Advanced Practice Psychiatric Nurses

29-1141.02

Median wage $97,550/yr3,379,720 employed (US)Rank #681 of 923 scored · top 74% by substitution

Assess, diagnose, and treat individuals and families with mental health or substance use disorders or the potential for such disorders. Apply therapeutic activities, including the prescription of medication, per state regulations, and the administration of psychotherapy.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution20

Exposure21

Augmentation62

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

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

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

Tasks on the substitution scale

24 rated tasks, binned by substitution score.

Position among all scored occupations

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

Tasks with substitution ≥ 70

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%23

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

Cost vs. human wagew 15%24

panel mean rating 2.0/5 → substitution pressure 24/100

Adoption barriersw 20%inverted — strong barriers lower the score12

panel mean rating 4.5/5 (barrier strength) → substitution pressure 12/100

Sector adoption velocityw 10%23

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

Task breakdown (24 tasks)

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

Monitor the use and status of medical and pharmaceutical supplies.

CI 50–92 · exposure 75 · augmentation 63 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Healthcare organizations, especially hospital systems and large psychiatric facilities, have already adopted automated inventory and pharmacy management systems at scale over the past decade. Deployment is standard practice in digital-native health systems.
Sector adoption velocity	claude-sonnet-5	3/5	Healthcare has moderate digitization with automated dispensing and inventory systems common in hospitals, but psychiatric outpatient and smaller practice settings lag in adoption.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist nurses by providing real-time supply alerts, predictive low-stock warnings, and automated reorder suggestions, thereby reducing time spent on manual checks. However, final physical verification and clinical judgment about supply adequacy remain with the human.
Augmentation potential	claude-sonnet-5	4/5	AI-enabled inventory systems significantly reduce manual counting and reordering effort, giving nurses real-time visibility and alerts while they retain responsibility for clinical and compliance decisions.
Task automatability	claude-haiku-4-5-20251001	5/5	Inventory tracking, supply-level monitoring, and stock status updates are routine data-logging and retrieval tasks that current AI and database systems can fully automate with significant time savings. This task involves no clinical judgment and maps directly to existing inventory management software with barcode/RFID integration.
Task automatability	claude-sonnet-5	3/5	Inventory and supply tracking is largely a data-monitoring task that AI-linked systems (inventory management software, RFID, EHR integration) can substantially automate, though clinical judgment about controlled substance discrepancies still requires human oversight.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Most healthcare organizations already use electronic health records and inventory systems, so regulatory and technical adoption barriers are low. However, institutional inertia, legacy system integration, and requirement for periodic human verification of physical counts create moderate friction.
Adoption barriers	claude-sonnet-5	3/5	Controlled substance tracking is subject to DEA and state regulations requiring documented human accountability, creating moderate compliance friction even though the underlying tracking can be automated.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated inventory systems cost pennies per transaction and require minimal human oversight, whereas a psychiatric nurse's loaded wage is $70–100+ per hour. The cost differential is at least 100-fold for straightforward supply monitoring.
Cost vs. human wage	claude-sonnet-5	3/5	Automated inventory systems are cost-effective at scale but require upfront integration, hardware (dispensing cabinets, barcode scanners), and ongoing maintenance, making savings moderate rather than order-of-magnitude for smaller practices.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Production-grade inventory management systems, pharmacy supply-tracking software, and automated stock-monitoring platforms are widely deployed in healthcare organizations today. These tools reliably track supply status, alert on low inventory, and generate reports without material error rates in routine use.
Technical feasibility today	claude-sonnet-5	3/5	Pharmacy inventory management systems and automated dispensing cabinets are widely deployed in clinical settings and reliably track supply levels, though full end-to-end monitoring including exception handling still involves staff review.

Interpret diagnostic or laboratory tests, such as electrocardiograms (EKGs) and renal functioning tests.

CI 29–67 · exposure 50 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Healthcare organizations are piloting and deploying AI for test interpretation, but adoption varies widely by institution and is often confined to screening or decision-support roles rather than independent automation; pace is middling compared to information-sector adoption.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric/nursing practice, adopts AI diagnostic tools cautiously due to regulatory hurdles and liability concerns, with pilots more common than widespread production use.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially augments clinician productivity by pre-screening tests, flagging abnormalities, and generating preliminary interpretations that the nurse reviews and refines, materially reducing time spent on routine interpretation while keeping the human in the loop for judgment and accountability.
Augmentation potential	claude-sonnet-5	4/5	AI-based EKG and lab analysis tools meaningfully speed up test review and flag anomalies, helping practitioners work faster while retaining final interpretive responsibility.
Task automatability	claude-haiku-4-5-20251001	4/5	AI systems can interpret EKGs and basic laboratory tests with high accuracy comparable to or exceeding human performance on many routine cases, achieving 50%+ time savings through automated flagging and preliminary reads; however, complex cases requiring clinical context and integration with patient history still need human review.
Task automatability	claude-sonnet-5	2/5	AI can flag abnormal values or provide pattern recognition on EKGs/labs, but integrating results into psychiatric clinical judgment and patient-specific context requires human interpretation and accountability, limiting full end-to-end automation.time savings are partial.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Regulatory and liability frameworks (FDA cleared algorithms, clinical validation requirements) create some friction; interpretation often must be reviewed and co-signed by a licensed provider, introducing organizational and legal oversight requirements that slow full automation.
Adoption barriers	claude-sonnet-5	4/5	Clinical interpretation of diagnostic tests for treatment decisions typically requires a licensed practitioner's sign-off, with significant liability exposure for misdiagnosis, creating strong regulatory and professional barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI inference cost for test interpretation is a small fraction of the clinician's loaded wage, especially when amortized across high-volume screening and routine cases; integration and oversight costs are modest relative to the labor replacement value.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted interpretation tools are relatively cheap per test, but required clinician oversight and liability review keep total cost closer to comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products for EKG interpretation (e.g., FDA-cleared AI algorithms) and lab result analysis exist and perform reliably in production settings; they are used clinically today though typically as decision-support rather than fully autonomous interpretation.
Technical feasibility today	claude-sonnet-5	2/5	Deployed clinical decision-support tools exist for EKG interpretation and lab flagging, but these are adjunct aids with notable error rates, not autonomous interpretive systems used standalone in psychiatric practice.

Monitor patients' medication usage and results.

CI 20–70 · exposure 50 · augmentation 88 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Large hospital systems and integrated health networks are rapidly adopting automated medication monitoring and EHR-based clinical alerts; adoption is most advanced in well-resourced sectors (health IT, large providers) though smaller practices lag.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, has been slow to adopt AI for clinical monitoring due to regulatory, liability, and workflow integration challenges.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI monitoring systems substantially augment nurse productivity by automating data aggregation, flagging priority cases, and highlighting drug interactions in real time; nurses stay in the loop for decision-making while spending far less time on manual record review and calculation.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by tracking medication logs, flagging anomalies, summarizing patient-reported outcomes, and alerting clinicians to potential issues, improving efficiency while the nurse retains decision authority.
Task automatability	claude-haiku-4-5-20251001	4/5	AI systems can reliably process medication records, dosages, lab results, and standard clinical outcomes through EHR integration and alert systems, flagging drug interactions and adherence issues automatically. Some clinical judgment about medication efficacy and side-effect interpretation remains necessary, but 50%+ time savings on routine monitoring is already demonstrable.
Task automatability	claude-sonnet-5	2/5	AI can flag adherence issues or drug interactions from structured data, but interpreting patient-reported symptoms, side effects, and adjusting judgment requires clinical assessment beyond current automation.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Nurses remain legally required to interpret results and make clinical decisions, and malpractice liability falls on the healthcare organization; regulations (FDA oversight of clinical decision support, state nursing scope laws) require human sign-off on medication changes, creating moderate friction.
Adoption barriers	claude-sonnet-5	5/5	Monitoring psychiatric medication effects and adjusting care legally requires a licensed prescriber, making this a hard regulatory and liability barrier.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-powered monitoring systems (inference + integration with EHR + alert management) cost a fraction of a nurse's labor per patient reviewed; once deployed at scale across patient populations, the cost per monitoring cycle is orders of magnitude cheaper than manual review.
Cost vs. human wage	claude-sonnet-5	2/5	AI monitoring tools add value but still require a licensed clinician to review and act, so overall cost savings versus the human labor loop are modest.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature healthcare monitoring systems with AI-assisted medication review and alert logic are deployed in production at major health systems; examples include integrated EHR clinical decision support and pharmacy monitoring tools. Error rates are material but acceptable in practice, as human oversight remains standard.
Technical feasibility today	claude-sonnet-5	2/5	Some clinical decision support tools flag interactions and monitor lab values, but no deployed product independently monitors psychiatric medication response and outcomes reliably at scale.

Document patients' medical and psychological histories, physical assessment results, diagnoses, treatment plans, prescriptions, or outcomes.

CI 25–62 · exposure 50 · augmentation 88 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Psychiatric practices and mental health clinics lag in AI adoption compared to primary care or radiology. Many rely on manual or semi-manual documentation workflows, and organizational resistance to purely AI-drafted psychiatric histories is high due to complexity, liability concerns, and specialty-specific workflows that resist standardization.
Sector adoption velocity	claude-sonnet-5	3/5	Ambient AI documentation tools are being piloted and adopted increasingly in outpatient and psychiatric settings, but adoption is uneven across health systems and slower than in pure information-sector industries.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists APNs by auto-populating templates, summarizing prior records, suggesting diagnostic codes, and organizing vital sign and lab data, substantially reducing documentation burden. The clinician remains the decision-maker and validator, but AI transforms the speed and completeness of note assembly, raising overall documentation productivity.
Augmentation potential	claude-sonnet-5	5/5	AI documentation assistants substantially reduce charting burden and administrative burnout while the clinician retains full responsibility for reviewing, editing, and finalizing records.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can draft documentation from structured inputs (vital signs, symptom checklists), psychiatric history and assessment require nuanced human judgment, synthesis of subjective patient reports, and clinical interpretation that current systems struggle to capture reliably. Partial automation of note templating is feasible, but end-to-end 50% time-saving at equal clinical quality is not demonstrable today.
Task automatability	claude-sonnet-5	4/5	AI scribe and documentation tools can transcribe encounters and draft structured notes (histories, assessments, treatment plans) with significant time savings, though clinician review and correction of clinical content is still needed.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Legal and regulatory barriers are substantial: most jurisdictions require a licensed clinician (APRN, MD, etc.) to attest to the accuracy and completeness of psychiatric documentation, and liability for missed diagnoses or treatment decisions falls on the clinician who must review and sign the record. Medical records are discoverable in malpractice and regulatory actions, creating high error-cost asymmetry.
Adoption barriers	claude-sonnet-5	4/5	Medical records, diagnoses, and prescriptions legally require sign-off by a licensed provider, and documentation errors carry clinical and legal liability, so a human must remain accountable even if AI drafts content.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI-assisted documentation still requires significant clinician oversight, review, and correction time. The cost of the AI system (licensing, integration, training, liability insurance) plus the clinician's residual labor often approaches or exceeds the cost of direct clinician documentation, especially in low-volume or specialized psychiatric settings.
Cost vs. human wage	claude-sonnet-5	4/5	AI scribe subscriptions cost a small fraction of a nurse practitioner's hourly documentation time, offering substantial though not order-of-magnitude-plus savings once oversight and EHR integration costs are included.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Clinical documentation AI products exist (EHR-integrated drafting tools, speech-to-text), but they produce incomplete or inaccurate psychiatric narratives without substantial clinician review and correction. Liability, regulatory review requirements, and the need for clinician sign-off mean no mature product reliably performs the full psychiatric assessment documentation independently.
Technical feasibility today	claude-sonnet-5	4/5	Ambient clinical documentation products (e.g., AI scribes integrated with EHRs) are deployed at scale in healthcare systems today and reliably generate draft notes, though psychiatric nuance and prescription accuracy require clinician verification.

Teach classes in mental health topics, such as stress reduction.

CI 25–30 · exposure 25 · augmentation 75 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare and nursing education remain slow to adopt AI-led instruction due to regulatory conservatism, accreditation requirements, and institutional preference for human clinical expertise. Pilots exist but production displacement of nurse instructors is minimal.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare and patient education sectors are slower adopters of AI-led instruction compared to fields like tech or finance, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can substantially assist a psychiatric nurse instructor by generating lesson outlines, providing evidence-based references, creating supplementary videos, and offering real-time suggestions during class delivery. This augmentation significantly raises the instructor's productivity and content quality while maintaining human authority and clinical judgment.
Augmentation potential	claude-sonnet-5	4/5	AI is highly useful for generating class curricula, slides, handouts, and answering prep questions, substantially speeding up a nurse's preparation while they remain the live instructor.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can generate educational content and lesson plans on mental health topics, but cannot replicate the live interpersonal dynamics, real-time responsiveness to student questions, and therapeutic modeling essential to teaching stress reduction classes. A human instructor remains necessary for genuine pedagogical and emotional engagement.
Task automatability	claude-sonnet-5	2/5	AI can generate educational content and even deliver static presentations, but live, adaptive teaching involving reading a group, answering nuanced questions, and adjusting to participants' emotional states remains largely human-driven.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Healthcare education and credentialing typically require a licensed clinician to deliver instruction, particularly for stress reduction and mental health topics where therapeutic framing and professional accountability matter. Organizational and regulatory norms strongly favor human-led instruction in clinical mental health education.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing mandate requires a nurse to teach general wellness classes, but clinical credibility, liability for health advice, and audience trust in a human presenter create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Generating lecture notes and slides has become cost-efficient, but integrating AI oversight with human instruction, platform hosting, and quality assurance remains comparable to or more expensive than a single nurse instructor teaching classes, especially for specialized mental health content.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply draft materials, but delivering the actual class still requires a paid clinician's time for facilitation, credibility, and liability coverage, keeping overall costs comparable to human-led delivery.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can draft educational materials and some educational platforms exist, no deployed system reliably delivers end-to-end classroom teaching with the clinical expertise, adaptive instruction, and interpersonal credibility required of a psychiatric nurse instructor. Current products are limited to content generation or asynchronous modules.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI-generated slide decks/scripts exist and are used for content creation, but no deployed product independently teaches mental health classes to patient/public audiences at scale in clinical settings.

Educate patients and family members about mental health and medical conditions, preventive health measures, medications, or treatment plans.

CI 25–29 · exposure 25 · augmentation 75 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare organizations, especially in psychiatry, move slowly toward AI automation for patient-facing education due to liability concerns, regulatory oversight (state nursing boards), and the clinical premium placed on human clinician relationships. Adoption remains in the pilot phase with minimal production displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially mental health, has been slower to adopt AI-driven patient communication tools due to regulatory, liability, and trust concerns compared to other professional service sectors.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist Advanced Practice Nurses by drafting personalized educational materials, summarizing medication information, generating family-friendly explanations, and preparing talking points—all while the nurse delivers, contextualizes, and ensures understanding. This assistive use raises nurse productivity and frees time for deeper engagement without removing the nurse from the interaction.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by drafting educational materials, summarizing conditions in plain language, and providing quick reference information, which the clinician can then customize and deliver to patients.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate educational content about medications and treatment plans, delivering this task end-to-end requires genuine relationship-building, assessing individual comprehension levels, addressing emotional concerns, and adapting explanations in real-time to each patient's needs and literacy—capabilities that remain beyond current AI systems. Partial automation (drafting materials, initial information delivery) is possible but cannot substitute for the interactive, empathetic human judgment that defines effective patient education.
Task automatability	claude-sonnet-5	2/5	AI can generate patient-education content and answer general questions, but tailoring education to a specific patient's clinical situation, family dynamics, and emotional state requires human judgment and rapport that current systems cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong barriers protect this task: psychiatric care requires trust, accountability, and often licensure; patients in crisis or with complex mental health histories need a qualified clinician; malpractice and error liability fall on the provider, not an AI tool; and family education often involves difficult conversations that demand professional judgment and legal responsibility. Regulations and the therapeutic relationship itself create hard constraints on substitution.
Adoption barriers	claude-sonnet-5	4/5	Psychiatric nurse practitioners are licensed clinicians and this education task is often legally and clinically tied to their scope of practice, especially regarding medication and treatment plan discussions requiring professional accountability.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools (chatbots, content generation) cost far less than a psychiatric nurse's labor, but the all-in cost of integration, clinical oversight, liability management, and the need for human nurses to still conduct meaningful education makes the comparison closer than the cost of infrastructure alone suggests. The human role cannot be fully displaced without clinical and safety costs.
Cost vs. human wage	claude-sonnet-5	3/5	AI-generated educational content is cheap to produce, but the need for clinician review and personalization for safety keeps effective cost closer to comparable rather than dramatically cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed products exist for generating patient-facing educational materials and chatbots that answer health questions, but they have significant limitations in handling complex psychiatric contexts, managing emotional dynamics, ensuring clinical accuracy, and maintaining provider accountability. No production system reliably performs the full task of educating patients and families in a psychiatric care setting at the standard required of an Advanced Practice Nurse.
Technical feasibility today	claude-sonnet-5	2/5	Patient-facing chatbots and educational materials exist, but no deployed product reliably conducts personalized clinical psychoeducation at scale without clinician oversight due to safety and accuracy concerns in mental health contexts.

Develop practice protocols for mental health problems, based on review and evaluation of published research.

CI 25–29 · exposure 25 · augmentation 75 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare remains among the more conservative sectors for high-stakes clinical decision-making. While some organizations pilot AI-assisted literature review, end-to-end protocol development via AI remains rare in production; most settings still rely on traditional clinician-led, committee-based processes.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially clinical protocol development, adopts AI more slowly than other professional services due to regulatory caution and safety concerns, though literature-review tools are gaining pilot use.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI excels at augmenting this task: it can rapidly search, filter, and summarize large research corpora, flag key evidence, and generate draft synthesis. Advanced practice nurses using current AI tools for literature review and preliminary drafting can work substantially faster while maintaining full clinical control over protocol validation and finalization.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially speed up literature review, summarization, and drafting of protocol language, meaningfully augmenting the clinician's productivity while they retain final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can summarize and extract information from published research at scale, developing practice protocols requires synthesizing evidence into clinical guidance with accountability for patient safety. Current AI systems cannot independently create validated protocols that meet professional and institutional standards; they lack the clinical judgment and legal responsibility that clinicians bear.
Task automatability	claude-sonnet-5	2/5	AI can synthesize literature and draft protocol outlines, but final protocol development requires clinical judgment, integration of local context, and accountability that current systems cannot fully replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Practice protocols require sign-off by licensed clinicians and often institutional review boards, regulatory compliance (Joint Commission, state nursing boards), and legal liability for outcomes. Healthcare organizations cannot substitute AI output for clinician-authored protocols without human professional accountability, creating a strong licensing and legal barrier.
Adoption barriers	claude-sonnet-5	4/5	Protocols affecting patient care typically require sign-off by licensed advanced practice nurses/physicians and compliance with clinical governance standards, creating strong professional and liability barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI can reduce the labor cost of literature review and initial synthesis substantially, but the advanced practice nurse must still perform critical appraisal, contextualization, and validation. The savings are meaningful but offset by necessary human oversight, making overall cost roughly comparable to traditional protocol development.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cut literature review time significantly, but the human oversight, clinical validation, and liability review needed keep overall costs closer to comparable rather than order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools can assist with literature review and summarization, but no deployed product reliably develops complete, clinically validated practice protocols end-to-end. Existing systems perform narrow subtasks (literature retrieval, synthesis) but cannot produce the integrated, professionally vetted clinical documents that organizations require for implementation.
Technical feasibility today	claude-sonnet-5	2/5	Products like literature-summarization and clinical decision-support tools exist, but no deployed system reliably produces validated, clinically sound practice protocols without extensive expert review.

Develop and implement treatment plans.

CI 20–25 · exposure 25 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare and psychiatry are laggard sectors in AI adoption due to regulatory, liability, and human-contact requirements. While documentation and administrative tools are deployed, clinical decision-making tools see slow, cautious adoption with frequent pilot status rather than deep production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, has been slower and more cautious in adopting AI for core clinical decision-making compared to information/finance sectors.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist APRNs by rapidly synthesizing evidence for specific diagnoses, suggesting evidence-based interventions, flagging drug interactions, and organizing clinical data—substantially raising productivity while the nurse retains clinical judgment and patient responsibility. This augmentation is already starting to appear in EHR-integrated tools and clinical decision support.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by summarizing patient records, suggesting evidence-based interventions, flagging drug interactions, and drafting documentation, improving clinician efficiency while the nurse remains the decision-maker.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in drafting treatment components and suggesting evidence-based interventions, developing and implementing psychiatric treatment plans requires nuanced clinical judgment about complex patient presentations, ongoing adjustment based on patient response, and integration of therapeutic relationship. Current AI cannot reliably perform the full task end-to-end with 50% time savings at equal quality.
Task automatability	claude-sonnet-5	2/5	Treatment plan development requires clinical judgment, patient history synthesis, risk assessment, and legal accountability that current AI cannot fully replicate end-to-end at equal quality.n.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Psychiatric treatment planning is a regulated scope of practice for licensed APRNs; autonomous AI delegation faces legal and liability barriers. Patients expect and often require human clinical judgment; insurers and regulatory bodies mandate licensed provider sign-off on treatment plans, creating hard barriers to substitution.
Adoption barriers	claude-sonnet-5	5/5	Only a licensed advanced practice nurse (or physician) can legally develop, prescribe, and implement psychiatric treatment plans, making this a hard regulatory and licensure barrier.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI inference for treatment planning support is cheap, but the overhead of clinical validation, oversight, and revision by the APRN nurse remains substantial. The cost of errors in psychiatric treatment planning is high, so oversight costs offset any raw inference savings, keeping total cost comparable to or higher than the human-alone baseline.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools may cheaply generate draft suggestions, but the clinician's oversight, liability, and implementation work still dominate cost, keeping overall savings modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed clinical product reliably performs autonomous psychiatric treatment planning at production scale. AI systems exist for treatment suggestions and documentation support, but they operate as narrow assistants with material limitations in handling comorbidity, individualization, and dynamic adjustment required in real psychiatric practice.
Technical feasibility today	claude-sonnet-5	2/5	Some clinical decision-support tools can suggest treatment options or draft plan templates, but no deployed product autonomously develops and implements psychiatric treatment plans in production.

Collaborate with interdisciplinary team members, including psychiatrists, psychologists, or nursing staff, to develop, implement, or evaluate treatment plans.

CI 20–25 · exposure 25 · augmentation 50 · importance 4.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare adoption of AI in clinical decision-making remains cautious and slow; psychiatric treatment planning is particularly risk-averse due to liability and the complexity of individualizing care. Most healthcare organizations are still in pilot phases for clinical AI, not production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, adopts AI slowly due to regulatory, liability, and interoperability constraints, with pilots more common than production use for this specific collaborative task.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist nurses by summarizing patient histories, flagging evidence-based guidelines, or drafting documentation templates, moderately raising efficiency in plan formulation. However, the collaborative judgment and accountability required limit how transformative the assistance can be without the nurse directing every output.
Augmentation potential	claude-sonnet-5	3/5	AI tools can aid by synthesizing patient records, flagging risks, or drafting treatment plan summaries, usefully augmenting the team's efficiency without replacing the human collaboration itself.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help draft treatment plan components or summarize clinical information, the core task requires real-time judgment about complex psychiatric cases, negotiation among professionals with competing perspectives, and accountability for treatment decisions. End-to-end automation of plan development/implementation/evaluation with ≥50% time saving at equal quality is not demonstrated.
Task automatability	claude-sonnet-5	2/5	Interdisciplinary collaboration requires real-time relational communication, clinical judgment, and shared accountability among licensed professionals that current AI cannot replace end-to-end, though it can support documentation and information sharing.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Advanced practice nurses typically operate under state licensure and scope-of-practice regulations; treatment planning decisions carry legal liability and require a licensed clinician's professional judgment and signature. Psychiatric care also maintains strong institutional expectations that humans remain accountable for plan safety and appropriateness.
Adoption barriers	claude-sonnet-5	5/5	Treatment planning for psychiatric care legally requires licensed clinicians to collaborate and sign off, making this a hard regulatory and liability barrier to substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI inference costs for clinical documentation assist are modest, but the task requires extensive human oversight, liability review, and potential re-work when AI outputs miss nuances. All-in costs remain comparable to or exceed direct clinician time because the nurse cannot be removed from the loop.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply generate summaries or notes, but the core task still requires paid clinical staff time for meetings and judgment, so overall cost savings are modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs interdisciplinary psychiatric treatment planning at scale. AI can assist with documentation and evidence retrieval, but clinical collaboration systems remain research-stage; existing products lack the contextual judgment and professional accountability this task demands.
Technical feasibility today	claude-sonnet-5	2/5	Some clinical decision-support and documentation tools exist that summarize patient data for teams, but no deployed product actually participates in or replaces the collaborative treatment-planning process itself.

Evaluate patients' behavior to formulate diagnoses or assess treatments.

CI 20–20 · exposure 25 · augmentation 63 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Mental health organizations are cautiously piloting AI screening tools but adoption remains slow due to regulatory constraints, liability concerns, and clinician resistance; production deployment of AI-driven diagnostic assessment is rare compared to information or financial sectors.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, has been slow to adopt AI for diagnostic decision-making due to regulatory, ethical, and trust barriers, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist APNs by flagging behavioral patterns, organizing clinical data, suggesting differential diagnoses, and streamlining documentation, substantially raising clinician productivity while the nurse retains core diagnostic authority and clinical judgment.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with documentation, symptom tracking, screening questionnaires, and literature-based treatment suggestions, meaningfully aiding but not replacing the clinician's evaluative judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can assist with structured behavioral screening and pattern recognition from clinical notes, but psychiatric diagnosis requires nuanced interpretation of complex, context-dependent behavioral cues, patient rapport, and real-time clinical judgment that current systems cannot reliably replicate end-to-end at 50% time savings with equal quality.
Task automatability	claude-sonnet-5	2/5	Diagnostic formulation requires synthesizing observed behavior, patient history, rapport, and clinical judgment in real-time interaction, which current AI cannot reliably do end-to-end despite some decision-support capability.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Psychiatric diagnosis and treatment assessment are legally and ethically restricted to licensed mental health professionals; jurisdictional regulations require a credentialed human clinician to conduct, validate, and sign off on psychiatric evaluations, creating hard barriers to full automation.
Adoption barriers	claude-sonnet-5	5/5	Psychiatric diagnosis and treatment assessment legally require a licensed advanced practice provider; this is a regulated clinical judgment task with high liability exposure.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-powered behavioral assessment tools require significant clinical oversight, validation, and integration costs; the total cost per evaluation approximates or exceeds the cost of clinician time, particularly when accounting for liability and supervision.
Cost vs. human wage	claude-sonnet-5	2/5	AI could cheaply assist with note-taking or screening, but full diagnostic evaluation still requires costly human oversight, licensure, and liability coverage, keeping overall cost comparable to human-only workflows.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI products exist for mental health screening and decision support, no deployed system reliably performs independent psychiatric behavioral evaluation and diagnosis at clinical-grade reliability; products show material error rates and cannot yet replace clinician judgment in real psychiatric practice.
Technical feasibility today	claude-sonnet-5	2/5	AI tools exist for clinical documentation and screening flagging, but no deployed product independently performs behavioral evaluation and diagnostic formulation in psychiatric practice.

Distinguish between physiologically- and psychologically-based disorders, and diagnose appropriately.

CI 20–20 · exposure 25 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare adoption of AI for independent clinical diagnosis remains slow due to liability concerns, regulatory caution, and organizational resistance; most psychiatric settings treat AI as a decision-support tool only, not a replacement.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric diagnosis, remains a slow-adopting sector for autonomous AI decision-making due to regulatory caution, though AI-assisted documentation and triage tools are spreading.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI-powered diagnostic suggestion systems and symptom checkers can substantially augment a psychiatric nurse's efficiency by organizing differential diagnoses and highlighting physiological red flags, reducing time spent on literature review and initial case structuring while keeping the clinician in the loop.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by surfacing differential diagnoses, flagging symptom overlaps, and summarizing patient history, improving clinician efficiency and diagnostic breadth while the clinician retains final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in differential diagnosis by processing clinical data and suggesting patterns, end-to-end distinction between physiological and psychological disorders requires nuanced clinical judgment, integration of patient history, physical exam findings, and real-time adaptive questioning that current systems cannot reliably perform autonomously at equal quality.
Task automatability	claude-sonnet-5	2/5	Differential diagnosis requires physical exam integration, longitudinal patient history, and clinical judgment under uncertainty that current AI cannot reliably perform end-to-end without a licensed clinician driving the process., only partial support is possible today.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Legal and regulatory frameworks require a licensed independent practitioner (psychiatrist or advanced practice nurse with specific credentials) to diagnose and take legal responsibility; diagnosis cannot be delegated to automated systems without human sign-off.
Adoption barriers	claude-sonnet-5	5/5	Diagnosis is a legally restricted act requiring a licensed APRN/physician, with strong liability exposure for misdiagnosing physiological vs. psychiatric conditions, making this a hard-barrier task.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Psychiatric nurse diagnostic time is relatively low-cost compared to physician consultation, and the cost of AI infrastructure plus required oversight by a licensed clinician approaches or exceeds the direct labor savings in most settings.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools can cheaply generate differential lists, but the required clinician verification, liability exposure, and integration with physical exam data keep effective all-in cost comparable to or only modestly below clinician-driven diagnosis.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Clinical decision support tools exist to flag differential diagnoses, but no deployed AI system reliably performs independent diagnostic distinction in production psychiatric settings; systems remain narrow in scope and require substantial human validation and override.
Technical feasibility today	claude-sonnet-5	2/5	Clinical decision-support tools and LLM-based diagnostic aids exist but are used as adjuncts, not autonomous diagnosers; no deployed product independently distinguishes physiological from psychological etiology in production psychiatric care.

Refer patients requiring more specialized or complex treatment to psychiatrists, primary care physicians, or other medical specialists.

CI 11–25 · exposure 17 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare adoption of AI decision support is growing but remains cautious and incremental; most organizations use AI as an assistive tool within clinician workflows rather than autonomous referral systems, reflecting both regulatory caution and clinical risk aversion.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, is a historically slow adopter of autonomous AI decision-making due to regulatory, ethical, and liability constraints, though documentation tools are spreading.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by flagging red flags, suggesting relevant specialists based on comorbidities, or integrating clinical guidelines; however, the nurse retains decision authority and the gain is moderate support rather than transformative productivity change.
Augmentation potential	claude-sonnet-5	3/5	AI can assist by summarizing patient history, flagging risk indicators, or suggesting appropriate specialists, helping the clinician make faster, better-informed referral decisions while retaining final judgment.
Task automatability	claude-haiku-4-5-20251001	1/5	Referral decisions require nuanced clinical judgment, understanding of patient history, system knowledge, and professional accountability that current AI cannot reliably perform end-to-end. AI cannot independently assess complex psychiatric presentations or legally authorize specialist referrals.
Task automatability	claude-sonnet-5	2/5	Referral decisions require clinical judgment about symptom severity, risk, and specialist fit that current AI cannot reliably perform end-to-end, though AI can help draft referral letters or flag cases needing escalation.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Referral authority is legally and professionally vested in licensed clinicians; liability for incorrect referrals rests with the nurse, and regulatory frameworks (state nursing boards, Medicare/Medicaid oversight) require human accountability for specialist routing decisions.
Adoption barriers	claude-sonnet-5	5/5	Referral is a licensed clinical act tied to scope-of-practice regulations and malpractice liability, requiring an APRN's independent judgment and legal authority to make care decisions.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for decision support are relatively inexpensive, but oversight, integration into EHR workflows, and clinical validation add cost; the human nurse remains essential, making full cost displacement implausible and not advantageous.
Cost vs. human wage	claude-sonnet-5	2/5	The cognitive assessment and liability of making a referral decision requires licensed clinical involvement, so AI cannot substitute the core judgment cheaply; at best it reduces administrative time slightly.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can suggest specialists based on diagnostic codes or flags, no deployed product reliably makes clinically sound referral decisions independently. Systems exist to support triage but require human oversight and clinical discretion; they cannot replace the nurse's judgment.
Technical feasibility today	claude-sonnet-5	2/5	Clinical decision-support tools exist that suggest referrals or triage based on symptom checklists, but no deployed product autonomously makes psychiatric referral decisions in production without clinician oversight.

Develop, implement, or evaluate programs such as outreach activities, community mental health programs, and crisis situation response activities.

CI 7–25 · exposure 13 · augmentation 50 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare and mental health sectors show slower AI adoption for autonomous decision-making; most adoption is in back-office and routine tasks. Program development and crisis response remain heavily dependent on credentialed clinicians, with AI in supporting rather than leadership roles.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare and community mental health services are historically slow AI adopters, with pilots for administrative support but little penetration into program design or crisis response operations.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist with data gathering, literature review, outcome tracking, scheduling, and preliminary analysis to inform program decisions. However, the core work of stakeholder engagement, clinical judgment, and adaptive implementation remains human-centered, limiting the transformative potential of augmentation.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with drafting program plans, analyzing outreach data, summarizing evaluation metrics, and researching best practices, improving efficiency while humans retain core responsibility.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with program design, data analysis, and documentation, developing, implementing, and evaluating community mental health programs requires ongoing judgment, stakeholder engagement, regulatory navigation, and adaptive management. Current AI systems cannot autonomously manage the full lifecycle of such programs with equal quality.
Task automatability	claude-sonnet-5	1/5	Program development, implementation, and evaluation for community mental health and crisis response requires situated judgment, stakeholder coordination, and real-world adaptation that current AI cannot execute end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong barriers exist: mental health program implementation typically requires licensure (RN/APRN credentials), regulatory compliance (mental health licensing boards, accreditation standards), and organizational accountability for program outcomes. Liability and clinical responsibility cannot be fully transferred to AI.
Adoption barriers	claude-sonnet-5	4/5	Clinical licensure, liability for crisis intervention decisions, and regulatory/ethical requirements around mental health programming create strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of AI systems plus required human oversight, validation, and program management labor remains comparable to or higher than hiring a qualified psychiatric nurse for program-level work. AI reduces some administrative burden but does not displace the core expertise.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the human labor and relationship-building required, so no meaningful cost displacement exists; any AI use is supplementary at added cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs end-to-end program development and implementation in mental health settings. AI tools exist for components (scheduling, data tracking, outcome measurement) but not for the integrated program leadership, stakeholder coordination, and crisis response adaptability this task demands.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously designs, runs, and evaluates mental health outreach or crisis-response programs; this remains firmly in the human professional domain.

Diagnose psychiatric disorders and mental health conditions.

CI 11–20 · exposure 17 · augmentation 63 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Mental health organizations adopt decision-support and screening tools at moderate rates, but autonomous or AI-driven diagnosis remains rare in production; cultural and regulatory caution, combined with liability concerns, limit velocity of AI adoption for this core clinical task.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, has been slow to adopt autonomous AI diagnostic tools due to regulatory, ethical, and safety concerns, with pilots more common than production deployment.'
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist psychiatric nurses by highlighting relevant DSM criteria, flagging symptom patterns, or organizing patient history, thereby reducing assessment time and supporting thoroughness; however, the interpretive and relational core of diagnosis remains human-centered.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist with intake summarization, symptom pattern recognition, documentation, and flagging risk indicators, improving efficiency while the clinician retains diagnostic authority.'
Task automatability	claude-haiku-4-5-20251001	1/5	Psychiatric diagnosis requires sustained clinical judgment, differential reasoning across complex presentations, and integration of patient history, behavioral observation, and subjective symptom reports. Current AI systems cannot reliably perform end-to-end diagnostic assessment meeting the >50% time-saving threshold at equal quality.
Task automatability	claude-sonnet-5	2/5	AI can assist with symptom analysis and differential suggestions using structured data, but the actual diagnostic act requires clinical judgment, patient interaction, and integration of nuanced context that current systems cannot fully replicate end-to-end.'
Adoption barriers	claude-haiku-4-5-20251001	5/5	Psychiatric diagnosis by licensed professionals (physicians, psychiatric nurse practitioners, psychologists) is legally mandated in most jurisdictions; malpractice liability, duty of care, and regulatory requirements around diagnostic authority create hard barriers to automation or unsupervised AI substitution.
Adoption barriers	claude-sonnet-5	5/5	Psychiatric diagnosis is a licensed clinical act requiring an advanced practice nurse's legal authority and accountability, making unauthorized AI substitution legally prohibited.'
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI screening and decision-support tools cost far less than a clinical encounter, but they do not replace the diagnostic function itself; they require expert clinician review and integration, making the all-in cost higher than the human performing diagnosis independently.
Cost vs. human wage	claude-sonnet-5	2/5	While AI inference is cheap, the required human oversight, liability review, and clinical validation keep total cost comparable to or only modestly less than a clinician's time for this task.'
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI tools exist for symptom screening and decision support (e.g., symptom checkers, DSM-5 criterion matching), no deployed product reliably performs independent psychiatric diagnosis. Existing systems lack the nuance to handle comorbidity, cultural context, and the subjective elements core to mental health assessment.
Technical feasibility today	claude-sonnet-5	2/5	Some clinical decision-support tools exist for flagging risk or suggesting differentials, but no deployed product independently and reliably diagnoses psychiatric conditions in production without clinician oversight.'

Provide routine physical health screenings to detect or monitor problems such as heart disease and diabetes.

CI 0–25 · exposure 13 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Healthcare remains among the slowest sectors to adopt autonomous AI systems due to regulatory constraints, licensing requirements, and the direct human-contact nature of clinical assessment. Adoption of AI for screening remains largely in pilot phases, not production displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially direct patient physical care, adopts AI more slowly than other professional sectors due to regulatory, safety, and workflow integration constraints.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist nurses by flagging abnormal vital signs, suggesting differential diagnoses based on screening results, or prompting additional tests—helping productivity within the screening workflow while the nurse retains clinical decision-making authority.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by flagging risk factors, analyzing wearable/vital sign data, and supporting documentation and decision-making during screenings, enhancing clinician efficiency.
Task automatability	claude-haiku-4-5-20251001	1/5	Physical health screenings require hands-on clinical assessment (vital signs, physical examination, patient interaction) that current AI cannot perform. While AI can assist with interpretation of lab results or health history analysis, the core task of direct patient screening cannot be automated end-to-end today.
Task automatability	claude-sonnet-5	2/5	Physical screenings require hands-on examination, vital sign measurement, and physical contact (e.g., palpation, auscultation) that current AI cannot perform; AI can assist with data interpretation but not the physical act itself.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Advanced practice nurses (or equivalent licensed clinicians) are legally required to perform physical health screenings in most jurisdictions. Regulatory requirements, patient contact necessity, and liability for missed diagnoses create hard barriers to substitution.
Adoption barriers	claude-sonnet-5	4/5	Physical health screenings typically require licensed clinical judgment and hands-on assessment, with liability and regulatory expectations that a qualified practitioner perform or supervise the exam.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of implementing AI infrastructure, oversight systems, and liability safeguards for autonomous screening would exceed the loaded wage of an advanced practice nurse performing routine screenings, particularly given the low-automation feasibility.
Cost vs. human wage	claude-sonnet-5	2/5	AI could cheaply analyze data inputs, but the physical examination component still requires a human clinician, keeping overall costs comparable to or only marginally below human-only delivery.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product performs autonomous physical health screenings. This task requires licensed clinical judgment and direct patient contact, which remains firmly within human clinical practice without demonstrated AI replacement systems in production.
Technical feasibility today	claude-sonnet-5	2/5	Deployed AI tools can analyze vitals, ECGs, or lab results, but no product performs the full hands-on physical screening process autonomously in clinical practice today.

Assess patients' mental and physical status, based on the presenting symptoms and complaints.

CI 3–20 · exposure 13 · augmentation 75 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Healthcare adoption of AI remains cautious and heavily regulated; clinical assessment tools are mostly in pilot or experimental phases within institutional settings rather than mainstream production use, reflecting slow velocity in high-liability domains.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially psychiatric care, remains a slower-adopting sector for full clinical task automation despite growing use of AI for administrative support.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist by auto-populating symptom checklists, highlighting red flags, suggesting differential diagnoses, and organizing patient history, thereby raising nurse efficiency in data gathering and documentation while the nurse retains clinical decision-making authority.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by suggesting differential diagnoses, summarizing patient history, flagging risk indicators, and drafting documentation, enhancing clinician efficiency and thoroughness.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can support symptom documentation and flag potential conditions, but cannot reliably perform comprehensive mental and physical status assessment end-to-end. The task requires nuanced clinical judgment, observation of nonverbal cues, and real-time response to patient presentation that current systems cannot consistently replicate at parity with human clinicians.
Task automatability	claude-sonnet-5	1/5	Comprehensive mental and physical status assessment requires hands-on physical exam, direct patient interaction, and clinical judgment integrating nonverbal cues, history, and risk factors that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Legal and regulatory frameworks require licensed nurses or physicians to perform clinical assessment; liability for missed diagnoses creates strong gatekeeping, and professional standards mandate human accountability for patient safety decisions.
Adoption barriers	claude-sonnet-5	5/5	This is a licensed clinical function requiring an advanced practice nurse's legal scope of practice, with direct liability for diagnosis and patient safety, making substitution legally prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The integration costs, model maintenance, and required human oversight for clinical-grade assessment make AI economically comparable to or more expensive than a nurse's time, especially when liability and quality assurance are factored in.
Cost vs. human wage	claude-sonnet-5	1/5	Since AI cannot substitute for the licensed clinician performing this task, there is no viable cost comparison—human labor remains mandatory.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI tools exist for symptom screening and documentation support, no deployed product reliably performs full clinical assessment independently. Systems show material error rates in complex cases and typically require human clinician validation; they function as assistants rather than autonomous performers.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product independently conducts psychiatric mental status and physical exams in production; AI tools at best support documentation or symptom-checking, not the actual clinical assessment.

Conduct individual, group, or family psychotherapy for those with chronic or acute mental disorders.

CI 9–11 · exposure 9 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Psychiatric and mental health sectors show minimal production adoption of AI for independent therapy delivery; most healthcare adoption remains in triage and administrative tasks. Clinical conservatism and regulatory caution in mental health slow any shift toward automation of direct therapeutic contact.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially behavioral health, remains a slower-adopting sector for autonomous AI due to regulatory, ethical, and safety concerns, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist nurses in drafting session notes, suggesting evidence-based interventions, or tracking symptom patterns, moderately raising documentation and research efficiency. However, augmentation remains limited because the core therapeutic work—rapport, real-time clinical judgment, crisis response—remains wholly human-dependent.
Augmentation potential	claude-sonnet-5	3/5	AI can help with documentation, session note summarization, treatment planning suggestions, and psychoeducational content, moderately boosting efficiency while the clinician remains the primary therapeutic agent.
Task automatability	claude-haiku-4-5-20251001	1/5	Psychotherapy fundamentally requires sustained emotional engagement, empathic attunement, and dynamic responsiveness to patient cues that current AI cannot reliably replicate. While AI can generate supportive text, it cannot establish the therapeutic alliance or make real-time clinical judgments needed for safety in acute psychiatric contexts.
Task automatability	claude-sonnet-5	1/5	Delivering psychotherapy requires nuanced clinical judgment, real-time relational attunement, and legal accountability for patient safety that current AI cannot replicate end-to-end for chronic or acute mental disorders.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Psychotherapy is a licensed, regulated practice requiring credentialed nurses or therapists; law and professional standards mandate human accountability for psychiatric care decisions. Liability asymmetry is extreme—algorithmic harm in psychiatric contexts carries high medical and legal consequences.
Adoption barriers	claude-sonnet-5	5/5	Psychotherapy for psychiatric patients requires licensure, clinical supervision, and legal accountability; APRNs face strict scope-of-practice, malpractice, and patient-safety regulations that block non-human delivery of core therapy.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	While AI inference costs are low, integrating safe, regulated, clinically supervised AI psychotherapy would require extensive validation, liability coverage, and human oversight that narrows the cost advantage. Human therapists remain substantially cheaper when accounting for regulatory and malpractice frameworks.
Cost vs. human wage	claude-sonnet-5	2/5	While AI chat tools are cheap per interaction, they cannot substitute for the full therapeutic task, so any comparison undercounts the necessary human oversight and liability costs required to make AI safe for this use.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed AI system reliably conducts independent psychotherapy in production settings; chatbots may provide some psychoeducational content but lack the clinical judgment, liability insurance, and regulatory approval to practice therapy. Research demonstrates AI conversational agents underperform human therapists on core therapeutic outcomes.
Technical feasibility today	claude-sonnet-5	2/5	Some chatbot-based mental health apps exist (e.g., Woebot, Wysa) and show narrow utility for mild symptoms, but no deployed product reliably conducts full psychotherapy sessions for acute or chronic psychiatric conditions.

Write prescriptions for psychotropic medications as allowed by state regulations and collaborative practice agreements.

CI 0–7 · exposure 5 · augmentation 63 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Autonomous or semi-autonomous prescription writing for psychotropic drugs is not being adopted in production healthcare systems because regulatory and liability barriers prevent it. Adoption remains restricted to human-supervised clinical decision support, not displacement of the prescriber's role.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare broadly adopts AI for documentation and decision support, but actual prescribing authority automation is essentially absent and highly regulated, keeping adoption in this specific task very slow.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by summarizing patient histories, flagging drug interactions, suggesting evidence-based medication options, or automating documentation—genuinely useful productivity gains for a human prescriber. However, the final clinical judgment and legal authority remain with the nurse, limiting augmentation to moderate productivity improvement.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by suggesting dosing options, flagging drug interactions, and drafting prescription documentation, while the licensed nurse retains final decision-making and legal responsibility.
Task automatability	claude-haiku-4-5-20251001	1/5	Writing psychotropic medication prescriptions requires real-time clinical judgment about patient history, contraindications, drug interactions, and individualized treatment plans. Current AI systems lack the authority, accountability, and contextual clinical reasoning to perform this end-to-end, and no AI today can legally or safely substitute for a licensed practitioner's prescriptive decision-making.
Task automatability	claude-sonnet-5	1/5	Prescribing requires licensed clinical judgment, patient assessment, and legal authority that current AI cannot exercise independently; no time-saving substitution meets the equal-quality bar end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Prescribing psychotropic medications is heavily regulated; state nursing boards, DEA regulations, and collaborative practice agreements explicitly require a licensed advanced practice nurse to authorize and sign prescriptions. Legal liability and malpractice exposure are asymmetric—errors in psychiatric medication can cause serious harm—making human oversight non-negotiable.
Adoption barriers	claude-sonnet-5	5/5	Prescribing psychotropic medications is tightly regulated, requiring a licensed advanced practice nurse with prescriptive authority under state law and collaborative practice agreements—a hard legal barrier to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI-assisted prescription drafting or decision support remains expensive relative to the nurse's hourly cost, especially when accounting for required human oversight, validation, and the liability costs of errors in psychopharmacology. Full autonomy is legally impossible, so cost comparison favors the human clinician.
Cost vs. human wage	claude-sonnet-5	1/5	Since AI cannot legally or safely perform this task independently, there is no valid AI cost comparison—the human must remain the prescriber, making AI substitution cost-inapplicable/more expensive when accounting for liability.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can draft clinical notes or suggest drug classes based on symptom descriptions, no deployed product reliably performs independent prescription writing for psychotropic medications. Any such system would face severe regulatory and liability barriers; what exists in production are decision-support tools that assist human prescribers, not autonomous prescription generators.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously writes and issues psychotropic prescriptions in production; this remains firmly a research/no-go area due to legal and clinical constraints.

Treat patients for routine physical health problems.

CI 0–7 · exposure 5 · augmentation 63 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Healthcare remains a laggard sector for task automation due to regulatory gatekeeping, liability concerns, and the requirement for human accountability; adoption of AI for autonomous patient treatment is negligible in practice.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare delivery of hands-on clinical care remains a laggard sector for AI substitution despite adoption of AI in documentation and decision support tools.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI meaningfully assists APRNs in this task through symptom checkers, drug interaction databases, clinical decision support, and evidence summaries that raise diagnostic accuracy and efficiency while the nurse retains responsibility for examination and clinical judgment.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with differential diagnosis suggestions, documentation, and evidence lookup during the visit, but the core physical treatment task itself sees limited augmentation.
Task automatability	claude-haiku-4-5-20251001	1/5	Treating patients for routine physical health problems requires clinical assessment, physical examination, patient interaction, and nuanced clinical judgment that current AI cannot perform end-to-end. While AI can assist with differential diagnosis or drug information, the core task involves hands-on patient care, physical examination, and therapeutic decision-making that remains beyond full automation.
Task automatability	claude-sonnet-5	1/5	Physical examination, hands-on assessment, and treatment decisions require physical presence, manual dexterity, and licensed clinical judgment that current AI cannot replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	This task is protected by strong regulatory and legal barriers: state licensure laws require an Advanced Practice Registered Nurse (APRN) to perform or directly oversee patient treatment, prescribing authority is legally restricted to licensed practitioners, and malpractice liability attaches to the human clinician, not AI systems.
Adoption barriers	claude-sonnet-5	5/5	This is a licensed clinical act requiring an APRN's legal scope of practice, malpractice liability, and direct patient contact, making it one of the most protected task types.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The all-in cost of AI infrastructure, clinical validation, oversight, and liability coverage for independent treatment of physical health problems would exceed the loaded wage of an advanced practice nurse, especially given current regulatory and safety requirements.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical exam and treatment delivery, so there is no viable cost comparison—the human clinician remains essential and AI adds cost as a supplement rather than a replacement.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed AI product reliably performs independent patient treatment for physical health problems in production. AI diagnostic support tools exist but require human clinicians to examine patients, assess findings, and execute treatment—they do not substitute for the licensed practitioner's role.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product independently treats patients for physical health problems; at best AI assists with documentation or triage, not the actual clinical treatment task.

Participate in treatment team conferences regarding diagnosis or treatment of difficult cases.

CI 0–6 · exposure 0 · augmentation 38 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Healthcare organizations, particularly in mental health, show laggard adoption of AI in clinical decision-making roles. Conference participation requires human licensure and accountability, creating structural barriers to any AI substitution or displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially behavioral health, has been slower and more cautious in adopting AI for clinical decision-making tasks compared to information/finance sectors.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI might assist by pre-populating case summaries or flagging relevant diagnostic criteria before the conference, but the core task—live synthesis, judgment, and team collaboration—offers limited augmentation value while the human remains fully responsible.
Augmentation potential	claude-sonnet-5	3/5	AI can assist by summarizing patient records, flagging relevant literature, or drafting case notes ahead of conferences, improving preparation efficiency even though it cannot participate in the judgment itself.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires real-time collaborative clinical judgment, interpersonal dynamics, and synthesizing nuanced patient information in a group setting. Current AI cannot meaningfully participate in live conferences or contribute novel diagnostic/treatment insights that would meet the 50% time-saving threshold.
Task automatability	claude-sonnet-5	1/5	This requires synthesizing complex clinical judgment, live discussion, and interpersonal negotiation among a care team, which current AI cannot substitute for end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Clinical liability, regulatory requirements, and professional standards mandate that licensed psychiatric nurses must participate in treatment team conferences and bear accountability for clinical decisions. Patients and care teams expect human clinical judgment in real-time diagnostic and treatment deliberations.
Adoption barriers	claude-sonnet-5	5/5	Advanced practice psychiatric nurses are licensed clinicians whose clinical judgment and legal accountability in diagnosis/treatment decisions cannot be delegated to non-licensed systems.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of AI systems capable of even partial conference support, combined with required human oversight and integration, would exceed the cost of the nurse's actual participation in the meeting.
Cost vs. human wage	claude-sonnet-5	2/5	Since AI cannot perform the core task, any cost comparison favors human staff; at best AI could cheaply provide note summarization support, not replace participation.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product reliably performs this task. While AI can summarize case information or suggest differential diagnoses from text, active participation in multidisciplinary team conferences with real clinical authority and accountability remains outside production AI capability.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously participates in psychiatric treatment team conferences as a clinical decision-maker; this remains outside current product scope.

Consult with psychiatrists or other professionals when unusual or complex cases are encountered.

CI 0–3 · exposure 0 · augmentation 38 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Healthcare remains a laggard sector for clinical decision-making automation due to regulatory oversight, liability concerns, and the irreplaceability of licensed professional judgment. Adoption of AI for case consultation decisions is minimal.
Sector adoption velocity	claude-sonnet-5	2/5	Healthcare, especially clinical decision-making in psychiatry, is a slow-adopting sector for autonomous AI decision tasks despite growing use of AI documentation tools.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could provide modest support by surfacing similar cases or highlighting atypical findings, but the core decision to consult and the consultation itself require human professional expertise. Augmentation potential is limited because the task inherently depends on clinical judgment.
Augmentation potential	claude-sonnet-5	3/5	AI can help by summarizing patient history, surfacing relevant literature, or drafting case summaries to prepare for consultation, aiding but not replacing the interaction.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires clinical judgment, interpersonal collaboration, and the ability to recognize when a case exceeds one's scope—capabilities that current AI cannot perform autonomously. AI cannot reliably determine case complexity or initiate meaningful consultations with licensed professionals.
Task automatability	claude-sonnet-5	1/5	This is an inherently interpersonal, judgment-driven professional consultation between licensed clinicians about complex clinical decisions; AI cannot substitute for this interaction end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	5/5	This task is legally and professionally gated: only licensed Advanced Practice Psychiatric Nurses can determine case complexity and make consultation decisions. Professional liability, scope of practice regulations, and the requirement for human professional judgment create insurmountable barriers to automation.
Adoption barriers	claude-sonnet-5	5/5	Licensed clinicians must legally consult on complex cases under scope-of-practice and liability rules, and clinical judgment/accountability cannot be delegated to a non-licensed system.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI consultation tools do not yet exist as a cost-competitive replacement for inter-professional clinical consultation. The task's value lies in professional judgment and accountability, which cannot be substituted by AI at lower cost.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI product replacing this task, so no cost comparison favors AI; human professional consultation remains the only viable option.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product performs this task end-to-end. While AI can flag certain patterns in patient data, initiating and conducting professional consultations requires human clinical expertise and professional accountability that current systems lack.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs peer clinical consultation on complex psychiatric cases autonomously; AI tools at best support information lookup, not the consultation itself.

Direct or provide home health services.

CI 0–3 · exposure 0 · augmentation 38 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Home health nursing remains a human-centric, heavily regulated field with minimal AI adoption in production; practitioners operate in distributed, in-home settings with complex patient populations where regulatory and practical barriers prevent meaningful automation.
Sector adoption velocity	claude-sonnet-5	2/5	Home health care is a physically-delivered, in-person service sector with historically low AI/automation penetration compared to office-based information work.
Augmentation potential	claude-haiku-4-5-20251001	2/5	While AI tools might assist with documentation, telehealth scheduling, or clinical decision support in the office, they offer limited augmentation for the core task of directing and providing hands-on home health services, which is inherently human-dependent.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with care coordination, documentation, remote monitoring alerts, and administrative direction of home health services, improving efficiency even though the hands-on care itself remains human-delivered.
Task automatability	claude-haiku-4-5-20251001	1/5	Home health services require direct, in-person clinical assessment, medication management, therapeutic interaction, and physical care that cannot be performed remotely by AI systems. Current AI cannot substitute for the physical presence, hands-on nursing actions, and real-time clinical judgment essential to this task.
Task automatability	claude-sonnet-5	1/5	Directing or providing home health services requires physical presence, hands-on assessment, and real-time clinical judgment in a patient's home environment, none of which current AI can perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Home health services are tightly regulated, requiring state licensure as an Advanced Practice Registered Nurse (APRN), direct patient contact mandates, liability requirements, and legal accountability that cannot be delegated to or substituted by autonomous AI systems.
Adoption barriers	claude-sonnet-5	5/5	Home health services require licensed nursing practice, direct patient contact, legal accountability, and regulatory oversight that mandate human provision and supervision.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Home health nursing is labor-intensive and requires licensed personnel whose presence and judgment cannot be replaced by AI inference; the cost of maintaining human delivery is substantially lower than any hypothetical AI alternative that would need human oversight and backup.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this physical service, so AI cost is not comparable—human labor remains the only viable option, making AI relatively 'more expensive' by default since it cannot do the task.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product can autonomously deliver or direct home health services; this task fundamentally depends on licensed nursing presence and direct patient contact that no current AI system can replicate.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product provides or directs home health nursing visits; this remains a research-stage concept at best, far from physical care delivery.

Participate in activities aimed at professional growth and development, including conferences or continuing education activities.

CI 0–0 · exposure 0 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	This task occurs across diverse healthcare settings, many of which operate with moderate digitization and institutional resistance to automation. Professional development participation is driven by regulatory requirements and human judgment, not by sectoral adoption of AI.
Sector adoption velocity	claude-sonnet-5	1/5	This is not a task sector adoption trends affect, since it is an individual licensure/certification obligation that cannot be delegated to AI or automated workflows.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by identifying relevant conferences, summarizing session content, organizing learning materials, or tracking continuing education credits, which would enhance a nurse's efficiency in managing their professional development pipeline.
Augmentation potential	claude-sonnet-5	3/5	AI can help find relevant courses, summarize research, or curate learning content, offering moderate assistance in identifying and preparing for professional development opportunities.
Task automatability	claude-haiku-4-5-20251001	1/5	This task is inherently about human professional development and attendance at learning events. While AI can help identify relevant conferences or summarize content, it cannot meaningfully participate in or attend continuing education activities on behalf of a healthcare professional, nor can it replace the human act of engaging with professional development.
Task automatability	claude-sonnet-5	1/5	This task fundamentally requires the human clinician to attend, engage with, and internalize professional development activities; AI cannot experience or complete this on the nurse's behalf.9d It is inherently a self-improvement obligation tied to licensure, not a deliverable AI can produce.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Professional licensing and credentialing bodies require licensed healthcare providers to directly participate in continuing education to maintain their credentials. Many professional development activities require human attendance and engagement as a regulatory and ethical mandate for psychiatric nurses.
Adoption barriers	claude-sonnet-5	5/5	Licensing boards mandate that the individual practitioner personally complete continuing education credits and maintain certification, making this a hard, non-delegable requirement.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	This task involves registration fees, travel, time investment, and the nurse's continued salary during participation. AI cannot reduce these costs since it cannot substitute for the nurse's attendance and engagement in learning activities.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this task, so cost comparison is not meaningful; the human must personally complete the requirement regardless of AI cost.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No current AI system can autonomously participate in conferences or continuing education programs. AI tools may assist with content discovery or summarization, but the core task—attending and engaging in professional growth activities—requires human presence and participation that deployed products do not address.
Technical feasibility today	claude-sonnet-5	1/5	No product substitutes for a human's actual participation in continuing education or conferences since this is a personal professional requirement, not an output-based task.

Administer medications, including those administered by injection.

CI 0–0 · exposure 0 · augmentation 25 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	No meaningful AI adoption for medication administration is occurring because the task is legally restricted to licensed practitioners and requires physical presence and clinical judgment in real time.
Sector adoption velocity	claude-sonnet-5	1/5	Healthcare's physical, hands-on clinical tasks show minimal AI adoption for direct physical administration, as this remains a highly regulated, in-person function.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could assist marginally with medication record-keeping, dosage verification, or patient history review before administration, but the core clinical act—injection and medication delivery—receives minimal augmentation benefit.
Augmentation potential	claude-sonnet-5	2/5	AI can assist with dosage calculations, medication interaction checks, or documentation support, but offers minimal direct assistance with the physical act of administering medication or injections.
Task automatability	claude-haiku-4-5-20251001	1/5	Administering medications by injection requires physical manipulation in a clinical setting, real-time patient assessment, and direct patient contact—capabilities entirely outside current AI scope. This task cannot be performed end-to-end by AI systems today.
Task automatability	claude-sonnet-5	1/5	Physical administration of medications, especially injections, requires hands-on patient contact and cannot be performed end-to-end by current AI systems.a
Adoption barriers	claude-haiku-4-5-20251001	5/5	Hard regulatory and legal barriers exist: only licensed nurses (and in some jurisdictions, physicians) are authorized to administer medications and injections. Liability and scope-of-practice laws firmly protect this task.
Adoption barriers	claude-sonnet-5	5/5	Administering medications, particularly injections, requires licensed clinical authority, direct patient contact, and legal/regulatory scope-of-practice requirements for nurses.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI has no meaningful cost advantage because the task involves physical action and direct patient care that only humans can perform; comparison is not applicable, making AI substantially more expensive when considering the need for human oversight.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing the physical act, so AI cost comparison is not applicable/AI is not cheaper since it cannot perform the task at all.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product can physically administer medications or injections. This remains a human-performed clinical procedure with no mature automation products in production.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product physically administers medications or injections; this remains entirely a human physical task.

Related occupations — Healthcare Practitioners & Technical

How to read this

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

What would change this score

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