exposure 0 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While neurology increasingly adopts diagnostic AI tools and EHR integration, actual treatment decisions and procedures remain heavily human-driven; adoption is slow and confined to supportive roles rather than autonomous performance.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialized clinical treatment, adopts AI slowly due to regulatory, liability, and safety constraints, with actual production deployment for treatment tasks remaining rare.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist neurologists by analyzing sleep studies, imaging data, genetic profiles, and literature synthesis, moderately improving diagnostic and treatment-planning efficiency; however, the specialized nature limits broad augmentation gains.
Augmentation potentialclaude-sonnet-53/5AI can assist with diagnostic imaging interpretation, literature review, treatment planning support, and genetic data analysis, meaningfully aiding but not replacing the neurologist's specialized treatment role.
Task automatabilityclaude-haiku-4-5-202510011/5Specialized neurological treatments involve patient-specific diagnosis, personalized intervention decisions, and hands-on clinical procedures that require real-time adaptation and deep medical judgment; current AI cannot perform these end-to-end with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-51/5This involves direct clinical treatment delivery, specialized procedures, and patient management requiring physical presence, licensed judgment, and hands-on care that current AI cannot execute end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Neurological treatments are legally reserved to licensed physicians; FDA, state medical boards, and professional liability regulations create hard legal barriers that prevent autonomous AI substitution.
Adoption barriersclaude-sonnet-55/5Delivering specialized medical treatment requires a licensed neurologist with malpractice liability, board certification, and legal authority to treat patients, creating hard regulatory and licensing barriers.
Cost vs. human wageclaude-haiku-4-5-202510011/5The specialized expertise and malpractice liability required for these treatments, combined with the need for licensed physician oversight and direct patient care, make current AI deployment far more expensive than the value of automated outputs.
Cost vs. human wageclaude-sonnet-51/5AI cannot perform the treatment itself, so the relevant cost comparison is not applicable in AI's favor; any AI use adds cost as a supplementary tool alongside the physician's labor.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product reliably performs specialized neurological treatments (sleep disorders, neuro-oncology, behavioral neurology, neurogenetics) in production; clinical decision support exists but does not substitute for the full treatment task.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product performs specialized neurological treatments; AI exists only as decision-support or diagnostic aid in research/pilot contexts, not as treatment-delivering systems.

Prescribe or administer treatments, such as transcranial magnetic stimulation, vagus nerve stimulation, and deep brain stimulation.

1

CI 0–3 · exposure 0 · augmentation 38 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Adoption of AI in this task is negligible because legal and safety requirements mandate human neurologist involvement in every step. The sector shows no trend toward AI substitution in prescribing or administering controlled neuromodulatory therapies.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially specialized neurology and neurosurgery, adopts AI slowly for procedural and prescriptive decisions due to regulatory oversight, safety concerns, and the physical nature of the interventions.
Augmentation potentialclaude-haiku-4-5-202510012/5AI may provide marginal assistance via treatment protocol suggestions or patient history analysis, but the core tasks—clinical judgment, device calibration, and direct patient management—remain wholly human-driven, limiting transformative augmentation potential.
Augmentation potentialclaude-sonnet-53/5AI can assist by helping identify candidate patients, optimizing stimulation parameters, or analyzing treatment response data, but it does not replace the physician's decision-making or hands-on administration.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires real-time clinical judgment, patient-specific device calibration, and direct patient interaction to monitor safety and efficacy—none of which current AI can perform end-to-end. Prescribing and administering invasive neuromodulatory treatments demand human accountability and cannot be delegated to AI systems today.
Task automatabilityclaude-sonnet-51/5Deciding on and administering neuromodulation therapies like TMS, VNS, and DBS requires physical procedures (device implantation, electrode placement) and clinical judgment integrating patient history, imaging, and risk assessment that AI cannot perform end-to-end today.
Adoption barriersclaude-haiku-4-5-202510015/5This task is protected by strong regulatory and licensing barriers: only licensed physicians can prescribe neuromodulatory treatments, FDA-cleared devices require physician oversight, malpractice liability rests with the clinician, and direct patient contact is mandatory for safety monitoring and informed consent.
Adoption barriersclaude-sonnet-55/5Prescribing and administering these treatments requires licensed physician authority, surgical credentials, and carries high liability; regulatory and legal frameworks mandate human medical professionals for these decisions and procedures.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI cannot reduce the cost of this task because it cannot replace the neurologist's core role: the prescribing decision, device programming, and patient supervision are all human-performed and legally required, making the full labor cost unavoidable.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this physical, procedural task, so AI cost cannot be compared favorably—human specialists and surgical teams remain the only means of delivery.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product independently prescribes or administers TMS, VNS, or DBS. These are strictly physician-directed interventions requiring licensing, physical manipulation of medical devices, and real-time clinical oversight that only human neurologists can legally and safely provide.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product independently prescribes or administers these invasive/procedural neurological treatments; this remains entirely within physician and surgical team practice.

Supervise medical technicians in the performance of neurological diagnostic or therapeutic activities.

1

CI 0–3 · exposure 0 · augmentation 38 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5This task is legally and professionally protected; supervision of medical personnel by AI is not occurring in healthcare practice and is unlikely given regulatory constraints and liability frameworks.
Sector adoption velocityclaude-sonnet-52/5Healthcare adoption of AI for clinical decision support is growing, but supervisory and legal responsibility functions in medicine adopt automation very slowly due to regulation and liability.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could assist by flagging anomalies in technician-recorded data or reminding the neurologist of protocol checklists, but the core supervisory decision-making must remain with the licensed neurologist, limiting transformative augmentation potential.
Augmentation potentialclaude-sonnet-53/5AI can help by monitoring diagnostic data quality, flagging anomalies, or assisting documentation, giving some productivity benefit while the neurologist retains supervisory responsibility.
Task automatabilityclaude-haiku-4-5-202510011/5Supervising medical technicians requires real-time judgment, accountability, and legal oversight of clinical personnel performing diagnostic/therapeutic activities. Current AI cannot replace this supervisory and accountability function, which is inherently human and regulatory in nature.
Task automatabilityclaude-sonnet-51/5Supervision of clinical staff involves real-time judgment, accountability, in-person oversight, and hands-on correction that current AI cannot perform end-to-end.hood
Adoption barriersclaude-haiku-4-5-202510015/5Hard regulatory and legal barriers exist: only a licensed neurologist can supervise medical technicians in neurological diagnostic/therapeutic activities. Medical boards, liability law, and institutional credentialing require a named, accountable human supervisor.
Adoption barriersclaude-sonnet-55/5Medical licensing, liability, and regulatory requirements mandate that a qualified physician supervise technicians performing neurological procedures, making this a hard legal barrier.
Cost vs. human wageclaude-haiku-4-5-202510011/5This task is inseparable from a neurologist's license and legal responsibility. Any AI assistance would still require full human supervision, making the cost ratio strongly favor retaining the human neurologist rather than attempting any automation.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this supervisory role, so no meaningful cost comparison favors AI; a licensed physician's oversight cannot be replaced by inference costs.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product can assume supervisory responsibility for medical technicians' clinical work. This requires a licensed neurologist's legal accountability and real-time decision-making authority, which no AI system is authorized or designed to provide.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product supervises medical technicians performing diagnostic or therapeutic procedures; this remains a human management function.

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.