Medical Assistants
31-9092.00Median wage $45,690/yr817,870 employed (US)Rank #332 of 923 scored · top 36% by substitution
Perform administrative and certain clinical duties under the direction of a physician. Administrative duties may include scheduling appointments, maintaining medical records, billing, and coding information for insurance purposes. Clinical duties may include taking and recording vital signs and medical histories, preparing patients for examination, drawing blood, and administering medications as directed by physician.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution33
Exposure32
Augmentation52
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
20 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
20%
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%33
panel mean rating 2.3/5 → substitution pressure 33/100
Technical feasibility todayw 20%29
panel mean rating 2.1/5 → substitution pressure 29/100
Cost vs. human wagew 15%34
panel mean rating 2.4/5 → substitution pressure 34/100
Adoption barriersw 20%inverted — strong barriers lower the score36
panel mean rating 3.6/5 (barrier strength) → substitution pressure 36/100
Sector adoption velocityw 10%30
panel mean rating 2.2/5 → substitution pressure 30/100
Task breakdown (20 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.
Perform general office duties, such as answering telephones, taking dictation, or completing insurance forms.
80CI 72–87 · exposure 83 · augmentation 100 · importance 4.5/5 · click for rater detail
Perform general office duties, such as answering telephones, taking dictation, or completing insurance forms.
80| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare organizations are actively deploying AI-driven phone systems, transcription services, and administrative automation; adoption is well underway in larger practices and health systems, with measurable cost-displacement trends. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare administration is adopting AI transcription and scheduling tools steadily, but overall digitization of small clinical offices lags behind finance or tech sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 5/5 | AI assistants meaningfully augment medical assistants by handling call routing, transcribing notes in real time, and pre-filling forms, allowing the human to focus on complex inquiries and clinical support tasks. |
| Augmentation potential | claude-sonnet-5 | 5/5 | AI dictation, transcription, and form-autofill tools substantially speed up these clerical tasks while assistants retain oversight and handle exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Current AI systems can reliably answer routine telephonic inquiries via voice assistants, transcribe dictation with high accuracy (speech-to-text), and auto-populate insurance forms using OCR and document processing, collectively saving >50% time at comparable quality. |
| Task automatability | claude-sonnet-5 | 4/5 | Answering calls, transcribing dictation, and filling insurance forms are largely structured, language-based tasks well-suited to current AI voice agents, speech-to-text, and document automation, though some in-office coordination remains. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | HIPAA compliance and data-handling requirements create moderate friction, and some patient-facing telephone screening may need human validation, but these are largely procedural barriers rather than legal prohibitions on automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensure is required for office/clerical tasks, but healthcare data privacy (HIPAA) and payer-specific insurance form accuracy requirements create moderate compliance friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Inference costs for answering phones, transcription, and form automation are a fraction of a medical assistant's loaded wage; even with integration and oversight, the per-task cost is roughly one to two orders of magnitude lower. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated transcription, call handling, and form-filling software cost a small fraction of a medical assistant's loaded wage per unit of work, though integration and oversight add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed products (HIPAA-compliant voice systems, transcription services like Dragon/Ambient, form-filling automation) perform these tasks in production across healthcare, though integration complexity and compliance overhead introduce some friction. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Deployed AI phone agents, medical dictation/transcription tools (e.g., Nuance DAX), and automated insurance claim/form processing are already in production use across healthcare settings, though not universally adopted for full-office duties. |
Schedule appointments for patients.
79CI 71–87 · exposure 83 · augmentation 75 · importance 4.2/5 · click for rater detail
Schedule appointments for patients.
79| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare is rapidly adopting scheduling automation; major health systems and digital-first practices have deployed AI scheduling agents in production, particularly in large, well-resourced settings, though smaller or traditional practices lag. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare administrative functions are adopting automation steadily but unevenly, with many practices still relying on human staff for scheduling due to legacy systems and patient preference. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI scheduling tools significantly assist medical assistants by automating data entry, conflict checking, and reminder sending, allowing humans to focus on complex scheduling needs and patient communication, meaningfully raising their throughput. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling tools significantly boost efficiency for medical assistants by handling routine bookings, reminders, and rescheduling, freeing staff for more complex patient interactions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Scheduling appointments is largely routine and rule-based: checking availability, recording patient details, and sending confirmations can be automated end-to-end by current AI systems (chatbots, calendar integration APIs) with minimal human oversight, easily achieving >50% time savings. |
| Task automatability | claude-sonnet-5 | 5/5 | Appointment scheduling is a well-structured, rules-based task involving calendar management and simple decision logic, which off-the-shelf AI scheduling assistants can handle end-to-end with significant time savings. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | While scheduling can be fully automated, many practices maintain human schedulers for liability concerns, patient preference for human contact, and integration friction with legacy EHR systems; regulatory requirements (HIPAA compliance, authorization) create moderate friction but not a legal bar to automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensure is required for scheduling, but some patients prefer human contact for complex scheduling needs, insurance verification, or sensitive situations, creating moderate organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | The all-in cost of an AI scheduling system (inference + EHR integration + light oversight) is dramatically lower than a full-time medical assistant's loaded salary, typically 1–5% of the annual cost per task completed. |
| Cost vs. human wage | claude-sonnet-5 | 5/5 | Automated scheduling software costs a small fraction of a medical assistant's loaded wage per interaction, especially at scale, making AI dramatically cheaper for this narrow task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Deployed scheduling AI and conversational agents already handle appointment booking in production across many healthcare systems; platforms like Accolade, Zocdoc, and embedded EHR scheduling assistants demonstrate reliable performance at scale, though some complex requests may require human intervention. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Numerous deployed products (e.g., automated scheduling systems, chatbots integrated with EHRs like Epic's MyChart, Phreesia, and voice AI schedulers) reliably handle patient scheduling in production today, though edge cases still require human intervention. |
Keep financial records or perform other bookkeeping duties, such as handling credit or collections or mailing monthly statements to patients.
74CI 72–75 · exposure 75 · augmentation 75 · importance 3.6/5 · click for rater detail
Keep financial records or perform other bookkeeping duties, such as handling credit or collections or mailing monthly statements to patients.
74| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare billing automation is already widespread and mature; practices routinely deploy practice management systems that automate statement mailing and payment posting. Adoption in larger health systems is near-universal, though small independent practices lag. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare administrative back-office functions have moderate automation adoption via EHR/billing software, but many small practices still rely on manual processes or hybrid staff-software workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered billing assistants (automated payment reminders, predictive collections prioritization, flagged accounts) substantially raise a human's productivity on financial record-keeping while the assistant handles routine posting and statement generation. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-enabled billing software substantially reduces manual entry and statement generation effort while staff retain oversight for exceptions, disputes, and patient communication. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI can handle most of this task end-to-end: invoice generation, payment tracking, statement preparation, and even basic collections workflows are routine for accounting software and AI-enabled RPA systems. The primary constraints are integration with existing patient management systems and occasional judgment calls on collections, but the core time-saving threshold is readily met. |
| Task automatability | claude-sonnet-5 | 4/5 | Bookkeeping tasks like financial record-keeping, generating statements, and tracking collections are highly structured, rule-based data processing tasks that current accounting/billing software and AI-augmented tools handle well end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minimal legal barriers exist for automating bookkeeping tasks; no licensing requirement mandates human involvement. However, HIPAA compliance, audit trails, and organizational resistance to removing the human touch on collections add modest friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for bookkeeping itself, though some oversight is needed for accuracy and patient communication sensitivity, creating mild organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Billing automation (software licenses and integration) costs a fraction of a full-time medical assistant's wage; monthly statement generation and basic collections can run on $50–200/month in tooling versus $3,500–4,500/month in loaded labor cost. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated billing/bookkeeping software costs a small fraction of a human's hourly wage for equivalent volume of transactions once integrated with practice management systems. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature accounting software (QuickBooks, NextGen, Athenahealth) and billing platforms already automate statement generation, payment recording, and collections tracking in medical practices at scale. Performance is reliable for routine transactions, though edge cases and dispute resolution still require human oversight. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Practice management and medical billing software (e.g., automated statement generation, AR tracking, collections workflows) already perform these functions reliably in production across many clinics. |
Inventory and order medical, lab, or office supplies or equipment.
73CI 67–79 · exposure 75 · augmentation 75 · importance 4.1/5 · click for rater detail
Inventory and order medical, lab, or office supplies or equipment.
73| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare is a digitized, information-rich sector with strong ROI incentives and mature procurement platforms already embedded; adoption of AI-assisted or fully automated ordering is rapid in hospital systems, clinics, and larger practices. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare administrative functions are adopting digital inventory tools steadily, but many small practices and clinics still rely on manual or semi-manual processes, placing this in the middle of adoption curves. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI powerfully assists medical assistants by flagging low-stock alerts, predicting demand, auto-generating orders, and surfacing supplier comparisons, allowing staff to focus on approval, exception-handling, and relationship management rather than routine data entry. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-driven inventory software significantly reduces the burden of tracking stock levels and generating orders, letting medical assistants focus on patient-facing duties while still reviewing and confirming orders. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Medical assistants' inventory and ordering tasks are largely routine, data-driven workflows involving stock level checks, purchase order generation, and vendor communication—processes current AI systems handle reliably via APIs and RPA today, achieving >50% time savings for standard procedures. |
| Task automatability | claude-sonnet-5 | 4/5 | Inventory tracking and reordering follows predictable rules and thresholds, which off-the-shelf inventory management and procurement software can handle with significant time savings, though initial setup and supplier integration require effort. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Healthcare regulatory requirements (FDA compliance, traceability for some items) and institutional approval workflows create moderate friction; inventory automation is commonplace but often requires oversight of high-cost or controlled items, and some organizations prefer human sign-off on supply chain decisions. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or clinical judgment is required for this administrative task, though controlled substances or certain medical devices may have some regulatory tracking requirements adding minor friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI-powered inventory and procurement systems are substantially cheaper per transaction than human labor once deployed, with marginal cost per order often cents rather than tens of dollars in loaded medical assistant wages. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated inventory software costs a small fraction of the labor hours a medical assistant would spend manually counting and ordering supplies, especially at scale across a practice. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Mature inventory management and procurement software with AI integration are widely deployed in healthcare settings; systems like NetSuite, SAP Ariba, and cloud-based inventory platforms routinely automate reordering with high reliability, though some edge cases (special approvals, regulatory holds) require human judgment. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Inventory management systems with automated reordering (par-level triggers, barcode scanning, EHR-integrated supply modules) are widely deployed in clinics and hospitals today, though many small practices still do this manually. |
Contact medical facilities or departments to schedule patients for tests or admission.
61CI 52–70 · exposure 58 · augmentation 75 · importance 4.0/5 · click for rater detail
Contact medical facilities or departments to schedule patients for tests or admission.
61| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Healthcare organizations, especially larger hospital systems and outpatient networks, are rapidly adopting scheduling automation and conversational AI for patient intake and appointment management. Adoption is accelerating in the digital health sector, though smaller practices lag. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare administrative functions are adopting AI slower than typical information-sector tasks due to legacy systems, interoperability issues, and cautious IT/compliance environments. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI significantly assists medical assistants by auto-populating scheduling forms, suggesting optimal time slots, flagging conflicts, and handling routine appointment confirmations, freeing them to focus on complex cases and patient communication. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scheduling assistants, automated reminders, and voice-to-text tools can meaningfully speed up the coordination process even while a human medical assistant remains responsible for confirming and resolving exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Current AI systems can reliably handle scheduling tasks using calendaring APIs, phone systems, and patient management platforms. Natural language understanding and scheduling logic are mature, though integration with legacy systems and handling complex multi-party coordination may require some setup, approaching the ≥50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 3/5 | Scheduling calls/coordination involve structured data exchange that AI voice agents and scheduling systems can handle, but real-world variability (insurance, facility-specific requirements, human negotiation) limits full end-to-end automation today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While patient contact preferences and organizational workflow inertia create modest friction, there are no legal or licensing barriers preventing AI from scheduling on behalf of a medical facility. HIPAA compliance is a consideration but is increasingly built into AI products, not a hard blocker. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for scheduling itself, though patient data privacy (HIPAA) and the need for accurate handling of medical test/admission logistics create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | AI scheduling systems cost far less to operate than a human medical assistant's wage once deployed at scale—inference is cheap, integration is one-time overhead. The cost ratio strongly favors automation, though implementation and maintenance require some upfront investment. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Automated scheduling tools can reduce staff time, but integration with disparate facility systems, phone-based coordination, and exception handling still require human oversight, keeping costs roughly comparable rather than dramatically cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 3/5 | AI scheduling assistants and chatbots exist in healthcare (e.g., Zocdoc, various EHR integrations), but they typically work within closed systems or require human fallback for complex scenarios, insurance verification, or special accommodations. Deployment is growing but remains somewhat narrow in scope and error-prone at the edges. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | AI scheduling assistants and voice bots are deployed in some healthcare systems for appointment coordination, but interfacility scheduling requiring negotiation across different EHRs/departments still has notable error rates and narrow scope. |
Record patients' medical history, vital statistics, or information such as test results in medical records.
61CI 50–71 · exposure 62 · augmentation 75 · importance 4.7/5 · click for rater detail
Record patients' medical history, vital statistics, or information such as test results in medical records.
61| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Healthcare is digitizing rapidly and many large health systems pilot ambient documentation and automated intake, but adoption is uneven: large hospital systems move faster than small practices, and regulatory caution limits deployment velocity compared to non-medical sectors. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare documentation AI adoption is growing steadily but healthcare overall lags digitally-native sectors; pilots are common while full-scale deployment across small practices remains limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assistance on data recording tasks (auto-populating fields, flagging missing data, drafting summaries from source documents) materially improves medical assistant productivity while keeping human oversight in place for accuracy and clinical judgment on what matters. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI scribes and structured-data tools meaningfully speed up documentation and reduce administrative burden while medical assistants verify and finalize entries. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Recording structured patient data (vitals, test results, medical history) from existing sources is highly automatable via OCR, HL7 parsing, and structured data extraction. Current AI systems can capture and populate electronic health records from documents and lab reports with high accuracy, achieving well over 50% time savings on data entry work. |
| Task automatability | claude-sonnet-5 | 3/5 | AI ambient scribes and voice-to-EHR tools can transcribe and structure vitals/history, but capturing accurate vitals requires physical measurement and verification, limiting full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | HIPAA compliance, data validation requirements, and organizational policies around medical record integrity create meaningful oversight friction. While no license requirement legally mandates human record-keeping, healthcare institutions typically implement verification workflows and liability concerns slow adoption below the technical ceiling. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for data entry itself, but HIPAA compliance, EHR integration requirements, and accuracy/liability concerns for medical records create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | AI inference and integration for medical record entry costs pennies per record, while a medical assistant's loaded hourly wage (typically $35–45k annual) makes human data entry substantially more expensive for high-volume recording tasks. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Ambient AI scribe subscriptions are cheaper than dedicated scribe labor but still require licensing fees plus human review, keeping costs roughly comparable rather than order-of-magnitude lower. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Multiple deployed EHR systems and medical AI products (e.g., ambient clinical documentation tools, automated intake systems) reliably perform data extraction and recording in production healthcare settings. Error rates on structured fields are low, though human verification remains common practice. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Deployed ambient documentation products (e.g., Nuance DAX, Abridge) are used in clinics for history/note capture, but vitals entry from devices and error-checking still require staff involvement. |
Greet and log in patients arriving at office or clinic.
50CI 25–75 · exposure 50 · augmentation 63 · importance 4.6/5 · click for rater detail
Greet and log in patients arriving at office or clinic.
50| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Although healthcare is increasingly digital, adoption of autonomous check-in systems remains patchy; most clinics still rely primarily on human medical assistants for greeting and intake, with kiosks only supplementing rather than replacing staff. |
| Sector adoption velocity | claude-sonnet-5 | 4/5 | Healthcare administrative front-desk functions have seen fast, broad adoption of digital check-in systems, especially post-pandemic, even though clinical care itself lags in AI adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered check-in kiosks and pre-populated digital forms significantly accelerate the intake process and reduce manual data entry, allowing medical assistants to focus on patient care and complex verification tasks rather than routine greeting and typing. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI check-in tools free medical assistants to focus on other tasks, but the assistant still often verifies information, greets patients personally, and handles exceptions. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Greeting and logging in patients requires interaction with variable, often unpredictable people and systems integration (EHR access, patient identification). While AI could handle the scripted greeting component, reliable end-to-end automation of check-in including verification, payment/insurance checks, and patient data entry faces significant barriers and would not meet the 50% time-saving threshold without substantial human oversight. |
| Task automatability | claude-sonnet-5 | 4/5 | Digital check-in kiosks, tablets, and AI-driven intake systems already handle patient greeting and logging with minimal human involvement, meeting the time-saving threshold for most routine visits. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Healthcare regulations (HIPAA, patient privacy), liability for errors in patient identification or record entry, and the expectation of human contact for vulnerable patients seeking medical care create substantial legal and organizational friction against full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for greeting/logging patients, though some patients (elderly, disabled, non-English speakers) still require human assistance, creating moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Hardware, software licensing, maintenance, and integration costs for a reliable check-in automation system, combined with necessary human oversight, are comparable to or exceed the loaded hourly wage of a medical assistant, especially given the relatively low per-interaction cost of a human performing the task. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Kiosk and software-based check-in systems cost far less per patient interaction than staff time once implemented, though initial setup and occasional human backup add some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Some kiosks and chatbots exist for initial greeting and triage, but they rarely handle full check-in autonomously in production healthcare settings. Error rates around patient identification, insurance verification, and EHR data entry remain too high for reliable deployed systems; most require human staff backup. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Self-service kiosks, patient portals, and automated check-in software are widely deployed in clinics and hospital systems today, though some patients still need staff assistance. |
Explain treatment procedures, medications, diets, or physicians' instructions to patients.
29CI 25–34 · exposure 25 · augmentation 63 · importance 4.7/5 · click for rater detail
Explain treatment procedures, medications, diets, or physicians' instructions to patients.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains a laggard sector for high-stakes task automation; while administrative chatbots are deployed, clinical patient education automation is rare in production due to regulatory caution and liability aversion. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare administrative settings are adopting AI documentation and communication tools, but direct patient-facing instructional roles remain slow to adopt full automation due to compliance and trust concerns. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting explanation templates, summarizing physician notes, or generating multilingual patient materials, which a medical assistant then personalizes and delivers—raising efficiency on preparation without replacing the human-patient communication. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can draft clear, personalized explanations of treatments, medications, and diet plans that medical assistants can review and deliver, meaningfully speeding up preparation and improving consistency. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can generate explanations of procedures and medications, the task requires contextual adaptation to individual patient understanding, health literacy, and emotional state—factors that demand real-time judgment. Current systems struggle with reliable, legally safe patient communication without human oversight. |
| Task automatability | claude-sonnet-5 | 2/5 | While AI chatbots can generate patient-friendly explanations of treatments and medications, in-person delivery, answering follow-up questions, and adapting to patient comprehension in real time still requires a human presence in most clinical settings today.atement |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: patient safety regulations, institutional liability frameworks, scope-of-practice rules, and patient expectation of human contact for medical guidance. Healthcare organizations face legal and reputational risk automating patient education without licensed personnel verification. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing barrier prevents an AI from generating explanatory content, but liability concerns, patient trust, and the need for a qualified staff member to verify accuracy of medical instructions create moderate friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Integration, compliance review, and mandatory human oversight for patient safety make AI+human cost comparable to or higher than direct human explanation, particularly given liability concerns in healthcare settings. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | AI-generated instructions are cheap to produce, but the task as described includes live interpersonal communication, so cost savings are partial rather than a full substitution of the human's wage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No mature production system reliably handles unsupervised patient education on medical matters; chatbots exist but carry unacceptable liability and compliance risk. Deployed products require human oversight and are not trusted for independent patient instruction. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Patient education chatbots and AI-generated after-visit summaries exist, but they are largely supplementary tools rather than replacements for the in-person explanation medical assistants provide during visits. |
Perform routine laboratory tests and sample analyses.
26CI 25–28 · exposure 25 · augmentation 50 · importance 4.6/5 · click for rater detail
Perform routine laboratory tests and sample analyses.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Hospital and clinical lab systems have adopted automated analyzers for high-volume assays, but adoption is uneven across smaller practices and remains equipment-focused rather than AI-driven; the pace is moderate, with labs still relying heavily on human technicians. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare support settings are generally slower adopters of full automation for physical clinical tasks compared to information-based professional services, though point-of-care automated analyzers have long been used for specific tests. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist medical assistants by flagging abnormal results, suggesting follow-up tests, and automating data logging and reporting, improving their efficiency and reducing errors on these assistive tasks without removing them from the workflow. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-assisted diagnostic tools and automated analyzers can speed up result interpretation and flag abnormal values, meaningfully aiding the medical assistant's workflow even though the physical task remains human-performed. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI can assist with sample preparation protocols and data interpretation, but the physical handling, collection, and analysis of biological samples—which are core to this task—requires wet-lab robotics and cannot achieve 50% time savings end-to-end with off-the-shelf systems today. |
| Task automatability | claude-sonnet-5 | 2/5 | Physical sample handling and running lab tests requires manual dexterity, equipment operation, and physical presence that current AI systems cannot perform end-to-end; only data interpretation portions are automatable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Laboratory testing is heavily regulated (CLIA, CAP) and typically requires licensed personnel (medical technologists, lab technicians) to perform, validate, and sign off on results; certification and quality assurance mandates create substantial legal and compliance barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Clinical lab testing is subject to regulatory oversight (e.g., CLIA in the US), quality control requirements, and liability concerns that necessitate human oversight and certification for many test types. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Lab automation systems are capital-intensive and require specialized equipment, maintenance, and operator oversight, making their all-in cost comparable to or higher than trained medical assistant labor for routine work. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Automated lab equipment carries high capital and maintenance costs and does not eliminate the need for a human operator for many routine point-of-care tests, so overall cost savings versus a medical assistant's wage are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Automated analyzers exist for specific assays, but they require human technicians for sample handling, quality control, and troubleshooting; no deployed AI product reliably performs the full scope of routine lab tests without human supervision and intervention. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some automated lab analyzers exist for specific tests but they are specialized hardware devices, not general AI agents, and human medical assistants still perform sample collection, prep, and machine operation in most settings. |
Interview patients to obtain medical information and measure their vital signs, weight, and height.
25CI 25–25 · exposure 25 · augmentation 50 · importance 4.8/5 · click for rater detail
Interview patients to obtain medical information and measure their vital signs, weight, and height.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains a laggard sector for AI automation due to regulatory conservatism, patient preference for human contact, and liability concerns; while some telemedicine and kiosk-based intake exists, production displacement of medical assistants remains minimal. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare administrative support is adopting AI slowly for scheduling and documentation, but hands-on clinical tasks like vital sign measurement remain largely untouched by automation in production settings. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted forms (auto-populated templates from prior records, prompted symptom checklist) and connected vital-sign devices can modestly accelerate data entry and flag abnormal readings, but the human medical assistant's judgment and communication remain central to the task. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled intake forms, voice transcription, and connected vital-sign devices can speed up data entry and pre-visit history gathering, meaningfully assisting the medical assistant even though the physical exam is done manually. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Vital sign measurement (BP, temperature, pulse) requires physical hardware interaction that current AI cannot perform; patient interviews can be partially automated via chatbots but lack the nuance, empathy, and ability to probe follow-up symptoms that human assessment provides, and quality parity is not achievable end-to-end. |
| Task automatability | claude-sonnet-5 | 2/5 | Interviewing patients and physically measuring vital signs, weight, and height requires hands-on contact and real-time rapport-building that current AI cannot perform end-to-end; digital intake forms and connected devices can partially automate history-taking but not the physical measurements.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Clinical practice standards, patient privacy (HIPAA), liability for misdiagnosis or measurement error, and healthcare organizational culture all require human contact and sign-off; many clinics and insurers mandate a licensed or certified human obtain the initial history and vitals. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Direct physical patient contact, liability for inaccurate vital signs, and clinical workflow norms create strong barriers to full automation, though intake questionnaires face fewer restrictions. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Vital sign devices (manual or automated) require capital investment and ongoing calibration; interview automation still needs human oversight and correction, keeping total cost per patient encounter close to or above the hourly wage of a medical assistant. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Physical measurement equipment and in-person presence still require staffing costs comparable to a medical assistant; digital intake tools reduce some interview time but don't replace the full cost of the task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While some isolated components (scheduling reminders, basic questionnaire administration) exist in deployed systems, end-to-end patient interviewing and vital sign capture remains primarily manual; no mature product reliably combines accurate symptom elicitation with physical measurement in production. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some clinics use digital check-in kiosks or chatbots for pre-visit history collection, but actual vital sign measurement still requires a human or dedicated medical device operated in-person, so no product performs this whole task reliably in production. |
Clean and sterilize instruments and dispose of contaminated supplies.
19CI 5–34 · exposure 20 · augmentation 25 · importance 4.8/5 · click for rater detail
Clean and sterilize instruments and dispose of contaminated supplies.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Adoption remains slow; most clinical settings still use manual or semi-automated sterilization with human inspection. While large hospital systems are exploring robotic solutions, production deployment is limited and pilots are uncommon. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical, hands-on healthcare support tasks like this show minimal AI/robotic adoption; sterilization remains manual in nearly all clinical settings. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted inventory tracking, scheduling optimization, and real-time monitoring alerts for sterilization cycles can improve workflow efficiency, but the core manual tasks (physical handling, visual inspection, compliance verification) remain human-dependent. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no meaningful assistance for the physical cleaning, sterilizing, and disposal actions involved in this task. |
| Task automatability | claude-haiku-4-5-20251001 | 3/5 | Instrument sterilization (autoclaving, chemical disinfection) can be partially automated with programmed cycles, but the physical handling, inspection for cleanliness, loading/unloading, and compliance verification still require human oversight and dexterity, limiting time savings to roughly 40–60% of the workflow. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical manipulation task requiring handling instruments, autoclaves, and biohazard disposal—no current AI system (software or robotic) can perform this end-to-end in a clinical setting. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Medical instrument sterilization is heavily regulated (OSHA, CDC, state health boards); sterility assurance and chain-of-custody documentation typically require licensed/authorized personnel sign-off, and liability for contamination-related harm creates strong legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Infection control, biohazard disposal, and sterilization protocols are governed by strict regulatory and safety standards (e.g., OSHA, CDC) requiring trained personnel, creating strong barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Automated sterilization equipment has high capital and maintenance costs; integrating robotic handling and compliance oversight would be expensive relative to the hourly wage of medical assistants, making the all-in cost roughly comparable or higher. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI/robotic substitute being deployed for this task, so AI cost cannot be meaningfully compared as cheaper than human labor. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While some specialized cleaning/sterilization equipment exists with automation features, no deployed AI-powered robotic system reliably handles the full end-to-end task (inspection, sorting, loading, monitoring, disposal compliance) in production medical settings at the scale required. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs medical instrument sterilization or contaminated waste disposal autonomously in clinical practice today; this remains firmly a manual task. |
Authorize drug refills and provide prescription information to pharmacies.
19CI 11–28 · exposure 20 · augmentation 63 · importance 4.4/5 · click for rater detail
Authorize drug refills and provide prescription information to pharmacies.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains highly regulated and cautious on liability-bearing tasks. While EHR routing and flagging are adopted, actual AI-driven authorization without human sign-off is not being deployed in production healthcare settings due to legal and malpractice concerns. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Healthcare administrative workflows are adopting AI-assisted EHR tools at a moderate pace, with pilots for refill routing and clinical messaging becoming more common but not yet universal. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | EHR systems and AI-assisted tools can flag refill requests, check for drug interactions, and highlight expired prescriptions, meaningfully assisting the medical assistant or clinician in their review process without replacing the required human decision. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI can effectively pre-populate refill requests, flag inconsistencies, and draft pharmacy communications, meaningfully speeding up the medical assistant's workflow while a human retains final authorization. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could draft or flag routine refill requests, the actual authorization requires a licensed clinician's legal sign-off. Current AI systems cannot independently decide whether a refill is medically appropriate, making full end-to-end automation infeasible. |
| Task automatability | claude-sonnet-5 | 2/5 | Verifying refill eligibility against chart data and communicating with pharmacy systems could partially be automated, but authorization requires clinical judgment often reserved for licensed staff, limiting full automation today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Federal and state pharmacy laws explicitly require a licensed healthcare provider (physician, nurse practitioner, or physician assistant) to authorize drug refills. This legal mandate creates a hard barrier that prevents full substitution regardless of technical capability. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Prescription authorization is tightly regulated; only licensed clinicians or supervised staff acting under protocol can approve refills, creating strong legal and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Deploying AI for this task would require significant oversight infrastructure and clinician review time, adding cost rather than reducing it. The savings from automating routine checks are offset by liability exposure and required human validation. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While software-based refill routing is cheap, the human oversight and liability review needed for authorization keeps blended costs closer to human-comparable rather than order-of-magnitude cheaper. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably performs this task autonomously today. EHR systems can route requests and flag duplicates, but a human clinician must review and authorize each refill—this is a regulatory requirement, not a technical limitation that AI has overcome. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some EHR systems have automated refill-request routing and pharmacy interoperability tools, but reliable end-to-end handling of authorization decisions in production is narrow and still requires human sign-off. |
Operate x-ray, electrocardiogram (EKG), or other equipment to administer routine diagnostic tests.
15CI 5–25 · exposure 13 · augmentation 38 · importance 4.0/5 · click for rater detail
Operate x-ray, electrocardiogram (EKG), or other equipment to administer routine diagnostic tests.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare adoption of automation for equipment operation remains low outside imaging centers; most clinics and hospitals retain human technicians. Regulatory caution, capital constraints, and infection-control/safety concerns slow adoption relative to information-intensive sectors. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical, hands-on medical procedures in outpatient/clinical settings show minimal AI-driven automation or displacement; this is a low-digitization, high-physical-contact task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by suggesting optimal positioning, flagging obvious artifacts, prompting quality checks, and providing preliminary interpretation; these augmentations raise technician efficiency modestly but do not transform the core task since the human remains essential for hands-on operation and judgment. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with image interpretation or EKG waveform analysis after acquisition, but offers little help with the physical operation and administration of the test itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While image acquisition itself is partially automatable (e.g., positioning guidance), the task requires hands-on patient interaction, equipment calibration, safety protocols, and quality control that resist full end-to-end automation. Current AI cannot handle patient positioning, consent, artifact detection, and re-acquisition decisions without substantial human oversight. |
| Task automatability | claude-sonnet-5 | 1/5 | Operating diagnostic imaging and EKG equipment requires physical manipulation of patients and machines, positioning, and hands-on procedural steps that current AI cannot perform without a robotic embodiment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory and liability barriers are high: X-ray and EKG equipment operation is often state-licensed or requires supervised certification; patient safety, radiation exposure, and clinical liability create strong asymmetry favoring human accountability. Medical device regulation and patient contact requirements form substantial legal friction. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Operating radiologic equipment typically requires certification/licensure and adherence to safety regulations (e.g., radiation safety), and hands-on patient contact is often legally restricted to trained personnel. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The capital and integration cost of robotic equipment operation, safety systems, and regulatory compliance exceeds the wage of a medical assistant. Oversight and maintenance overhead further reduces cost advantage relative to deploying human staff. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI system performing this physical task, so cost comparison favors the human by default since AI cannot yet substitute at all. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product performs the full task autonomously in clinical settings. AI assists with image interpretation (post-hoc) but does not operate equipment, position patients, or manage the clinical workflow independently. Products exist for analysis but not for autonomous equipment operation. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously operates x-ray or EKG machines on patients today; this remains a physical, human-performed task with no robotic substitute in clinical use. |
Collect blood, tissue, or other laboratory specimens, log the specimens, and prepare them for testing.
13CI 0–25 · exposure 13 · augmentation 38 · importance 4.7/5 · click for rater detail
Collect blood, tissue, or other laboratory specimens, log the specimens, and prepare them for testing.
13| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains a conservative sector; adoption is limited to digital logging tools and bar-coding in some labs, with virtually no production deployment of automated specimen collection or preparation. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Healthcare's physical, hands-on clinical tasks show minimal AI adoption; this sub-task involves direct patient contact with no robotic deployment at scale. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted specimen tracking, automated logging systems, and quality-control imaging offer moderate productivity gains for medical assistants, but the core manual collection and handling tasks remain human-dependent. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with specimen tracking, labeling accuracy, and lab information system logging, but offers little assistance for the physical collection itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While specimen logging and data entry can be automated, the physical collection of blood and tissue samples requires human dexterity, patient interaction, and real-time procedural judgment that current robots cannot reliably perform end-to-end in clinical settings. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical manipulation (venipuncture, tissue collection) and dexterous handling of biological specimens, which current AI systems cannot perform end-to-end; only logging/documentation portions are automatable. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Significant regulatory and liability barriers exist: clinical laboratory standards (CLIA), specimen chain-of-custody requirements, patient consent protocols, and medical responsibility create hard requirements for human accountability and oversight. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Phlebotomy and specimen handling require certified/trained personnel under clinical safety and regulatory standards, with direct patient contact and liability concerns making automation legally and practically infeasible. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Partial automation (logging, labeling) is cost-competitive, but the capital and integration cost of robotic collection systems, combined with still-required human oversight, makes the all-in cost comparable to or higher than human phlebotomists. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute for physical specimen collection, so AI cost is not comparable—human labor is the only viable option for this physical task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Automated systems exist for specimen tracking and labeling, but no deployed product performs the full task (collection, logging, preparation) reliably without human intervention; robotic phlebotomy remains experimental and impractical at scale. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs blood draws or specimen collection autonomously; this remains a manual clinical skill performed by trained humans. |
Prepare treatment rooms for patient examinations, keeping the rooms neat and clean.
10CI 5–15 · exposure 0 · augmentation 13 · importance 4.7/5 · click for rater detail
Prepare treatment rooms for patient examinations, keeping the rooms neat and clean.
10| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Medical practices have shown minimal adoption of automation for room preparation. The sector remains conservative, labor-intensive, and reliant on human workers for these physical, context-dependent tasks. No measurable displacement trend is evident. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical facility upkeep in healthcare settings shows minimal AI/robotic adoption; this is a low-digitization, manual task with no meaningful automation trend in medical offices. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI offers minimal assistance for this task. While scheduling systems can help optimize room-prep timing, they provide marginal productivity gains. The core physical work of cleaning and organizing remains fundamentally human-dependent. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI tools offer essentially no assistance for physically tidying and preparing a treatment room, as this is not a cognitive or information-processing task. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Preparing treatment rooms requires physical manipulation of objects (cleaning, organizing supplies, adjusting furniture) in a physical space—tasks that current AI systems cannot perform. No end-to-end automation with 50% time saving is feasible without robotics, which are not yet deployable at scale in medical settings. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical cleaning and setup task requiring manual manipulation of equipment, surfaces, and supplies in real space, which no current AI system can perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Healthcare facilities have regulatory and infection-control standards that mandate specific cleaning and preparation protocols, often requiring human verification and sign-off. Additionally, direct physical presence in patient-facing spaces creates organizational friction and regulatory requirements that limit substitution. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing specifically governs room cleaning, but infection control standards and clinical safety protocols create some procedural requirements around how spaces must be prepared and sanitized. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Robotic systems capable of medical-room preparation remain far more expensive than hiring a medical assistant, particularly when accounting for integration, maintenance, and oversight costs. Human labor remains the cost-effective solution. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-driven substitute for this physical task, so the human wage remains the only real cost input; any hypothetical robotic solution would be far more expensive than a medical assistant's labor. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed products reliably perform this task. While cleaning robots exist in limited contexts, they are not standard in medical offices and require extensive customization. The task requires contextual judgment about medical room-specific requirements that current systems cannot handle. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product cleans and stages medical treatment rooms; this remains purely a human physical labor task with no robotic products in production for this specific application. |
Help physicians examine and treat patients, handing them instruments or materials or performing such tasks as giving injections or removing sutures.
10CI 0–20 · exposure 8 · augmentation 38 · importance 4.6/5 · click for rater detail
Help physicians examine and treat patients, handing them instruments or materials or performing such tasks as giving injections or removing sutures.
10| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare remains heavily regulated and conservative; adoption of autonomous clinical robotics for routine MA tasks is minimal despite sector digitization. Most adoption is in high-acuity surgical settings with human control, not task automation. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical, hands-on healthcare support roles are among the least digitized and slowest to adopt AI/robotics for direct patient physical care tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist MAs by suggesting appropriate instruments, flagging patient allergies, or drafting documentation, but the hands-on clinical work (injections, sutures, patient handling) cannot be meaningfully augmented by current AI systems. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can help with documentation, scheduling reminders, or prep-list generation around these encounters, but offers minimal direct assistance to the physical act of handing instruments or performing injections. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could assist with some preparatory steps (e.g., retrieving instrument information), the core task involves physical manipulation, patient contact, and real-time clinical judgment that require human presence and dexterity today. Surgical robotics exist but are tele-operated by humans, not autonomous. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical presence, hand dexterity, and real-time physical coordination with a physician during patient exams and procedures like injections and suture removal—no current AI system can perform these physical manipulations. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Strong legal and regulatory barriers protect this role: only licensed healthcare providers can supervise; liability for patient harm; infection control and sterility requirements; direct patient contact and informed consent; state licensure requirements for medical assistants themselves. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Direct patient contact, invasive procedures like injections, and infection control protocols create substantial regulatory, liability, and training barriers that require certified human personnel. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current medical robots are extremely expensive (hundreds of thousands to millions) compared to a medical assistant's annual salary. Integration, training, and liability overhead remain prohibitive for routine tasks. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute for the physical task, so cost comparison favors the human by default; any hypothetical robotic solution would require far more capital investment than a medical assistant's wage justifies. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product reliably performs independent patient examination, injection, or suture removal at clinical standards. Robotic systems in surgery require human control and are not autonomous, task-substituting agents. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical clinical assistance tasks like handing instruments, giving injections, or removing sutures; this remains firmly outside current robotic or AI product capability for general clinical settings. |
Prepare and administer medications as directed by a physician.
7CI 0–15 · exposure 8 · augmentation 50 · importance 4.5/5 · click for rater detail
Prepare and administer medications as directed by a physician.
7| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Healthcare lags in AI adoption for clinical tasks; medication administration remains heavily manual in most settings. Hospitals pilot robotic dispensing and IV systems in specialized units, but broad production adoption remains limited by regulation and risk aversion. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Healthcare direct-care settings show minimal AI adoption for hands-on clinical tasks like medication administration, remaining a laggard area due to physical and regulatory constraints. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI assists medication preparation through barcode verification, dose checking, and drug interaction alerts—raising safety and speed. However, augmentation is limited to pre-administration steps; the supervised administration itself remains primary and human-dependent. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can assist with medication reminders, dosage calculations, drug interaction checks, and documentation, improving safety and efficiency even though the physical act remains human-performed. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While medication preparation (checking doses, labeling) could be partially automated, administering medications requires direct patient contact, real-time clinical judgment, and physical execution that current AI cannot perform. Most of the task's value—safety verification, patient interaction, and actual delivery—remains human-dependent. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on physical and clinical task requiring direct patient contact, dosage verification, and injection/administration skills that current AI systems cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Administering medications is tightly regulated and typically requires a licensed healthcare professional (RN, LPN, or authorized medical assistant) to verify and deliver. Liability, licensure requirements, and legal accountability create hard barriers to full substitution. |
| Adoption barriers | claude-sonnet-5 | 5/5 | Medication administration is tightly regulated, requires licensure/certification, and carries high liability for errors, making this a hard legal barrier to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current automation (barcode scanning, electronic verification) has high upfront integration costs relative to the labor it displaces. Full-task robotics would be extremely expensive compared to medical assistant wages, and oversight costs remain substantial. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so cost comparison favors the human by default since AI cannot deliver the output at all. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product reliably administers medications end-to-end in production. Robotic systems exist in research and niche hospital settings but are not mainstream; rule-based verification systems handle only narrow preparation steps. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product physically prepares or administers medications; this remains firmly in the physical/manual domain outside current AI product capabilities. |
Set up medical laboratory equipment.
5CI 5–5 · exposure 0 · augmentation 25 · importance 3.7/5 · click for rater detail
Set up medical laboratory equipment.
5| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Medical settings remain low in adoption of autonomous AI for hands-on physical tasks. Setup of lab equipment is typically performed by certified or trained medical staff in regulated environments where robotic automation is rare and high-risk. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Physical healthcare support tasks show minimal AI/robotic adoption in production settings; this is a low-digitization, hands-on task typical of laggard adoption sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can provide marginal assistance through documentation lookup, checklist management, or alert systems for out-of-spec conditions, but these add limited productivity gains to the core physical and calibration work that dominates the task. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI could potentially assist with equipment manuals, troubleshooting guidance, or checklists, but offers little direct enhancement to the physical setup process itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Setting up medical laboratory equipment requires physical manipulation of hardware, calibration with precision, and contextual judgment about equipment state—capabilities that current AI systems cannot perform end-to-end. While AI can assist in reading setup instructions or documenting configurations, the hands-on assembly, calibration, and verification remain firmly in human domain. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a hands-on physical task involving manipulating and calibrating physical lab equipment, which current AI systems cannot perform without robotic embodiment far beyond deployed capability. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Medical equipment setup often falls under regulatory oversight (FDA, CLIA) and may require certified personnel sign-off on calibration and readiness. Liability for equipment malfunction is high, and many jurisdictions require a licensed human to validate that equipment is properly configured before clinical use. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Medical lab equipment setup often involves calibration standards, safety protocols, and sometimes certification requirements that create strong barriers to non-human handling. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Automation is not yet feasible, so cost comparison is moot; however, the overhead of any attempted robotic solution would far exceed the cost of a medical assistant performing the task directly. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so any AI cost comparison is moot and the human remains the only cost-effective option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system can autonomously set up medical laboratory equipment in production settings. The task requires physical manipulation, spatial reasoning, and real-time error detection that robotics and AI have not yet achieved reliably at the scale and precision required in clinical environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical setup of medical laboratory equipment; this remains firmly in the domain of human technicians. |
Show patients to examination rooms and prepare them for the physician.
4CI 0–7 · exposure 0 · augmentation 25 · importance 4.6/5 · click for rater detail
Show patients to examination rooms and prepare them for the physician.
4| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare remains among the slowest sectors to adopt full automation of tasks requiring physical presence and direct patient interaction, with strong regulatory and cultural preferences for human staff in clinical roles. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Healthcare support/clinical settings adopt AI slowly for hands-on patient care tasks, though administrative adjacent tasks see more uptake. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist marginally with scheduling or documentation workflows, but the core task of physically guiding and preparing patients offers limited augmentation value—AI cannot meaningfully enhance a human's ability to escort a patient to a room. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can assist with scheduling, patient intake forms, or reminders beforehand, but offers minimal support for the physical escorting and preparation itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical presence to guide patients, physical space navigation, and interpersonal communication—capabilities far beyond current AI systems. No end-to-end automation is feasible without a humanoid robot in a clinical setting, which is not deployable at scale today. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physically escorting patients, physical presence, and hands-on preparation (vitals, gowning, positioning) that current AI cannot perform end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Medical settings operate under strict regulatory frameworks (HIPAA, state medical practice acts), require direct human contact and presence for patient comfort and safety, and involve liability for patient welfare that necessitates a licensed human in the care chain. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Direct patient contact, physical safety, privacy, and clinical workflow norms create strong barriers to any non-human substitution for this task. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Robotic systems capable of performing this task reliably would cost far more than the annual wage of a medical assistant, and integration into clinical workflows would be prohibitively expensive. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute performing this physical task, so AI cost is not comparable—human labor remains the only viable option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No current deployed AI system can autonomously escort patients to examination rooms, prepare them (taking vitals, documenting, positioning), and adapt to patient needs. This requires embodied presence and real-time human judgment. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product physically guides patients or prepares them for physician examination; this remains entirely a human physical-presence task. |
Change dressings on wounds.
3CI 0–5 · exposure 0 · augmentation 25 · importance 4.4/5 · click for rater detail
Change dressings on wounds.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Healthcare adoption of physical automation for wound care remains minimal; robotic systems are confined to surgery centers, and bedside wound dressing remains manual across the vast majority of clinical settings. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Healthcare physical care tasks show very slow AI/robotic adoption due to safety, regulatory, and physical manipulation challenges. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist via image analysis of wound photographs for triage or documentation, but real-time augmentation is limited since the core task—sterile handling and physical application—requires direct human execution. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can support documentation, wound image analysis for healing tracking, or protocol reminders, but offers minimal direct assistance during the hands-on dressing change itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Changing wound dressings requires physical manipulation of sterile materials, assessment of wound condition in real-time, and tactile feedback in a clinical environment. Current AI systems lack embodied manipulation capabilities and cannot perform this hands-on procedure. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical manipulation of dressings on a patient's body, involving fine motor skills, tactile assessment, and adaptive judgment that current AI systems cannot perform end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Strict regulatory requirements (FDA, state scope-of-practice laws) mandate that licensed or supervised clinical personnel perform wound care; patient safety liability and infection-control protocols create hard legal and organizational barriers to automation. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Direct patient contact involving wound care carries infection control, liability, and often requires trained/certified personnel, creating strong practical and regulatory barriers to non-human performance. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The capital cost of a robotic system capable of sterile wound care would far exceed the loaded wage of a medical assistant performing this task, with minimal production deployment to amortize costs. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so any hypothetical automation (specialized robotics) would be far more costly than a medical assistant's wage. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product reliably performs autonomous wound dressing changes. While robotic surgery exists for controlled settings, wound assessment and dressing changes in diverse clinical contexts remain beyond current robotic medical systems. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed robotic or AI product performs wound dressing changes in clinical settings; this remains firmly in the research/prototype stage for medical robotics. |
Related occupations — Healthcare Support
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.