Occupational Therapy Assistants

31-2011.00
Median wage $72,300/yr51,290 employed (US)Rank #703 of 923 scored · top 76% by substitution

Assist occupational therapists in providing occupational therapy treatments and procedures. May, in accordance with state laws, assist in development of treatment plans, carry out routine functions, direct activity programs, and document the progress of treatments. Generally requires formal training.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution19
Exposure18
Augmentation46

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

22 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

9%

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%20

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

Technical feasibility todayw 20%14

panel mean rating 1.5/5 → substitution pressure 14/100

Cost vs. human wagew 15%19

panel mean rating 1.7/5 → substitution pressure 19/100

Adoption barriersw 20%inverted — strong barriers lower the score24

panel mean rating 4.0/5 (barrier strength) → substitution pressure 24/100

Sector adoption velocityw 10%15

panel mean rating 1.6/5 → substitution pressure 15/100

Task breakdown (22 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 any needed educational or treatment supplies.

76

CI 65–87 · exposure 78 · augmentation 75 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Healthcare and educational institutions are actively adopting AI-driven procurement and inventory management systems. Many larger therapy clinics and hospital systems already use automated or semi-automated ordering; adoption is accelerating in the professional services and healthcare sectors.
Sector adoption velocityclaude-sonnet-52/5Healthcare and allied health settings, especially smaller clinics and schools where OT assistants work, tend to be slower adopters of automated procurement systems compared to digital-native sectors.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly assist occupational therapy assistants by automating routine reorder detection, suggesting items based on patient caseload, flagging budget thresholds, and pre-populating orders—allowing staff to focus on clinical oversight and exception handling rather than manual data entry.
Augmentation potentialclaude-sonnet-54/5AI-enabled inventory tracking and reorder reminders can meaningfully streamline the ordering process, letting the assistant focus on identifying specific therapeutic needs rather than administrative tracking.
Task automatabilityclaude-haiku-4-5-202510015/5Ordering supplies is a well-defined, highly automatable task: identify needed items, check inventory, select vendors, process purchase orders, and arrange delivery. Current AI systems can handle supplier research, SKU matching, cost comparison, and order placement end-to-end, easily achieving >50% time savings with equal quality.
Task automatabilityclaude-sonnet-54/5Ordering supplies is a structured, transactional task (identifying needed items, checking inventory, placing orders) that off-the-shelf procurement/e-commerce and AI agent tools can largely handle with proper integration.
Adoption barriersclaude-haiku-4-5-202510012/5Few legal or regulatory barriers prevent automation; supply ordering does not require professional licensure or human sign-off. Minor friction exists around institutional policies and vendor relationships, but these are organizational, not regulatory constraints.
Adoption barriersclaude-sonnet-52/5No licensure requirement to order supplies, though budget authority, purchasing approval workflows, and employer purchasing policies create some organizational friction.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated ordering via AI agents incurs minimal ongoing cost (platform subscription, API calls, light human oversight) compared to the fully-loaded hourly wage of an occupational therapy assistant performing this clerical task manually—easily an order of magnitude cheaper.
Cost vs. human wageclaude-sonnet-54/5Automated ordering/inventory software is inexpensive relative to staff time spent on manual ordering, offering substantial cost savings once integrated.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple deployed solutions (e-procurement platforms, inventory management systems with AI-assisted ordering, vendor-integrated purchasing tools) reliably perform supply ordering at scale in healthcare and educational settings. Minor gaps remain in handling edge cases like customized supplies or complex institutional approval workflows, but core ordering is production-ready.
Technical feasibility todayclaude-sonnet-53/5Automated procurement and inventory-reorder systems exist and are used in many healthcare/clinical settings, but they are typically generic supply-chain tools not specialized for OT contexts and often still require human sign-off.

Perform clerical duties, such as scheduling appointments, collecting data, or documenting health insurance billings.

71

CI 67–75 · exposure 75 · augmentation 75 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Healthcare is a high-digitization sector with rapid EHR and RPA adoption. Appointment scheduling automation and billing AI are already commonplace in practices and clinics; this is a mainstream adoption trajectory, not a pilot phase.
Sector adoption velocityclaude-sonnet-53/5Healthcare administrative functions have moderate AI adoption, with many clinics using digital scheduling and billing systems, though full automation with AI agents remains uneven across smaller practices.
Augmentation potentialclaude-haiku-4-5-202510014/5Even where humans remain, AI substantially raises clerical productivity by auto-populating forms, flagging billing errors, suggesting appointment slots, and organizing data. Human oversight of AI-prepared scheduling or billing documentation is far faster than manual entry.
Augmentation potentialclaude-sonnet-54/5AI scheduling assistants and automated billing coding tools significantly reduce administrative burden on occupational therapy assistants, freeing time for patient-facing work.
Task automatabilityclaude-haiku-4-5-202510014/5Scheduling appointments, data collection, and billing documentation are highly structured, rule-based tasks where current AI systems (email integration, calendar APIs, form automation, RPA) can achieve significant time savings at equal or better quality. End-to-end automation is feasible for routine cases, though some human verification may be needed for edge cases.
Task automatabilityclaude-sonnet-54/5Scheduling, data collection, and billing documentation are structured, repetitive clerical tasks well within the capability of current AI-based scheduling and billing automation tools, meeting the time-saving threshold for most of the workflow.
Adoption barriersclaude-haiku-4-5-202510013/5HIPAA compliance and data security requirements, plus institutional preference for human oversight of billing and insurance matters, create meaningful friction. However, no licensing requirement mandates human performance of these clerical tasks, and most healthcare systems are actively automating them.
Adoption barriersclaude-sonnet-52/5No licensure is required for clerical/billing tasks, though healthcare privacy regulations (HIPAA) and insurer-specific billing rules create some compliance friction.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated scheduling, data entry, and billing systems cost a fraction of a full-time clerical worker's loaded wage (typically $30–45K annually). AI-based solutions with integration can operate at roughly 10–20% of human labor cost for high-volume routine work.
Cost vs. human wageclaude-sonnet-54/5Automated scheduling and billing software costs a small fraction of clerical staff wages per transaction, though integration and oversight costs keep it from being an order of magnitude cheaper in all cases.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed products like calendar scheduling bots, EHR data-entry automation, and billing software with AI-assisted coding are widely used in healthcare organizations today. These systems reliably handle appointment scheduling and billing documentation in production, though some integration and oversight overhead remains.
Technical feasibility todayclaude-sonnet-54/5Practice management software with AI-assisted scheduling and automated billing/claims documentation is deployed widely in healthcare settings today, though occasional exceptions and edge cases still require human review.

Observe and record patients' progress, attitudes, and behavior and maintain this information in client records.

43

CI 25–60 · exposure 45 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and rehabilitation settings are digitizing, but adoption of autonomous AI for clinical documentation remains slow; most facilities still rely on templated EHRs with human input. Regulatory caution and liability concerns slow production deployment of fully automated observation recording.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially allied health and therapy services, has historically been slower to adopt AI due to regulatory, liability, and workflow integration challenges, with pilots more common than full production deployment.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly augment OTA productivity by auto-transcribing observations, suggesting relevant behavioral codes or progress categories, and organizing notes—allowing the assistant to focus on actual patient interaction and more nuanced assessment rather than administrative data entry.
Augmentation potentialclaude-sonnet-54/5AI dictation and note-generation tools can meaningfully speed up the documentation portion of this task, letting assistants spend more time on direct patient observation while AI drafts records for review.
Task automatabilityclaude-haiku-4-5-202510014/5AI systems can reliably extract, classify, and document behavioral observations from clinical notes, video, or structured questionnaires with high accuracy, saving substantial setup and manual recording time. However, some nuanced behavioral interpretation and clinical judgment still benefits from human oversight, preventing a perfect 5.
Task automatabilityclaude-sonnet-52/5Observation of patient behavior and progress requires physical presence, direct interaction, and clinical judgment that current AI cannot perform; only the documentation/note-writing portion is automatable, not the observation itself.
Adoption barriersclaude-haiku-4-5-202510013/5HIPAA, state licensing rules, and organizational policy typically require a licensed clinician (OT or supervisor) to sign off on or review clinical records, limiting full automation. However, AI can assist in data capture and preliminary organization without legal prohibition.
Adoption barriersclaude-sonnet-54/5Clinical documentation in patient records typically requires a credentialed professional to verify accuracy and sign off, given liability, HIPAA, and regulatory compliance concerns in healthcare settings.
Cost vs. human wageclaude-haiku-4-5-202510014/5AI documentation and observation-logging systems cost far less per record than the fully-loaded labor cost of an occupational therapy assistant reviewing and manually writing progress notes. Integration and oversight overhead is modest relative to the time saved.
Cost vs. human wageclaude-sonnet-52/5AI transcription/summarization tools are cheap, but since the core observation task still requires the human therapist assistant, overall cost savings are limited to a minor documentation efficiency gain, not a full task substitution.
Technical feasibility todayclaude-haiku-4-5-202510013/5Electronic health record (EHR) systems with built-in templates and some AI-assisted documentation exist in clinical settings, but they require human data entry and validation. AI-driven clinical documentation tools are emerging but not yet universally deployed or fully autonomous in capturing behavioral nuance.
Technical feasibility todayclaude-sonnet-52/5AI scribe and documentation tools exist and are used in some clinical settings to draft notes from dictation, but no product independently observes patients and generates clinically valid progress records without a human present to gather the input.

Attend continuing education classes.

39

CI 0–77 · exposure 45 · augmentation 38 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and occupational therapy sectors have moderate digital maturity but remain relatively conservative on outsourcing compliance activities; adoption of AI for CE automation is still rare in production, though pilot interest is growing.
Sector adoption velocityclaude-sonnet-51/5There is no sector trend toward AI attending CE classes on behalf of professionals; this is not an area of AI adoption at all.
Augmentation potentialclaude-haiku-4-5-202510012/5AI can help an OTA assistant by summarizing content or highlighting key points, but since the task itself (attending and completing CE) requires minimal active human judgment, augmentation value is limited compared to tasks involving synthesis or decision-making.
Augmentation potentialclaude-sonnet-53/5AI can help by curating relevant courses, summarizing content, generating study notes, or assisting in exam prep, offering moderate productivity support around the task without replacing attendance itself.
Task automatabilityclaude-haiku-4-5-202510015/5Attending continuing education classes (accessing materials, watching videos, completing assignments, taking quizzes) can be fully automated by AI agents that can log into platforms, consume content, submit work, and document completion without human intervention, easily achieving 50%+ time savings.
Task automatabilityclaude-sonnet-51/5Physically or actively attending and engaging in continuing education classes is an activity performed by the human professional; AI cannot attend on their behalf to satisfy licensure requirements.attendance itself cannot be delegated to AI.
Adoption barriersclaude-haiku-4-5-202510013/5Many professional licensing bodies and employers require documented human attendance and genuine learning for continuing education credit; some jurisdictions have ethics/fraud concerns about delegating CE to machines, creating moderate friction despite no formal legal prohibition in most cases.
Adoption barriersclaude-sonnet-55/5Continuing education attendance is typically mandated by licensing boards and must be completed personally by the licensed/certified individual, making this a hard regulatory barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510015/5The inference and automation cost for an AI agent to attend online classes, complete assignments, and document hours is orders of magnitude cheaper than the loaded human wage for the same time investment.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot perform the actual task of attendance/certification, there is no valid cost comparison—AI cannot substitute at any cost.
Technical feasibility todayclaude-haiku-4-5-202510014/5Multiple AI systems can reliably automate educational platform interactions, content consumption, and basic assessments today; however, some continuing education programs include proctored exams or require verified human attendance, limiting full end-to-end automation in all contexts.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product attends CE classes for a professional; this is not a task category AI systems perform as a substitute for the human.

Report to supervisors, verbally or in writing, on patients' progress, attitudes, and behavior.

36

CI 25–48 · exposure 38 · augmentation 63 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI for clinical documentation and patient assessment remains cautious and pilot-stage; regulatory oversight, liability concerns, and institutional conservatism slow deep production deployment in occupational therapy settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially allied health and rehabilitation therapy, has been slower than finance or tech to adopt AI documentation tools at scale, though pilots are growing.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by drafting note templates, organizing observations, or flagging keywords from transcribed reports, helping the assistant structure and complete documentation faster while the human retains responsibility for accuracy and clinical judgment.
Augmentation potentialclaude-sonnet-54/5AI scribes and note-drafting assistants can meaningfully speed up writing progress reports, letting the assistant focus on verifying content and adding clinical nuance.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist with drafting progress notes from structured data (vital signs, test scores), but observing and interpreting nuanced patient behavior, attitudes, and emotional states requires human judgment and contextual understanding that current systems cannot reliably capture end-to-end.
Task automatabilityclaude-sonnet-53/5AI can draft progress notes and summaries from structured input or dictation, but synthesizing nuanced clinical observations about attitude and behavior still requires human judgment and verification before submission.
Adoption barriersclaude-haiku-4-5-202510014/5Healthcare liability, regulatory compliance (HIPAA, state licensure for clinical judgment), and organizational protocol requiring a qualified human to assess and vouch for patient status create substantial legal and procedural barriers to full automation.
Adoption barriersclaude-sonnet-53/5Clinical documentation often requires credentialed sign-off and is subject to healthcare record-keeping and privacy regulations, creating moderate compliance friction even though the writing itself isn't inherently licensed.
Cost vs. human wageclaude-haiku-4-5-202510012/5The oversight and correction required to ensure accurate, clinically sound reporting of patient progress and behavior—plus the liability exposure—currently makes AI assistance more costly than direct human reporting.
Cost vs. human wageclaude-sonnet-53/5AI transcription/summarization tools are cheap per note, but the necessary human review, correction, and accountability keep overall cost savings moderate rather than dramatic.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can generate template-based summaries from clinical data or transcribed observations, no deployed product reliably interprets complex behavioral and attitudinal nuances or replaces the observational and relational assessment that a trained assistant provides.
Technical feasibility todayclaude-sonnet-53/5Ambient documentation and clinical note-generation tools are deployed in some healthcare settings, but adoption in OT-specific contexts is narrower and accuracy still requires clinician review.

Assist educational specialists or clinical psychologists in administering situational or diagnostic tests to measure client's abilities or progress.

20

CI 20–20 · exposure 16 · augmentation 50 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and clinical settings, especially those involving diagnostic testing, adopt automation slowly due to regulatory constraints and professional gatekeeping. Most healthcare organizations continue to use human assistants for test administration rather than AI systems.
Sector adoption velocityclaude-sonnet-52/5Healthcare/allied health and educational assessment settings adopt AI slowly for direct client-facing diagnostic tasks, with pilots more common in documentation than test administration.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by scoring tests, flagging patterns in responses, documenting observations, and transcribing results—tasks that reduce administrative burden and allow the human assistant to focus on client interaction and clinical observation.
Augmentation potentialclaude-sonnet-53/5AI can help prepare materials, score standardized results, and draft progress notes, meaningfully aiding the assistant without replacing hands-on test administration.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could assist in test scheduling and result logging, administering diagnostic or situational tests requires real-time clinical observation, adaptive questioning, and rapport-building with clients—tasks that depend on human presence and judgment. Current AI cannot reliably conduct in-person assessments or handle the contingencies inherent in testing.
Task automatabilityclaude-sonnet-52/5This involves hands-on assistance with physical test administration, client observation, and interpersonal rapport that current AI cannot perform end-to-end; only scoring/documentation sub-steps are automatable.
Adoption barriersclaude-haiku-4-5-202510014/5Clinical diagnostic and situational testing is heavily regulated by professional standards, licensing boards, and test publishers who restrict administration to qualified humans. Liability concerns around misadministration of assessments create legal barriers to full automation.
Adoption barriersclaude-sonnet-54/5Working under licensed psychologists/specialists with client contact and diagnostic implications creates supervision requirements and liability concerns that constrain full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5An occupational therapy assistant's loaded wage is modest, and oversight by a supervisor (clinical psychologist or specialist) would still be required. AI tools for test support (transcription, scoring) add cost without eliminating the need for the human assistant, making the all-in ratio unfavorable.
Cost vs. human wageclaude-sonnet-52/5AI could cheaply assist with scoring or report generation, but the physical presence and hands-on test administration still require paid human labor, limiting overall cost savings.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably administers diagnostic or situational tests to clients in a clinical setting. AI may support test scoring or data entry, but the core task of direct test administration and observation is not performed by current systems in production.
Technical feasibility todayclaude-sonnet-51/5No deployed product administers situational/diagnostic tests to clients in clinical settings; this remains a human-delivered, supervised clinical activity.

Assemble, clean, or maintain equipment or materials for patient use.

19

CI 5–33 · exposure 13 · augmentation 25 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare is a laggard sector for physical automation of equipment maintenance; most occupational therapy settings operate in small clinics or hospitals with limited robotics infrastructure, and the high customization required per facility slows adoption of standardized systems.
Sector adoption velocityclaude-sonnet-51/5Healthcare support occupations involving physical equipment handling show minimal AI/robotic adoption; this is a low-digitization, physical-labor task.
Augmentation potentialclaude-haiku-4-5-202510012/5Current AI/robotic systems offer minimal augmentation for equipment assembly and maintenance; assistants rely on domain knowledge, visual inspection, and patient feedback rather than AI-assisted workflows. Tools exist for inventory tracking but not for assisting the core physical maintenance task.
Augmentation potentialclaude-sonnet-52/5AI could provide checklists, maintenance schedules, or inventory tracking software to support this task, but it offers minimal direct assistance with the physical cleaning/assembly itself.
Task automatabilityclaude-haiku-4-5-202510012/5While some routine assembly or cleaning steps could be partially automated (e.g., robotic arms for standardized equipment cleaning), the task requires judgment about patient-specific needs, equipment condition assessment, and safety verification that current AI systems cannot reliably perform end-to-end. Physical manipulation of diverse materials and equipment in varied states of disrepair remains challenging for current robots.
Task automatabilityclaude-sonnet-51/5This is a physical hands-on task involving handling, cleaning, and assembling tangible therapy equipment, which current AI systems cannot perform without robotic embodiment.》 No off-the-shelf AI system performs this physical labor.。
Adoption barriersclaude-haiku-4-5-202510014/5Healthcare settings have strong regulatory and liability barriers: equipment used for patient care must meet safety and sanitation standards, often requiring human certification and sign-off. Patient safety and infection control requirements create high error-cost asymmetry that deters full automation.
Adoption barriersclaude-sonnet-52/5No licensing specifically governs equipment cleaning/assembly, but there are practical/organizational barriers since it requires physical presence and manual dexterity in a clinical environment.
Cost vs. human wageclaude-haiku-4-5-202510012/5Specialized robotic systems capable of equipment assembly and maintenance are capital-intensive and require significant integration costs, making them more expensive than paying occupational therapy assistants for routine equipment care in most healthcare settings.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute for this physical task, so any AI-based approach (e.g., robotics) would be far more costly than a human assistant performing it directly.
Technical feasibility todayclaude-haiku-4-5-202510012/5No mature deployed products today reliably assemble, clean, or maintain diverse therapeutic equipment across settings. Robotic systems for specific repetitive cleaning exist in research/narrow industrial contexts, but general-purpose equipment maintenance in healthcare settings lacks production-scale automation that meets safety and quality standards.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product cleans or assembles physical therapy equipment in clinical settings today; this remains purely a manual task.

Instruct, or assist in instructing, patients and families in home programs, basic living skills, or the care and use of adaptive equipment.

16

CI 7–25 · exposure 13 · augmentation 63 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare settings, especially outpatient therapy clinics and home-care agencies, have low digitization relative to tech sectors and remain heavily constrained by regulatory and liability concerns; adoption of AI for direct patient instruction is minimal and largely experimental.
Sector adoption velocityclaude-sonnet-52/5Healthcare/rehab settings have historically slow, cautious AI adoption for direct patient care tasks, though administrative AI use is growing.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist by generating illustrated home-exercise guides, creating adaptive-equipment use videos, drafting written instructions, and providing visual demonstrations that OT assistants can then refine and deliver to patients, substantially reducing preparation time while the human retains assessment and personalization.
Augmentation potentialclaude-sonnet-53/5AI can generate instructional materials, home program handouts, and video/visual aids to support the assistant's teaching, but cannot replace in-person demonstration and correction.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could generate instructional content or demonstrate adaptive equipment use via video, the task fundamentally requires real-time responsiveness to patient questions, physical adaptation of techniques to individual mobility/cognitive levels, and emotional rapport-building—elements current AI systems cannot reliably deliver end-to-end. Partial automation (scripted videos, written guides) saves time but does not meet the ≥50% threshold when the core demand is personalized, responsive instruction.
Task automatabilityclaude-sonnet-51/5This requires hands-on demonstration, physical guidance, and real-time assessment of patient ability and safety with adaptive equipment, which current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Occupational therapy instruction carries liability risk if adaptive equipment is misused or home programs cause harm; regulatory oversight (state licensure of OT professionals, insurance billing codes) and the requirement for human assessment of patient safety and comprehension create substantial legal and organizational barriers to full substitution.
Adoption barriersclaude-sonnet-54/5Requires credentialed OTA involvement, liability for patient safety with equipment, and typically direct human interaction mandated by care plans and reimbursement rules.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI tools (LLMs, video generators, chatbots) cost pennies per interaction but require significant human oversight, customization, and follow-up to ensure safety and efficacy; the all-in cost approximates or exceeds the value of an assistant's time on routine instruction tasks.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the physical, in-person instruction component, so human labor cost remains the only viable option for actual delivery.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed products reliably perform the full instructional and adaptive components of this task. Chatbots and video platforms exist but lack the ability to assess patient comprehension in real time, modify instructions based on physical demonstration, or build the trust relationship necessary for effective patient compliance.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously instructs patients/families in home programs or equipment use; this remains a human clinical task.

Monitor patients' performance in therapy activities, providing encouragement.

15

CI 5–25 · exposure 13 · augmentation 38 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare, especially rehabilitation and therapy, adopts automation cautiously; therapy assistant tasks remain largely manual and labor-intensive in practice, with minimal evidence of AI agent deployment in real clinical settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare/rehabilitation settings show slower AI adoption for hands-on patient care tasks compared to administrative or diagnostic support functions.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by flagging performance outliers, suggesting encouragement prompts, or tracking progress data, raising therapist awareness and efficiency; however, the core task of reading patient state and delivering authentic encouragement remains human-centered.
Augmentation potentialclaude-sonnet-52/5AI could potentially assist with tracking metrics or generating progress notes to support the assistant, but it does not enhance the core act of monitoring and encouraging patients during sessions.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could track performance metrics and generate encouragement scripts, the task fundamentally requires real-time perception of nuanced patient behavior, adaptive feedback calibrated to individual psychological state, and authentic human connection—elements that current AI cannot reliably substitute for at equal quality outcomes in a therapeutic context.
Task automatabilityclaude-sonnet-51/5This requires real-time physical presence, human rapport, and hands-on monitoring of a patient's physical movements and emotional state, which current AI cannot perform end-to-end.'
Adoption barriersclaude-haiku-4-5-202510014/5Therapeutic contexts carry high liability for incorrect feedback, regulatory oversight of patient safety, clinical licensing requirements for practitioners, and strong patient preference for human contact during vulnerable moments; these create substantial adoption friction.
Adoption barriersclaude-sonnet-54/5Direct patient care in therapy settings typically requires supervised, credentialed personnel and involves liability and safety concerns around physical monitoring and encouragement.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of monitoring AI (sensors, vision, LLM responses, oversight) into therapy workflows would require significant infrastructure and oversight costs that approach or exceed the loaded wage of therapy assistants in most markets.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical, relational task, so any comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system reliably monitors complex motor/cognitive therapy performance and delivers personalized, contextually appropriate encouragement in clinical settings; research prototypes exist but lack production validation and clinical safety evidence.
Technical feasibility todayclaude-sonnet-51/5No deployed product monitors in-person therapy performance and delivers human encouragement in a clinical setting; this remains outside current product capability.

Communicate and collaborate with other healthcare professionals involved with the care of a patient.

14

CI 7–20 · exposure 8 · augmentation 50 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains a heavily regulated, conservative sector with slow AI adoption in clinical workflows. While some health systems pilot documentation and communication tools, actual autonomous participation in clinical team coordination is rare and faces institutional resistance.
Sector adoption velocityclaude-sonnet-52/5Healthcare is a comparatively slower-adopting sector for full automation of interpersonal clinical coordination, though AI tools for documentation are spreading.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can assist with drafting communications, organizing information summaries, and flagging scheduling conflicts, moderately improving an OT assistant's coordination efficiency. However, the human's judgment, accountability, and relational presence remain central to effective multidisciplinary collaboration.
Augmentation potentialclaude-sonnet-53/5AI can assist with drafting notes, summarizing patient records, or scheduling communications, improving efficiency, but the core collaborative interaction remains human-led.
Task automatabilityclaude-haiku-4-5-202510012/5Limited automation is possible for routine scheduling or status updates, but genuine collaboration requires nuanced interpersonal communication, context-sensitivity, and real-time responsiveness that current AI systems cannot reliably replicate. Asynchronous message drafting could assist, but the core collaborative decision-making remains human-dependent.
Task automatabilityclaude-sonnet-51/5Interprofessional communication and care coordination requires real-time judgment, relationship building, and physical presence in clinical settings that current AI cannot replicate end-to-end.'
Adoption barriersclaude-haiku-4-5-202510014/5Healthcare licensure, HIPAA compliance, liability for clinical decisions, and professional standards of care create substantial legal and regulatory barriers. Healthcare institutions require human professionals to sign off on care coordination decisions and take responsibility for patient outcomes.
Adoption barriersclaude-sonnet-54/5Care coordination often involves licensed professionals making clinical judgments and documentation with legal/regulatory accountability, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI can generate template communications at near-zero marginal cost, but meaningful care coordination requires human oversight and decision-making, so the all-in cost of AI-assisted coordination remains high relative to the cost of direct human participation.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this collaborative task, so cost comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system reliably conducts independent healthcare collaboration with clinical teams in production environments today. While chatbots can draft communications, they cannot authentically participate in multidisciplinary case conferences or handle the liability and accountability expectations of live clinical coordination.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously handles interdisciplinary healthcare team communication and collaboration on patient care; this remains a human-driven activity.

Teach patients how to deal constructively with their emotions.

14

CI 11–16 · exposure 5 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and mental health remain relatively conservative in AI adoption outside diagnostic imaging and administrative tasks. Therapeutic delivery automation is still rare in production; most therapists and assistants continue to deliver emotional coaching directly.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially hands-on allied health and behavioral support, has been slow to adopt AI for direct patient emotional counseling due to regulatory, ethical, and trust concerns.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist by generating psychoeducational materials, suggesting evidence-based coping strategies, or helping structure session notes, allowing the human therapist to focus on rapport and personalized guidance. The human remains the core therapeutic agent.
Augmentation potentialclaude-sonnet-53/5AI can help OTAs prepare psychoeducational materials, suggest coping strategies, or role-play scenarios, providing moderate support while the human retains the core therapeutic relationship.
Task automatabilityclaude-haiku-4-5-202510011/5Teaching emotional coping requires genuine understanding of individual patient psychology, rapport-building, and real-time adaptive responsiveness to emotional states. Current AI cannot reliably conduct this kind of personalized emotional coaching or detect and respond to nuanced emotional cues in ways that meet clinical standards.
Task automatabilityclaude-sonnet-51/5Teaching patients to constructively manage emotions requires real-time empathetic human interaction, trust-building, and adaptive judgment that current AI cannot replicate end-to-end in a clinical therapeutic context.
Adoption barriersclaude-haiku-4-5-202510014/5Therapeutic interventions involving emotional support face significant regulatory, licensure, and liability barriers. An occupational therapy assistant must be credentialed, and substitution by unaccountable AI raises malpractice and patient-safety concerns that prevent straightforward automation.
Adoption barriersclaude-sonnet-54/5This is a clinical task within a regulated healthcare role typically requiring supervision and documentation, with high liability if emotional/behavioral guidance is mishandled, though it may not always require full independent licensure.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-based emotion-coaching tools remain relatively expensive to develop and integrate into clinical workflows, while occupational therapy assistants have moderate wage costs. Cost parity is not yet achieved at clinical quality.
Cost vs. human wageclaude-sonnet-52/5While AI chatbots are cheap per interaction, they cannot yet deliver equivalent clinical outcomes, so effective cost-per-quality-adjusted-output remains higher than a trained assistant despite lower nominal cost.
Technical feasibility todayclaude-haiku-4-5-202510012/5While chatbots can deliver generic psychoeducational content about emotion management, no deployed AI product reliably teaches emotional coping to actual patients in clinical settings with the individualization and safety this task demands. Some research and limited pilots exist but lack production-scale evidence.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently delivers this clinical, emotionally-attuned therapeutic instruction as a substitute for a trained OTA in production healthcare settings.

Evaluate the daily living skills or capacities of clients with physical, developmental, or mental health disabilities.

13

CI 0–25 · exposure 13 · augmentation 50 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare and rehabilitation are slow-moving sectors in AI adoption for clinical decision-making. While administrative tools are spreading, autonomous clinical evaluation remains rare; most organizations use AI only for scheduling or documentation support, not for replacing skilled evaluation.
Sector adoption velocityclaude-sonnet-51/5Healthcare/rehabilitation therapy is a slow-adopting, highly regulated, physically-grounded sector with minimal AI deployment for hands-on assessment tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could moderately assist an OTA by analyzing video of observed tasks, organizing assessment data, prompting for missed items in a checklist, or generating preliminary reports for clinical review. However, the core interpretive and observational work remains human-driven, limiting transformative productivity gain.
Augmentation potentialclaude-sonnet-53/5AI can assist with documentation, standardized assessment scoring, and generating summary reports, but the core evaluative interaction remains human-driven.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could potentially assist with documenting observations or analyzing standardized test responses, the task fundamentally requires direct observation of a client's actual performance in activities (dressing, grooming, cooking, etc.) combined with clinical judgment about functional capacity, safety, and disability-specific context. Current AI systems cannot reliably assess physical ability, fine motor skills, safety awareness, or mental health presentation through autonomous observation.
Task automatabilityclaude-sonnet-51/5Direct clinical observation, hands-on functional testing, and interpersonal rapport with disabled clients cannot be replicated end-to-end by current AI systems.
Adoption barriersclaude-haiku-4-5-202510014/5Regulatory and professional barriers are substantial: only licensed OTAs or occupational therapists can conduct standardized occupational therapy evaluations in most jurisdictions; liability for misassessment is high (risk of unsafe discharge, missed adaptive needs); and clients' disabilities often require human judgment, rapport, and adaptive assessment methods that AI cannot reliably provide.
Adoption barriersclaude-sonnet-55/5Requires licensed/credentialed practitioner judgment under supervision, involves direct physical contact and liability-sensitive clinical decisions, making substitution legally and practically barred.
Cost vs. human wageclaude-haiku-4-5-202510012/5The loaded cost of an OTA (typically $30–45k/year) is relatively low compared to the infrastructure, oversight, validation, and liability costs required to deploy AI for disability assessment. The clinical and legal stakes make human oversight mandatory, negating cost advantage.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this hands-on evaluation, so no meaningful cost comparison favors AI.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI product performs independent occupational therapy evaluation. While video analysis or screening tools exist in research settings, they lack the clinical validity, safety coverage, and scope required for real diagnostic or therapeutic decision-making. Organizations still require qualified humans to conduct these evaluations.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs physical/functional client evaluations autonomously; this remains a research-stage concept at best.

Design, fabricate, or repair assistive devices or make adaptive changes to equipment or environments.

11

CI 5–16 · exposure 8 · augmentation 38 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Occupational therapy remains a relatively low-digitization sector with fragmented settings (clinics, schools, homes) and strong preference for personalized human assessment. Adoption of autonomous AI in device design and fabrication is minimal compared to information-sector workflows.
Sector adoption velocityclaude-sonnet-51/5Occupational therapy involves physical, hands-on care with low digitization and minimal AI deployment for this specific fabrication/repair task.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist in generating design options, providing reference materials, or documenting modifications, helping occupational therapy assistants work more efficiently. However, the core task of physical fabrication and client-specific adaptation remains largely human-dependent.
Augmentation potentialclaude-sonnet-52/5AI/CAD tools and 3D printing software can assist in designing device specifications or generating models, but the core fabrication, fitting, and environmental adaptation still rely on human skill.
Task automatabilityclaude-haiku-4-5-202510012/5This task requires physical fabrication, spatial reasoning about individual client needs, and creative problem-solving. While AI can assist with design ideation and documentation, the hands-on fabrication and real-world testing of assistive devices remain fundamentally manual and context-dependent, preventing 50% time savings across the full task.
Task automatabilityclaude-sonnet-51/5Designing, fabricating, and repairing physical assistive devices requires hands-on measurement, physical craftsmanship, and iterative fitting with the patient, none of which AI can perform end-to-end today.
Adoption barriersclaude-haiku-4-5-202510014/5Occupational therapy assistants must work under supervision of licensed occupational therapists in most jurisdictions, and assistive devices carry liability and safety-critical requirements. Client safety and the need for hands-on fit and customization create strong adoption barriers.
Adoption barriersclaude-sonnet-54/5Fabrication and fitting of assistive devices typically requires a licensed practitioner's judgment and hands-on interaction with the patient, creating strong professional and liability barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of specialized fabrication equipment, materials, and the human oversight required to ensure safety and client fit means current AI solutions are more expensive than employing a skilled occupational therapy assistant to perform this task directly.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute for the physical labor and clinical judgment involved, so human labor remains the only cost-effective option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs end-to-end design, fabrication, or repair of custom assistive devices autonomously. AI tools exist for design drafting and visualization, but the integration of client assessment, material selection, fabrication execution, and safety validation requires human expertise in production settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously fabricates or fits assistive devices or modifies physical environments for patients; this remains a manual, therapist-driven task.

Select therapy activities to fit patients' needs and capabilities.

10

CI 0–20 · exposure 8 · augmentation 50 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare and therapy are laggard sectors in AI adoption for clinical decisions; occupational therapy practices are typically small, resource-constrained, and cautious about automation of patient-facing judgments. Regulatory and liability concerns slow meaningful adoption.
Sector adoption velocityclaude-sonnet-52/5Healthcare/rehabilitation settings adopt AI slowly for hands-on clinical decision-making, with adoption concentrated in documentation and administrative support rather than treatment selection.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could usefully assist by suggesting activity options based on condition databases, outcomes research, or patient profiles, helping therapists explore choices faster. However, the human therapist must remain the decision-maker given the need to assess patient readiness, motivation, and safety.
Augmentation potentialclaude-sonnet-53/5AI can help by suggesting evidence-based activity options, summarizing patient history, or tracking outcomes, but final activity selection still depends on the assistant's in-person judgment.
Task automatabilityclaude-haiku-4-5-202510012/5Selecting therapy activities requires understanding individual patient capabilities, limitations, and treatment goals—nuanced clinical judgment that current AI cannot reliably perform end-to-end. While AI could suggest activities from a database based on stated conditions, matching them to a specific patient's evolving capabilities and psychosocial context demands human assessment.
Task automatabilityclaude-sonnet-51/5Selecting therapy activities requires hands-on clinical judgment about a specific patient's physical, cognitive, and emotional state, which current AI cannot perceive or assess directly.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: occupational therapy is a regulated profession in most jurisdictions, therapists must assess patient capabilities and document selection rationale, and liability for inappropriate activity selection rests with the supervising licensed therapist. Direct patient contact and clinical accountability create hard friction.
Adoption barriersclaude-sonnet-55/5This is a licensed clinical task requiring supervision by an OT and hands-on patient assessment, with strong liability and regulatory requirements for qualified human judgment.
Cost vs. human wageclaude-haiku-4-5-202510012/5Even if AI could perform activity selection, the integration, clinical review, and liability costs would approach or exceed the wage of an occupational therapy assistant, especially given the low error-tolerance in patient care contexts.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this task, so cost comparison favors the human by default; any AI support tool adds cost rather than replacing labor.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs this task; it would require real-time integration with patient assessment data, therapist oversight, and clinical accountability. Existing AI systems lack the clinical depth and organizational embedding necessary for production use in therapy settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously selects and administers individualized therapy activities in clinical practice today; this remains a human clinical judgment task.

Demonstrate therapy techniques, such as manual or creative arts or games.

9

CI 5–14 · exposure 8 · augmentation 38 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare, particularly therapy services, shows slow adoption of automation for direct patient care tasks. Occupational therapy is a hands-on, relationship-dependent field with strong preference for human practitioners and limited digitization of core service delivery.
Sector adoption velocityclaude-sonnet-51/5Healthcare/rehabilitation therapy delivery is a low-digitization, physically embodied sector with minimal AI agent deployment for hands-on patient-facing tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by suggesting technique variations, providing video references, or generating adaptive practice modifications based on patient needs, but the core demonstration task requires human presence and real-time adjustment that an AI co-worker could partially enhance but not fully augment.
Augmentation potentialclaude-sonnet-52/5AI can help plan session content, suggest activities, or provide video references for the therapist to review, but it offers little direct assistance during the actual physical demonstration with a patient.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could generate descriptions or videos of therapy techniques, the task requires real-time, adaptive physical demonstration and rapport-building with patients that current AI cannot reliably perform. The interactive, embodied nature of demonstrating techniques to individuals with varying needs falls far short of the 50% time-saving threshold.
Task automatabilityclaude-sonnet-51/5This requires physically demonstrating manual therapy techniques, hands-on physical guidance, and interactive engagement with patients through games and creative activities—none of which current AI systems can perform end-to-end without a physical embodiment.
Adoption barriersclaude-haiku-4-5-202510014/5Strong barriers exist: occupational therapy assistants operate under supervision of licensed occupational therapists who typically oversee or approve demonstration methods; patient safety, liability, and the requirement for human therapeutic presence and adjustment during demonstrations are regulatory and practical constraints.
Adoption barriersclaude-sonnet-54/5OT assistants often work under supervision requirements and licensing frameworks, and hands-on physical therapy demonstration inherently requires human presence and touch, creating strong structural barriers to substitution.
Cost vs. human wageclaude-haiku-4-5-202510011/5Deploying robotic or VR systems capable of demonstrating therapeutic techniques would require significant capital investment and ongoing maintenance, far exceeding the cost of employing an occupational therapy assistant to demonstrate in-person.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical, in-person task, so cost comparison favors the human by default since no AI alternative exists to displace the labor cost.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs live, adaptive therapeutic technique demonstrations. Robotic systems and VR exist but are narrow in scope and not standard in occupational therapy settings; humans remain the only practical demonstrators in clinical care.
Technical feasibility todayclaude-sonnet-51/5No deployed product physically demonstrates manual therapy or interactive game-based techniques to patients; this remains outside current robotics/AI product capability at any meaningful scale.

Implement, or assist occupational therapists with implementing, treatment plans designed to help clients function independently.

7

CI 0–14 · exposure 8 · augmentation 50 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare, especially rehabilitation, has been slow to adopt AI agents in direct service delivery due to regulatory, liability, and client-contact requirements. Adoption remains at the pilot and documentation-support stage, not production-level task replacement.
Sector adoption velocityclaude-sonnet-51/5Healthcare/allied health physical therapy services show slow AI adoption for hands-on care delivery, being a low-digitization, physically-embodied sector.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by generating client progress summaries, suggesting evidence-based modifications to written plans, or helping schedule and track outcomes, which would ease administrative burden. However, augmentation is limited to planning and documentation periphery, not the core interactive therapeutic work itself.
Augmentation potentialclaude-sonnet-53/5AI can assist with documentation, progress tracking, exercise planning suggestions, and patient education materials, offering moderate productivity support around the core hands-on task.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could generate or refine written treatment plans, the hands-on implementation—adjusting adaptive equipment, providing real-time encouragement, monitoring patient response, and modifying interventions mid-session—requires physical presence and human judgment that current AI cannot provide. AI might assist in documenting or planning, but not autonomously execute the core therapeutic work.
Task automatabilityclaude-sonnet-51/5This task requires hands-on physical guidance, manual techniques, and real-time adaptive interaction with clients' bodies and behaviors, which current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Occupational therapy assistants work under license delegation and direct supervision; their work involves direct client contact, safety judgments, and liability around physical intervention. Regulatory frameworks require a licensed occupational therapist to authorize and oversee treatment, creating a hard barrier to full automation.
Adoption barriersclaude-sonnet-55/5OTA practice is licensed and requires supervision by occupational therapists, with direct physical client contact and liability considerations that legally require a credentialed human.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI tools that assist with documentation or planning are still marginal in cost savings relative to the loaded wage of an occupational therapy assistant. The labor cost per client contact hour remains far below the cost of AI infrastructure, oversight, and integration for this hands-on role.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing the physical treatment delivery, so cost comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product today can reliably perform in-person occupational therapy assistance or treatment implementation; this remains a human-delivered service. Research into teletherapy decision support exists, but production systems do not autonomously implement treatment with clients.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs hands-on occupational therapy implementation; this remains firmly in the physical/human-contact domain outside AI product scope.

Work under the direction of occupational therapists to plan, implement, or administer educational, vocational, or recreational programs that restore or enhance performance in individuals with functional impairments.

7

CI 0–14 · exposure 8 · augmentation 50 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare and therapeutic sectors exhibit slow, cautious AI adoption, with strong regulatory oversight and resistance to automating direct patient care; pilot programs exist but production displacement of therapy assistants remains minimal.
Sector adoption velocityclaude-sonnet-51/5Healthcare/rehabilitation services involving direct physical patient care show slow AI adoption due to regulatory, safety, and hands-on constraints.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist therapy assistants by drafting program outlines, suggesting adaptive activities, or organizing patient progress data, moderately raising documentation and planning efficiency while the human remains central to implementation and patient interaction.
Augmentation potentialclaude-sonnet-53/5AI can help with documentation, exercise-plan suggestions, progress tracking, and educational material generation, but the core administration of therapy remains human-led.
Task automatabilityclaude-haiku-4-5-202510012/5While AI could assist with program planning (drafting activity schedules, suggesting evidence-based interventions), the core requirement of implementing and administering programs for individuals with functional impairments demands real-time human responsiveness, safety monitoring, and adaptive modification based on patient reaction—tasks current AI cannot reliably perform end-to-end.
Task automatabilityclaude-sonnet-51/5This requires hands-on physical assistance, direct client interaction, adaptive real-time judgment, and supervised clinical decision-making that current AI cannot perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Healthcare and therapeutic contexts impose significant legal, licensing, and liability barriers; occupational therapy is regulated, and direct service delivery to vulnerable populations typically requires credentialed human professionals to ensure duty of care and legal accountability.
Adoption barriersclaude-sonnet-55/5Occupational therapy assistants are licensed/certified and must work under supervision of a licensed OT, with legal and liability requirements mandating human involvement in patient care.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of implementing AI-based program oversight plus required human supervision and liability would likely exceed the loaded wage of an occupational therapy assistant, making economic displacement implausible today.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this physical, supervised task, so cost comparison favors the human by default.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs the full task of administering therapeutic or vocational programs to individuals with impairments; this requires human judgment in real-time interaction, safety assessment, and personalized adaptation that exceeds current AI capabilities.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product administers or implements hands-on therapeutic programs; this remains firmly in the human physical-care domain.

Attend care plan meetings to review patient progress and update care plans.

4

CI 0–7 · exposure 0 · augmentation 38 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare remains a laggard sector for clinical task automation due to regulatory constraints, liability concerns, and the requirement for licensed professional judgment. Care planning meetings are not being automated in production settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially allied health/rehab settings, is a slower-adopting sector for AI displacement of clinical judgment tasks, though administrative AI tools are creeping in.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could modestly assist by preparing summaries of prior notes or flagging missing data elements pre-meeting, but the collaborative, real-time nature of care plan discussions limits meaningful augmentation of the meeting itself.
Augmentation potentialclaude-sonnet-53/5AI can assist with meeting transcription, summarization of patient progress notes, and drafting updated care plan language, saving the assistant preparation time even though the human still attends and decides.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires real-time interpersonal coordination, nuanced clinical judgment about patient progress, and collaborative decision-making with multidisciplinary teams. Current AI cannot reliably participate as a meeting attendee or contribute meaningfully to care plan adjustments based on subtle clinical cues and contextual patient information.
Task automatabilityclaude-sonnet-51/5This requires physical presence, live discussion, clinical judgment, and interpersonal collaboration with a care team, none of which current AI can perform end-to-end.rehab-specific decision-making is central.
Adoption barriersclaude-haiku-4-5-202510015/5Healthcare regulations (HIPAA, state licensure laws) and standard of care requirements legally mandate that qualified human professionals attend and contribute to care planning decisions. Liability and accreditation standards create hard barriers to substitution.
Adoption barriersclaude-sonnet-54/5Clinical documentation and care plan updates typically require credentialed practitioner involvement and accountability, and many settings mandate licensed staff participation in care planning.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI systems cannot currently perform this task, making cost comparison moot. The infrastructure and oversight required to validate AI-generated clinical contributions would exceed the cost of human attendance.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this task, so cost comparison favors the human by default; any AI note-taking aid doesn't replace the meeting itself.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product can autonomously attend and participate in care plan meetings with the clinical accuracy and judgment required. While AI can summarize documents or draft notes, it cannot serve as a substitute meeting participant making real-time clinical contributions.
Technical feasibility todayclaude-sonnet-51/5No deployed product attends multidisciplinary care meetings or updates patient-specific care plans autonomously; this remains outside product scope today.

Aid patients in dressing and grooming themselves.

3

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

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare assistant roles involve significant in-person, hands-on work in settings with strict regulatory oversight and minimal digital-first adoption patterns. Robots for personal care remain research-stage and are not being deployed at scale.
Sector adoption velocityclaude-sonnet-51/5Healthcare/rehabilitation physical care settings show very slow adoption of automation for hands-on patient care tasks compared to digital/information-based sectors.
Augmentation potentialclaude-haiku-4-5-202510012/5AI could potentially assist with scheduling, documenting progress, or suggesting adaptive techniques, but the core physical and interpersonal elements of aiding dressing and grooming resist meaningful AI augmentation of the hands-on worker.
Augmentation potentialclaude-sonnet-52/5AI can support scheduling, care planning, or documentation around this task, but offers minimal direct assistance during the actual physical act of dressing/grooming.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires physical contact, manual dexterity, and real-time adaptation to patient needs and abilities. Current AI systems lack embodied robotics capable of reliably assisting with clothing fasteners, grooming tools, and the fine motor coordination needed for safe, dignified personal care.
Task automatabilityclaude-sonnet-51/5This is a hands-on physical caregiving task requiring direct physical assistance and adaptive judgment; no current AI system can perform physical dressing/grooming assistance.
Adoption barriersclaude-haiku-4-5-202510015/5Direct physical contact with vulnerable patients, strong regulatory and licensing requirements for occupational therapy practice, patient safety and dignity concerns, and legal liability create hard barriers to automation. Human licensure and hands-on oversight are legally and practically required.
Adoption barriersclaude-sonnet-54/5Direct physical patient care involves safety, liability, dignity, and often licensure/certification requirements for the assistant, creating strong barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5A robot or AI system capable of safe physical assistance with dressing and grooming would require substantial hardware, maintenance, and liability infrastructure—far exceeding the loaded wage of a human occupational therapy assistant.
Cost vs. human wageclaude-sonnet-51/5Physical assistive robots for dressing are experimental and far more costly and slower than a human aide performing the same task, with no viable production deployment.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs this task reliably in real care settings. While research robots exist, they do not demonstrably provide production-grade assistance for dressing and grooming at the scale or reliability required in occupational therapy.
Technical feasibility todayclaude-sonnet-51/5No deployed AI or robotic product reliably performs physical dressing/grooming aid for patients in clinical or home settings today; this remains research-stage robotics at best.

Transport patients to and from the occupational therapy work area.

3

CI 0–5 · exposure 0 · augmentation 13 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare and therapy settings have not adopted AI-driven patient transport; the sector remains highly human-dependent, and robotics adoption in clinical environments is nascent and confined to research or pilot programs.
Sector adoption velocityclaude-sonnet-51/5Healthcare physical care tasks show minimal AI/robotic adoption; this is a low-digitization, hands-on task with no meaningful automation trend in production settings.
Augmentation potentialclaude-haiku-4-5-202510012/5Limited augmentation potential; AI could assist with routing or scheduling logistics, but the core physical transport task remains human-performed and offers minimal opportunity for AI-human joint productivity gains in current systems.
Augmentation potentialclaude-sonnet-51/5AI offers essentially no assistance for the physical act of transporting a patient between locations.
Task automatabilityclaude-haiku-4-5-202510011/5Transporting patients physically requires embodied navigation, awareness of patient mobility constraints, safety handling, and real-time environmental adaptation—capabilities that current AI systems lack entirely without specialized robotics, which are not yet deployed in clinical settings at scale.
Task automatabilityclaude-sonnet-51/5Physically transporting patients (e.g., via wheelchair) requires embodied physical action that no current off-the-shelf AI or general-purpose robot can reliably perform in clinical settings.
Adoption barriersclaude-haiku-4-5-202510015/5Patient transport involves direct human contact, physical safety liability, and regulatory oversight under healthcare and worker safety frameworks; healthcare facilities face liability concerns and legal requirements for human judgment in patient handling.
Adoption barriersclaude-sonnet-54/5Patient safety, liability for falls or injury during transport, and the need for trained personnel to handle mobility-impaired patients create strong practical and regulatory barriers to automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5Current mobile manipulation systems capable of patient interaction cost hundreds of thousands of dollars per unit, plus integration and maintenance, far exceeding the loaded cost of a therapy assistant performing patient transport.
Cost vs. human wageclaude-sonnet-51/5No viable AI/robotic solution exists for this task at any meaningful scale, so the human remains the only cost-effective option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI system today reliably performs patient transport in healthcare environments; mobile manipulation robots capable of safely assisting or replacing human transport are research-stage and not in production use in occupational therapy clinics.
Technical feasibility todayclaude-sonnet-51/5There are no deployed products performing patient transport in occupational therapy settings; this remains a physical, human-performed task in essentially all clinics today.

Alter treatment programs to obtain better results if treatment is not having the intended effect.

1

CI 0–3 · exposure 0 · augmentation 50 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare sectors, particularly rehabilitation services, remain conservative in automating clinical decision-making; adoption of AI for treatment program modification is minimal and heavily restricted to pilot research settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare/allied health sectors show slow, cautious AI adoption for clinical decision-making tasks, with most AI use confined to documentation or administrative support rather than treatment planning.
Augmentation potentialclaude-haiku-4-5-202510013/5AI could assist by flagging patterns in patient progress data or suggesting evidence-based intervention modifications for clinician review, but the human practitioner must retain full authority over treatment decisions.
Augmentation potentialclaude-sonnet-53/5AI can help by analyzing patient progress data, flagging lack of improvement, and suggesting evidence-based alternative interventions for the therapist to consider and validate.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires real-time clinical judgment, assessment of patient response, understanding of complex medical history, and ethical decision-making about treatment modifications—capabilities that current AI systems cannot reliably perform end-to-end in clinical settings.
Task automatabilityclaude-sonnet-51/5This requires hands-on clinical judgment, real-time patient assessment, and physical/emotional interaction to determine why a treatment isn't working and how to adjust it—far beyond current AI capabilities.
Adoption barriersclaude-haiku-4-5-202510015/5Strong regulatory barriers exist: occupational therapy treatment modification decisions fall under licensed scope of practice, and liability for adverse outcomes from program alterations rests with the credentialed practitioner.
Adoption barriersclaude-sonnet-55/5Treatment modification is a licensed clinical function requiring professional judgment and accountability under healthcare regulations; an unlicensed AI system cannot legally alter treatment plans.
Cost vs. human wageclaude-haiku-4-5-202510011/5The task involves specialized clinical expertise and liability that necessitates human involvement; any AI assistance still requires a paid occupational therapy professional to validate and implement changes.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing this task, so AI cost cannot be meaningfully compared—the human professional remains the only option.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably modifies occupational therapy treatment programs based on patient response without human clinical oversight; this remains a domain requiring licensed practitioner judgment.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product independently modifies occupational therapy treatment plans based on patient response; this remains a clinical decision made by licensed practitioners.

Maintain and promote a positive attitude toward clients and their treatment programs.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare and therapeutic settings remain among the slowest to adopt autonomous AI agents for direct client interaction, with strong organizational and regulatory resistance to replacing human emotional presence in care.
Sector adoption velocityclaude-sonnet-51/5Healthcare/therapy settings show slow adoption of AI for interpersonal and emotional aspects of care, and this task is not a target for automation efforts.
Augmentation potentialclaude-haiku-4-5-202510012/5AI tools might assist by suggesting communication strategies or reminders for the human assistant, but the core task—authentic emotional support and positive relational presence—is not meaningfully augmented by current systems; the human must provide the attitude itself.
Augmentation potentialclaude-sonnet-52/5AI might help with scheduling, documentation, or reminders that indirectly support positive client interactions, but it does not meaningfully enhance the practitioner's attitude or rapport itself.
Task automatabilityclaude-haiku-4-5-202510011/5Maintaining and promoting a positive attitude is an inherently human interpersonal and emotional act that requires genuine empathy, judgment, and relational presence. Current AI systems cannot authentically model or deliver the emotional engagement and personalized encouragement that characterizes this task.
Task automatabilityclaude-sonnet-51/5Maintaining a genuine positive attitude toward clients is an interpersonal, emotional disposition inherent to the human practitioner and cannot be performed by AI on someone's behalf.'
Adoption barriersclaude-haiku-4-5-202510015/5Therapeutic relationships and emotional support require a licensed or credentialed human present; healthcare regulations, patient dignity standards, and the fiduciary nature of occupational therapy create hard barriers to AI substitution for attitude and encouragement.
Adoption barriersclaude-sonnet-55/5This requires direct human presence, empathy, and professional relationship-building that is core to licensed clinical practice and cannot be delegated to a machine.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of integrating AI systems to simulate positive attitude and engagement, combined with required human oversight to ensure therapeutic appropriateness, exceeds the cost of direct human occupational therapy assistant labor.
Cost vs. human wageclaude-sonnet-51/5There is no AI alternative delivering this attitudinal function, so cost comparison is moot; the human is the only source of this output.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product reliably performs genuine attitudinal support or emotional encouragement in therapeutic contexts. AI chatbots can produce encouraging text, but they cannot maintain a sustained, authentic therapeutic relationship or respond with genuine empathetic understanding.
Technical feasibility todayclaude-sonnet-51/5No deployed AI product embodies or substitutes for a therapist's personal attitude or rapport with clients; this is not a task AI systems address.

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.