Nurse Practitioners

29-1171.00
Median wage $132,300/yr323,040 employed (US)Rank #576 of 923 scored · top 62% by substitution

Diagnose and treat acute, episodic, or chronic illness, independently or as part of a healthcare team. May focus on health promotion and disease prevention. May order, perform, or interpret diagnostic tests such as lab work and x rays. May prescribe medication. Must be registered nurses who have specialized graduate education.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution24
Exposure24
Augmentation69

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

27 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

4%

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

Why this score

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

Task automatabilityw 35%23

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

Technical feasibility todayw 20%27

panel mean rating 2.1/5 → substitution pressure 27/100

Cost vs. human wagew 15%28

panel mean rating 2.1/5 → substitution pressure 28/100

Adoption barriersw 20%inverted — strong barriers lower the score17

panel mean rating 4.3/5 (barrier strength) → substitution pressure 17/100

Sector adoption velocityw 10%29

panel mean rating 2.2/5 → substitution pressure 29/100

Task breakdown (27 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.

Provide patients or caregivers with assistance in locating health care resources.

73

CI 59–87 · exposure 70 · augmentation 75 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Health systems and insurance companies are actively adopting AI-driven resource navigation, patient portals, and chatbots to triage resource requests. Major EHR vendors and stand-alone health platforms now embed or offer this functionality as production features. Adoption is widespread in larger, digitized health systems.
Sector adoption velocityclaude-sonnet-53/5Healthcare adoption of AI is accelerating but remains uneven; navigation and referral tools are piloted in many systems but not yet universally deployed at scale.
Augmentation potentialclaude-haiku-4-5-202510014/5AI strongly augments nurse practitioners by instantly surfacing compliant, up-to-date resources; the practitioner can validate, contextualize, and explain options to patients. This raises the throughput and accuracy of resource assistance while keeping the clinician in the verification loop.
Augmentation potentialclaude-sonnet-54/5AI tools can significantly speed up locating appropriate resources, pre-filling referral forms, and summarizing options for NPs and patients, meaningfully boosting efficiency while the clinician retains oversight.
Task automatabilityclaude-haiku-4-5-202510015/5AI systems can reliably search and compile health care resources (clinics, hospitals, insurance networks, support services) based on patient location, insurance, and condition. Retrieval-augmented generation and knowledge base search can reduce the time to locate relevant resources by >50% compared to manual lookup, meeting the automatability threshold.
Task automatabilityclaude-sonnet-53/5AI can search, summarize, and match patients to relevant resources (clinics, specialists, financial aid, community programs) quickly, but personalized triage and follow-up communication still typically involve human judgment and rapport.
Adoption barriersclaude-haiku-4-5-202510012/5Minimal legal or licensing barriers exist for AI-assisted or autonomous resource lookup; it does not constitute medical judgment or treatment. Patient preference for human explanation may create organizational friction, and some health systems may prefer a human touchpoint, but no regulatory requirement mandates practitioner involvement.
Adoption barriersclaude-sonnet-52/5This task is largely administrative/informational rather than requiring licensed clinical judgment, so legal barriers are low, though patient trust and liability for inaccurate referrals create moderate friction.
Cost vs. human wageclaude-haiku-4-5-202510015/5Automated resource location (via chatbot or portal) costs pennies per query once integrated. The loaded wage for a nurse practitioner to spend 15–30 minutes locating resources is $30–100+ per patient. AI cost is at least 10–100× lower.
Cost vs. human wageclaude-sonnet-54/5Automated resource-lookup and referral tools cost far less per interaction than clinician time, though integration with EHRs and local resource databases adds some overhead.
Technical feasibility todayclaude-haiku-4-5-202510014/5Deployed AI systems (chatbots, decision-support tools, health information portals) already perform this task in many health systems today. Integration with provider directories and insurance formularies is standard, though occasional data staleness and edge cases remain. Production use is common but not universal.
Technical feasibility todayclaude-sonnet-53/5Deployed patient-navigation chatbots and resource-finder tools exist in some health systems, but coverage and reliability vary widely and most still require human verification for accuracy and appropriateness.

Schedule follow-up visits to monitor patients or evaluate health or illness care.

57

CI 40–75 · exposure 55 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510014/5Healthcare organizations have widely deployed electronic health records and appointment scheduling systems that automate parts of this workflow; many practices use automated reminders and rule-based scheduling, reflecting fairly rapid adoption in the digitized healthcare sector.
Sector adoption velocityclaude-sonnet-53/5Healthcare has moderate digitization with widespread EHR adoption, but many systems still rely on staff-mediated scheduling rather than fully autonomous AI agents.
Augmentation potentialclaude-haiku-4-5-202510014/5AI scheduling tools and EHR systems already assist nurse practitioners by managing calendar logistics, suggesting follow-up intervals based on clinical protocols, and automating reminders, meaningfully raising productivity while the practitioner retains clinical decision authority.
Augmentation potentialclaude-sonnet-54/5AI scheduling assistants and EHR reminder systems meaningfully reduce administrative burden on NPs while they retain oversight of clinical necessity and timing.
Task automatabilityclaude-haiku-4-5-202510012/5Scheduling appointments requires accessing calendars and patient records, which AI can handle, but the clinical decision of *when* follow-up is needed—timing based on diagnosis, treatment response, and individual patient factors—requires human clinical judgment that current AI cannot reliably replicate end-to-end.
Task automatabilityclaude-sonnet-54/5Scheduling follow-up visits based on clinical protocols or care plans is a structured, rules-based task that current scheduling software and AI assistants can handle end-to-end with significant time savings.ggf
Adoption barriersclaude-haiku-4-5-202510014/5Clinical decisions about follow-up care timing and monitoring strategy typically require accountability by a licensed practitioner; regulatory and liability frameworks generally expect a qualified nurse practitioner or physician to sign off on care plans, creating a legal barrier to full automation.
Adoption barriersclaude-sonnet-52/5No licensure is required to schedule an appointment, though clinical judgment about follow-up timing/urgency may still require NP sign-off in some workflows.
Cost vs. human wageclaude-haiku-4-5-202510014/5Automated scheduling tools have low marginal cost per appointment after initial setup, making them substantially cheaper than nurse practitioner time for the pure scheduling and calendar-management component, though clinical triage of follow-up decisions remains human-dependent.
Cost vs. human wageclaude-sonnet-54/5Automated scheduling systems cost a fraction of clinical staff time spent on manual calendar coordination, though some integration and EHR licensing costs remain.
Technical feasibility todayclaude-haiku-4-5-202510013/5Calendar and scheduling systems with basic rule-based logic exist and are deployed in healthcare settings, but they struggle with the clinical reasoning aspect of determining appropriate follow-up intervals and cannot independently assess when monitoring is necessary without human oversight.
Technical feasibility todayclaude-sonnet-54/5EHR-integrated scheduling systems and patient portals with automated follow-up reminders and booking are already deployed at scale in many health systems today.

Read current literature, talk with colleagues, or participate in professional organizations or conferences to keep abreast of developments in nursing.

46

CI 32–59 · exposure 42 · augmentation 88 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare organizations have adopted literature-monitoring and summarization tools to assist NPs, but adoption remains piecemeal. Many NPs still rely on traditional journal subscriptions, conference attendance, and informal networks rather than full AI-driven knowledge management.
Sector adoption velocityclaude-sonnet-53/5Healthcare is a moderate adopter of AI tools for literature review and CDS, with growing but not yet universal use of AI summarization in clinical CPD workflows.
Augmentation potentialclaude-haiku-4-5-202510014/5AI augmentation is strong here: literature summarization, search prioritization, and AI-generated topic alerts substantially raise an NP's efficiency in scanning and filtering developments without requiring the NP to attend every conference or read every journal.
Augmentation potentialclaude-sonnet-55/5AI substantially boosts efficiency in staying current by summarizing papers, flagging new studies, and answering questions, while the human still engages with colleagues and conferences.
Task automatabilityclaude-haiku-4-5-202510012/5AI can help surface and summarize nursing literature and conference abstracts, but cannot fully replace the nuanced judgment, contextual filtering, and collegial discussion that characterize staying 'abreast of developments.' A human must evaluate relevance and synthesize findings into practice.
Task automatabilityclaude-sonnet-53/5AI can summarize literature and surface relevant papers efficiently, but the task also includes networking, discussion, and conference participation that require human presence and judgment, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510014/5Professional credibility and licensure reinforce the expectation that NPs personally engage with current evidence and peers; regulatory standards emphasize continuing education and staying current as a personal obligation. Organizational and professional norms strongly favor human-driven professional development.
Adoption barriersclaude-sonnet-52/5No licensing requirement blocks use of AI for staying current, though professional norms around CE credits and human networking create some friction.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-powered literature monitoring and summarization tools are relatively inexpensive, but cannot eliminate the NP's time investment in reading, reflecting, and attending conferences. The human cost remains substantial compared to partial automation savings.
Cost vs. human wageclaude-sonnet-54/5AI-based literature synthesis and alert tools are inexpensive relative to a nurse practitioner's time spent manually scanning journals, though human interaction components still require paid time.
Technical feasibility todayclaude-haiku-4-5-202510013/5Deployed AI systems (literature search, summarization, conference alert tools) exist and perform parts of this task reliably, but rely on human curation and judgment. No product fully automates the decision-making about which developments matter for an individual practitioner.
Technical feasibility todayclaude-sonnet-53/5Deployed tools like AI literature summarizers, alerting services, and clinical decision support exist and are used by clinicians, but they cover only the reading/awareness portion, not colleague interaction or conference engagement.

Provide patients with information needed to promote health, reduce risk factors, or prevent disease or disability.

38

CI 25–51 · exposure 38 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of autonomous AI for direct patient education remains slow and cautious, with most deployments limited to supplementary or informational roles. Regulatory scrutiny, malpractice risk aversion, and patient-safety culture in healthcare slow experimental automation.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall remains a slower-adopting sector for patient-facing AI due to regulatory, liability, and workflow integration challenges, despite growing pilot programs.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially assist by summarizing evidence, generating draft educational materials tailored to risk profiles, and flagging relevant prevention guidelines—allowing the nurse practitioner to focus on personalized communication and relationship-building. This augmentation materially increases productivity without removing the human from clinical judgment.
Augmentation potentialclaude-sonnet-54/5AI tools substantially help NPs draft patient education materials, answer common questions, and personalize risk-reduction guidance, improving efficiency while the clinician remains responsible for accuracy and delivery.
Task automatabilityclaude-haiku-4-5-202510012/5Health promotion and disease prevention require contextual judgment, personalization to individual risk profiles, and adaptive communication—tasks at which current AI has limited capability. While AI can generate educational content or identify risk factors from data, the nuanced counseling and tailored guidance central to this task demand human expertise and presence.
Task automatabilityclaude-sonnet-53/5AI chatbots can generate personalized health education and risk-reduction information, but tailoring to a specific patient's clinical context and ensuring accuracy still requires clinician review, so only partial time savings are realized end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Strong legal and regulatory barriers exist: nurse practitioners must maintain clinical accountability for patient education and counseling. Liability concerns are high if harm results from incorrect health information, and many jurisdictions require a licensed practitioner to deliver or co-sign health promotion guidance.
Adoption barriersclaude-sonnet-53/5While NPs must exercise clinical judgment and maintain licensure for care decisions, providing general health information is less strictly regulated, though liability and patient trust still favor human delivery for many interactions.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration of reliable AI systems with oversight for patient safety would require substantial infrastructure and quality assurance. The cost of ensuring clinical accuracy and liability protection approaches or exceeds the labor cost of a nurse practitioner delivering this service directly.
Cost vs. human wageclaude-sonnet-54/5Generating standardized health information and counseling scripts via AI is very cheap compared to a nurse practitioner's time, though oversight costs reduce the full savings somewhat.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed AI system reliably performs autonomous health promotion counseling at clinical standard. Chatbots and educational apps exist but lack the clinical reasoning, patient rapport, and accountability necessary for practicing nurse practitioners to delegate this task, and they operate in narrow, pre-scripted domains.
Technical feasibility todayclaude-sonnet-53/5Patient education tools and AI chat assistants are deployed in some health systems for generating educational materials, but reliability, personalization, and liability concerns keep scope narrow rather than fully autonomous.

Maintain complete and detailed records of patients' health care plans and prognoses.

37

CI 30–45 · exposure 42 · augmentation 88 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5While health IT adoption is common, uptake of AI-driven documentation automation remains slow outside large health systems and academic medical centers. Regulatory caution, clinician skepticism about AI-generated narratives, and integration friction with legacy EHRs limit widespread production deployment.
Sector adoption velocityclaude-sonnet-53/5Healthcare has moderate AI documentation adoption via ambient scribing pilots and growing EHR integration, but overall sector digitization and AI adoption lag behind finance or tech.
Augmentation potentialclaude-haiku-4-5-202510014/5AI assistants that offer real-time clinical note drafting, structured data suggestions, and evidence-based plan templates meaningfully accelerate NP documentation workflow while the clinician remains the decision-maker and certifier. This is among the most mature augmentation use cases in healthcare today.
Augmentation potentialclaude-sonnet-55/5AI scribe and summarization tools substantially reduce time spent on documentation and improve note quality/completeness while the NP remains the responsible record-keeper.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with structured data entry and basic documentation templating, comprehensive health care plan and prognosis records require nuanced clinical judgment, patient context synthesis, and legal-compliant narrative that current systems struggle to generate end-to-end at the 50% time-savings threshold. Human review and revision remain necessary for accuracy and liability.
Task automatabilityclaude-sonnet-53/5AI can draft, summarize, and structure clinical documentation from visit notes or dictation, but final entry, accuracy verification, and clinical judgment about prognosis still require NP review, limiting full end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510014/5Health records are legally regulated (HIPAA, state scope-of-practice laws); NPs must personally author, review, and sign clinical documentation. Liability and patient safety standards mean an NP must certify the clinical accuracy and completeness of records, creating a hard requirement for human oversight.
Adoption barriersclaude-sonnet-54/5Medical record accuracy and legal liability mean a licensed clinician must ultimately verify and sign off on health records; HIPAA and malpractice concerns create strong barriers to full delegation to AI.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI-assisted documentation tools reduce typing time but require integration with EHRs, oversight by licensed clinicians, and ongoing maintenance. The all-in cost per record remains comparable to or higher than the time saved, especially when liability and rework are factored in.
Cost vs. human wageclaude-sonnet-53/5AI scribing tools carry meaningful subscription/integration costs and still require clinician oversight time, so savings are real but not an order-of-magnitude cheaper than the marginal cost of NP documentation time.
Technical feasibility todayclaude-haiku-4-5-202510013/5EHR systems with AI-assisted documentation (e.g., voice-to-text, template population) exist in production, but they still require significant NP oversight, manual editing, and clinical validation. Error rates and incomplete capture remain material for complex cases.
Technical feasibility todayclaude-sonnet-53/5Ambient documentation and AI scribe products (e.g., Nuance DAX, Abridge) are deployed in real clinical settings and reduce charting burden, but they still require clinician review/editing and are not universally adopted or error-free.

Maintain current knowledge of state legal regulations for nurse practitioner practice, including reimbursement of services.

34

CI 29–39 · exposure 30 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI for compliance is slower than other sectors; most NP practices rely on manual monitoring, professional organizations' alerts, and legal counsel rather than automated AI systems for regulatory tracking.
Sector adoption velocityclaude-sonnet-52/5Healthcare is generally a slower-adopting sector for AI-driven compliance and legal-tracking tasks compared to information/finance sectors, with most professionals still relying on professional associations, licensing boards, and manual updates.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially assist NPs by aggregating and summarizing regulatory updates from multiple state sources, flagging changes, and drafting practice policy memos, allowing the NP to focus review effort on interpretation and implementation rather than raw information gathering.
Augmentation potentialclaude-sonnet-54/5AI tools can efficiently aggregate, summarize, and flag changes in state regulations and reimbursement policies, significantly speeding up the research phase even though the practitioner must verify and apply the information.
Task automatabilityclaude-haiku-4-5-202510012/5AI can summarize and track regulatory changes across state sources, but verifying applicability to a specific NP's practice scope, licenses, and reimbursement contracts requires human judgment and contextual knowledge that AI cannot reliably automate end-to-end.
Task automatabilityclaude-sonnet-52/5AI can help surface and summarize regulatory updates, but continuously monitoring, interpreting, and applying legal changes to one's own practice requires ongoing human verification and judgment, so full end-to-end automation with equal quality is not yet achievable.
Adoption barriersclaude-haiku-4-5-202510014/5Nurse practitioners must maintain current licenses and comply with state regulations; failure to stay current can result in liability and license suspension, creating strong legal and professional incentives for human oversight and verification of regulatory information.
Adoption barriersclaude-sonnet-53/5There's no legal requirement that a human personally track regulations, but professional liability and accreditation/licensure obligations create pressure for the practitioner to personally verify legal compliance rather than fully delegate to AI.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-powered regulatory tracking services cost comparable to hiring dedicated compliance staff for larger practices, but setup and integration into clinical workflows adds friction that partly offsets the labor savings.
Cost vs. human wageclaude-sonnet-53/5AI-assisted research tools are cheap to query, but human review, verification against official sources, and liability concerns mean the effective cost including oversight is only modestly better than manual research.
Technical feasibility todayclaude-haiku-4-5-202510013/5Regulatory monitoring and summarization tools exist and are used in healthcare compliance, but they typically flag potential changes rather than fully interpreting their impact on individual NP practice; human review of state boards and payers remains standard.
Technical feasibility todayclaude-sonnet-52/5Legal research and summarization tools exist and can retrieve regulatory text, but no deployed product reliably tracks state-specific NP scope-of-practice and reimbursement changes with the accuracy needed for compliance-grade use.

Educate patients about self-management of acute or chronic illnesses, tailoring instructions to patients' individual circumstances.

29

CI 29–29 · exposure 25 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI for core clinical functions remains slow and cautious, with most deployments in administrative or supportive roles. Patient education remains a high-touch, relationship-dependent function where organizations are piloting automation but not yet moving to production replacement at scale.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall has been slower than information/finance sectors to deploy AI in direct patient-facing clinical education, with adoption mostly at pilot or decision-support stage.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can significantly assist NPs by drafting customized education materials, identifying evidence-based resources, tracking patient responses, and flagging comprehension gaps—allowing the clinician to focus on relationship-building and clinical judgment. This augmentation pattern is already emerging in healthtech products.
Augmentation potentialclaude-sonnet-54/5AI tools can efficiently draft tailored handouts, translate materials, and summarize condition-specific instructions, meaningfully speeding up the NP's preparation while the NP still delivers and personalizes the conversation.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can generate educational materials and draft tailored instructions, true end-to-end automation requires real-time assessment of patient comprehension, emotional state, health literacy, and iterative adjustment—elements that demand human clinical judgment and interpersonal presence. Current AI systems excel at content generation but cannot reliably perform the full patient education loop without significant human oversight.
Task automatabilityclaude-sonnet-52/5AI can generate personalized educational content, but the task requires assessing individual context, verifying understanding, adjusting for health literacy and emotional state in real time, which current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Patient education is often embedded in licensed clinical care; regulatory and liability frameworks expect a qualified clinician to verify that education is appropriate and effective. Standard of care and malpractice liability create strong pressure to retain human clinician responsibility and sign-off on patient education.
Adoption barriersclaude-sonnet-54/5Patient education is part of the licensed scope of practice tied to clinical judgment and liability for treatment plans, and regulatory/organizational expectations generally require a credentialed provider to deliver and document this counseling.
Cost vs. human wageclaude-haiku-4-5-202510013/5AI-driven educational tools have low marginal costs at scale, but integrating them into clinical workflows, ensuring clinical accuracy, and maintaining human oversight add meaningful costs that approximate the labor cost of a nurse practitioner in many settings.
Cost vs. human wageclaude-sonnet-53/5AI-generated educational materials are cheap to produce, but the human clinician time for tailoring and verification is still required, so overall cost savings are moderate rather than dramatic.
Technical feasibility todayclaude-haiku-4-5-202510012/5Chatbots and educational software exist for patient education, but they operate in narrow domains with limited personalization and lack the clinical assessment and real-world adaptability that characterize nurse practitioner-led education. No deployed product reliably substitutes for NP-directed, tailored education across the full range of acute and chronic conditions.
Technical feasibility todayclaude-sonnet-52/5Patient education chatbots and portals exist and are used for supplementary materials, but no deployed product independently and reliably conducts individualized clinical education replacing the NP conversation at scale.

Keep abreast of regulatory processes and payer systems, such as Medicare, Medicaid, managed care, and private sources.

29

CI 16–41 · exposure 17 · augmentation 63 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains relatively laggard in autonomous AI adoption due to regulatory constraints, liability concerns, and organizational conservatism. While some practices use regulatory monitoring services, end-to-end autonomous AI for this task is rare in production.
Sector adoption velocityclaude-sonnet-53/5Healthcare administrative and compliance functions are adopting AI tools at a moderate pace, with pilots for regulatory tracking more common than full production reliance.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist by curating and summarizing regulatory updates, flagging relevant policy changes, and organizing payer documentation—raising the speed and ease of monitoring—but the nurse practitioner must retain final interpretation and compliance responsibility.
Augmentation potentialclaude-sonnet-54/5AI can efficiently aggregate, summarize, and alert practitioners to regulatory and payer changes, meaningfully reducing the time needed to stay informed.
Task automatabilityclaude-haiku-4-5-202510011/5Staying abreast of regulatory processes requires continuous judgment about relevance, interpretation of policy nuance, and integration with clinical practice context. While AI can summarize regulatory updates, the task fundamentally demands human discretion to evaluate impact on patient care and practice operations—not a task that achieves ≥50% time savings end-to-end.
Task automatabilityclaude-sonnet-52/5Staying current on regulatory and payer policy changes involves continuous monitoring, interpretation, and application to practice that AI can support but not fully replace end-to-end.5
Adoption barriersclaude-haiku-4-5-202510014/5High barriers exist: nurse practitioners have direct liability for compliance with Medicare, Medicaid, and managed care regulations; errors in payer reimbursement and coverage rules can result in claim denials, fraud liability, and regulatory penalties. The human practitioner must legally understand and act on these requirements.
Adoption barriersclaude-sonnet-52/5No strict licensing barrier prevents using AI to track this information, though liability for missing critical compliance updates creates some caution.
Cost vs. human wageclaude-haiku-4-5-202510012/5Integration and ongoing oversight costs are substantial because regulatory interpretation requires expert review. A nurse practitioner's time remains cheaper than the combined cost of AI tools, integration, and the mandatory human sign-off on regulatory compliance decisions.
Cost vs. human wageclaude-sonnet-53/5AI tools can cheaply summarize and flag regulatory updates, but human verification and clinical judgment remain necessary, keeping costs roughly comparable when oversight is factored in.
Technical feasibility todayclaude-haiku-4-5-202510012/5No deployed product reliably performs this task independently. AI tools can draft summaries of regulatory changes or flag updates, but no production system credibly keeps a practitioner compliant and current across multiple payer systems without substantial human filtering and judgment.
Technical feasibility todayclaude-sonnet-52/5Some products (e.g., compliance/coding assistants, summarization tools) exist to track payer/regulatory updates, but none reliably serve as the sole source of truth for NPs in production at scale.

Counsel patients about drug regimens and possible side effects or interactions with other substances, such as food supplements, over-the-counter (OTC) medications, or herbal remedies.

27

CI 25–29 · exposure 25 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Health systems have adopted EHR-integrated drug interaction alerts and clinical decision support, but these remain narrow assistive tools; autonomous AI counseling of patients on regimens and side effects is not standard practice. Adoption remains pilot-stage and cautious due to liability and regulatory risk.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially direct patient care and clinical counseling, adopts AI more slowly than other sectors due to regulatory, liability, and workflow integration challenges, with most AI use limited to documentation or decision support.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist by instantly surfacing drug interactions, side effect profiles, contraindication flags, and evidence summaries, allowing the NP to focus dialogue on patient concerns and shared decision-making. This augmentation significantly raises NP productivity and safety during counseling.
Augmentation potentialclaude-sonnet-54/5AI tools substantially help NPs by surfacing interaction data, generating patient-friendly explanations, and flagging risks, meaningfully boosting counseling efficiency and accuracy while the NP remains the decision-maker.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can retrieve and cross-reference drug interaction data and side effect information rapidly, counseling requires nuanced risk assessment, patient-specific contraindication evaluation, and adaptive communication based on individual health history and concerns. Current systems cannot reliably replicate the clinical judgment and personalized dialogue needed to meet the ≥50% time-saving threshold at equal quality.
Task automatabilityclaude-sonnet-52/5AI can generate accurate drug interaction information but the counseling task requires personalized clinical judgment, rapport-building, and legal accountability that current systems cannot fully replicate end-to-end.'
Adoption barriersclaude-haiku-4-5-202510014/5Pharmacological counseling carries legal and clinical liability; regulators and malpractice doctrine expect a licensed provider (NP or pharmacist) to ensure accuracy and appropriateness of drug guidance. Direct patient contact and professional accountability create strong legal and organizational barriers to full automation.
Adoption barriersclaude-sonnet-54/5Prescribing counsel and medication management are within licensed scope of practice with legal liability attached, requiring a credentialed provider to deliver or supervise this counseling.
Cost vs. human wageclaude-haiku-4-5-202510012/5Current AI tools require significant clinical oversight and integration into workflows, and the liability and verification burden remain high. The all-in cost of AI-assisted drug counseling with necessary human review likely exceeds the loaded wage of a nurse practitioner handling the same task.
Cost vs. human wageclaude-sonnet-53/5AI-assisted lookup tools are cheap, but the overall counseling task still requires an NP's time for judgment and liability, so cost savings are partial rather than an order-of-magnitude reduction.
Technical feasibility todayclaude-haiku-4-5-202510012/5Products like drug interaction checkers and AI-assisted clinical decision support exist, but they function as narrow reference tools, not autonomous counselors. No deployed system reliably performs the full task—patient assessment, individualized risk communication, and follow-up dialogue—without material error risk or human oversight.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools and chatbots (e.g., drug interaction checkers) exist and are used as reference aids, but no deployed product independently counsels patients on regimens in place of an NP in production settings.

Recommend interventions to modify behavior associated with health risks.

27

CI 25–29 · exposure 25 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI is measured but cautious; clinical decision-support tools see pilot and limited production use, but full replacement or autonomous behavior-modification recommendation is rare. Regulatory and liability concerns slow deep adoption.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall lags in AI adoption for clinical decision-making due to regulatory caution, EHR integration challenges, and liability concerns, despite pilots in digital health coaching.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can substantially augment NP productivity by surfacing evidence-based behavior-change frameworks, predicting which interventions might suit a patient profile, and drafting talking points, allowing the NP to focus on patient engagement and clinical judgment. This is a strong augmentation scenario even if full automation is not feasible.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist NPs by surfacing evidence-based intervention options, generating patient education materials, and flagging risk factors, improving efficiency while the NP retains decision authority.
Task automatabilityclaude-haiku-4-5-202510012/5AI can generate behavior-modification suggestions based on clinical guidelines, but the task requires individualized risk assessment, rapport-building, and clinical judgment to tailor interventions to patient values, motivations, and psychosocial context—elements current systems struggle with reliably. The human NP must remain central to personalizing recommendations and assessing readiness for change.
Task automatabilityclaude-sonnet-52/5AI can draft generic behavior-change recommendations (smoking cessation, diet, exercise) but tailoring to patient history, comorbidities, and motivation requires clinical judgment and relational rapport that current systems cannot fully replicate end-to-end.
Adoption barriersclaude-haiku-4-5-202510014/5Nurse practitioners are licensed professionals; standards of care, liability, and malpractice law typically require a qualified human clinician to take responsibility for clinical recommendations. Regulatory and professional standards create strong friction against full automation.
Adoption barriersclaude-sonnet-54/5NPs are licensed providers whose recommendations carry legal and clinical accountability, and health behavior counseling often requires documented clinical judgment, creating substantial regulatory and liability barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tooling for clinical decision support has meaningful upfront and integration costs, and the task still requires NP time for assessment, patient interaction, and oversight. The cost savings are modest because the human remains the primary decision-maker.
Cost vs. human wageclaude-sonnet-53/5AI-generated counseling content is cheap to produce, but the overall cost includes clinician review, liability, and patient engagement, making the net savings moderate rather than dramatic.
Technical feasibility todayclaude-haiku-4-5-202510012/5Decision-support tools exist that suggest evidence-based interventions, but no deployed system reliably performs this task end-to-end without human oversight; clinical appropriateness and ethical safety require human judgment. Products are narrow and require significant NP input and validation.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools and chatbots offer lifestyle counseling suggestions, but no deployed product autonomously delivers and adapts behavioral health interventions reliably in real NP practice without clinician oversight.

Maintain departmental policies and procedures in areas such as safety and infection control.

25

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

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI for policy management remains minimal; most organizations still rely on manual policy review cycles and dedicated compliance staff, with only early-stage pilots of AI-assisted policy tools in select institutional settings.
Sector adoption velocityclaude-sonnet-52/5Healthcare administrative functions are adopting AI slowly relative to information/professional services sectors, with pilots for documentation more common than for policy governance tasks.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can meaningfully assist by drafting policy updates, identifying outdated language, flagging regulatory gaps, and suggesting compliance improvements, helping humans maintain policies more efficiently while they retain final decision-making authority.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by drafting policy language, summarizing regulatory updates, and flagging compliance gaps, substantially speeding the work while the nurse practitioner retains final judgment and accountability.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can help draft or update policy documents and flag compliance gaps, maintaining policies requires organizational judgment, stakeholder consultation, and accountability that remains fundamentally human; AI cannot independently ensure organization-wide policy adherence or adjust procedures based on evolving clinical contexts.
Task automatabilityclaude-sonnet-52/5Drafting and updating policy documents can be partly automated, but maintaining departmental policies requires ongoing clinical judgment, contextual knowledge, and institutional coordination that AI cannot fully replace end-to-end.dry
Adoption barriersclaude-haiku-4-5-202510014/5Healthcare policy maintenance and safety compliance are heavily regulated; legal responsibility for departmental safety and infection control typically rests with licensed practitioners or administrators, creating both regulatory and liability barriers to full automation.
Adoption barriersclaude-sonnet-54/5Healthcare policy-setting is subject to regulatory, accreditation, and liability requirements that typically mandate licensed clinical staff involvement and sign-off, creating strong barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5The task requires skilled clinical judgment and organizational accountability that AI cannot fully replace; integration costs for compliance systems plus ongoing human oversight likely approach or exceed the cost of a human policy manager, especially given error-cost sensitivity.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply draft or update text but the human oversight, clinical validation, and organizational buy-in required keep total cost comparable to or only modestly less than human-led policy maintenance.
Technical feasibility todayclaude-haiku-4-5-202510012/5Current AI products can assist with policy document generation and compliance checking, but no deployed systems reliably maintain departmental policies end-to-end; production implementations remain limited to narrow support functions rather than full policy stewardship.
Technical feasibility todayclaude-sonnet-52/5AI tools exist for drafting compliance documents and summarizing regulatory guidance, but no deployed product autonomously maintains and enforces departmental safety/infection control policy in production healthcare settings.

Order, perform, or interpret the results of diagnostic tests, such as complete blood counts (CBCs), electrocardiograms (EKGs), and radiographs (x-rays).

25

CI 20–30 · exposure 30 · augmentation 63 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Adoption of AI-assisted radiology interpretation is underway in larger health systems, but most primary care and rural settings have not widely deployed such tools. Test ordering remains manual and cognitive; adoption of automation here is slow due to regulatory, liability, and organizational inertia in healthcare.
Sector adoption velocityclaude-sonnet-52/5Healthcare is a highly regulated, historically slower-adopting sector for autonomous AI decision-making, though AI-assisted radiology and EKG tools are seeing growing but still limited deployment.
Augmentation potentialclaude-haiku-4-5-202510013/5AI augmentation is proven for image and EKG interpretation (second-read support, flagging abnormalities), raising NP diagnostic confidence and speed on those components. However, test ordering requires human clinical reasoning, so augmentation covers only part of the task and does not transform overall productivity substantially.
Augmentation potentialclaude-sonnet-54/5AI substantially augments interpretation speed and flags anomalies in EKGs, CBC patterns, and radiographs, helping NPs work faster and catch findings, while the clinician retains final diagnostic and ordering authority.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist with image interpretation (radiology) and some EKG analysis via proven algorithms, ordering tests requires clinical judgment about patient context and contraindications, and performing tests (EKGs, phlebotomy) requires hands-on clinical skills. No end-to-end AI system meets the 50% time-saving bar across all three components reliably today.
Task automatabilityclaude-sonnet-52/5AI can assist in interpreting some diagnostic images or lab patterns, but ordering tests requires clinical judgment integrating patient history, and final interpretation/responsibility cannot be fully offloaded to AI today.5
Adoption barriersclaude-haiku-4-5-202510014/5Nurse practitioners must be licensed to order and interpret diagnostic tests; liability and regulatory requirements (CLIA for labs, FDA oversight for imaging AI) create hard and soft barriers. Clinical responsibility for test selection and result interpretation cannot be delegated to an algorithm without legal and professional liability concerns.
Adoption barriersclaude-sonnet-55/5Ordering and interpreting diagnostic tests for treatment decisions is a licensed clinical act with strict liability and regulatory requirements; a qualified practitioner must legally take responsibility for these medical decisions.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI interpretation tools for imaging and EKG reduce per-case analysis cost modestly, but the full task (ordering, performing, interpreting) remains dominated by NP labor. Integration, regulatory compliance, and required clinician oversight mean all-in cost is still higher than or comparable to traditional NP workflow.
Cost vs. human wageclaude-sonnet-52/5AI diagnostic tools have licensing and integration costs plus required clinician oversight, so while per-read cost may be lower for narrow tasks, the full task including ordering and clinical correlation still requires the NP, keeping costs comparable rather than dramatically cheaper.
Technical feasibility todayclaude-haiku-4-5-202510013/5FDA-cleared AI products exist for radiology and EKG interpretation in production settings, but they function as assistive tools requiring clinician review, not independent performers. Test ordering and blood draws remain entirely human-dependent. Deployed systems have material error rates and narrow scope relative to full task scope.
Technical feasibility todayclaude-sonnet-52/5FDA-cleared AI tools exist for narrow tasks like EKG arrhythmia detection or radiograph triage, but these are decision-support adjuncts, not autonomous interpreters replacing clinician judgment across the full range of tests described.

Recommend diagnostic or therapeutic interventions with attention to safety, cost, invasiveness, simplicity, acceptability, adherence, and efficacy.

24

CI 20–28 · exposure 25 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Healthcare has moderate AI adoption in some domains (imaging, admin), but clinical decision-making adoption remains cautious due to liability, regulatory scrutiny, and clinician skepticism. Pilots are common; routine AI-led therapeutic recommendation is rare in production.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall is a slower-adopting, highly regulated sector; AI is used in pilots and as adjunct decision support but not as an autonomous recommender in mainstream practice.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist by synthesizing guidelines, flagging evidence, suggesting drug interactions, and cost-comparing options, thereby accelerating the evidence-gathering phase of recommendation. The NP remains the decision-maker, but productivity on information synthesis is substantially raised.
Augmentation potentialclaude-sonnet-54/5AI clinical decision support, literature synthesis, and diagnostic aids meaningfully help NPs weigh interventions faster and surface evidence, improving their productivity while they retain final judgment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in filtering evidence and suggesting treatment options, the multifactorial balancing of safety, cost, invasiveness, and patient-specific factors requires human clinical judgment. Current systems cannot reliably weigh these competing priorities or handle the full complexity of individual patient context end-to-end.
Task automatabilityclaude-sonnet-52/5This requires integrated clinical judgment across patient-specific factors, ethical/legal accountability, and nuanced tradeoffs that current AI cannot fully own end-to-end despite decision-support capabilities.
Adoption barriersclaude-haiku-4-5-202510014/5Strong regulatory and liability barriers exist: recommendations that cause harm carry legal and malpractice risk; NPs remain liable for diagnostic and therapeutic decisions; and clinical autonomy/oversight requirements are baked into scope-of-practice regulations and standard of care.
Adoption barriersclaude-sonnet-55/5Prescribing and clinical recommendations require licensed practitioner authority and legal accountability, making this a hard-barrier task under current regulatory and liability frameworks.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI decision support tools have modest upfront costs but require integration, maintenance, and clinician oversight. The savings are incremental (faster literature review, fewer errors), not order-of-magnitude, compared to NP salaries for this high-stakes cognitive work.
Cost vs. human wageclaude-sonnet-52/5AI decision-support can lower some cognitive load cheaply, but the human NP must still review, contextualize, and bear liability, so the overall cost of the full task remains close to the human baseline.
Technical feasibility todayclaude-haiku-4-5-202510012/5Decision support tools exist (e.g., clinical guidelines, drug interaction checkers), but no deployed product reliably performs independent therapeutic recommendation with the nuanced trade-off analysis this task requires. Systems may suggest options but require substantial human review and override rates remain high.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools exist and are used to suggest options, but no deployed product independently recommends interventions with full accountability for safety and patient-specific tradeoffs at scale.

Develop treatment plans, based on scientific rationale, standards of care, and professional practice guidelines.

21

CI 20–23 · exposure 25 · augmentation 75 · importance 4.9/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI is cautious, pilots are common, but autonomous treatment planning in production remains rare; regulatory, liability, and clinical-culture barriers keep adoption slow despite information-sector digitization.
Sector adoption velocityclaude-sonnet-53/5Healthcare is adopting AI decision support tools steadily but cautiously, with pilots and narrow deployments common rather than deep, fast-scaling automation as seen in finance or tech.
Augmentation potentialclaude-haiku-4-5-202510014/5AI tools demonstrably assist nurse practitioners by rapidly surfacing relevant guidelines, flagging drug interactions, and suggesting evidence-based options—transforming the speed and comprehensiveness of plan development while the practitioner retains decision authority.
Augmentation potentialclaude-sonnet-54/5AI clinical decision support and literature synthesis tools meaningfully speed up guideline lookup and draft plan generation, letting NPs focus on judgment and patient-specific customization.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can retrieve clinical guidelines and generate draft treatment plan components, the task requires integrating patient-specific context, clinical judgment, and standards of care with significant liability—no current system reliably produces treatment plans a nurse practitioner can use without substantial revision and human oversight.
Task automatabilityclaude-sonnet-52/5AI can suggest evidence-based treatment options and draft plans, but integrating patient-specific nuance, comorbidities, and shared decision-making requires clinical judgment beyond current end-to-end automation.
Adoption barriersclaude-haiku-4-5-202510015/5Treatment plans constitute a core licensed function; nurse practitioners bear legal and clinical liability for the plan, and regulations typically require a licensed provider to make and sign the clinical decision—an AI system cannot legally assume this responsibility.
Adoption barriersclaude-sonnet-55/5Treatment planning is a core licensed clinical act requiring an NP's legal authority and signature, with strong liability and regulatory constraints preventing full substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for clinical decision support is inexpensive, but full-task cost including human verification, liability insurance, and oversight still approaches or exceeds the marginal cost of a nurse practitioner's time given the critical nature of plan validation.
Cost vs. human wageclaude-sonnet-52/5AI can cheaply generate draft recommendations, but the need for licensed oversight, liability review, and EHR integration keeps overall cost comparable to human-driven workflows.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI clinical decision support tools exist and can assist with guideline-based recommendations, but deployed systems typically function as reference aids rather than autonomous treatment plan generators; production reliability remains limited for complex, individualized plans across diverse conditions.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools exist and are used to suggest guideline-based options, but no deployed product autonomously finalizes treatment plans without clinician review at scale.

Treat or refer patients for primary care conditions, such as headaches, hypertension, urinary tract infections, upper respiratory infections, and dermatological conditions.

21

CI 20–23 · exposure 25 · augmentation 75 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510013/5Health systems are adopting clinical decision support and EHR-integrated AI tools at moderate pace, but these are augmentative, not replacive. Production autonomous treatment systems do not exist; adoption is limited to pilot programs and narrow use cases with heavy human oversight.
Sector adoption velocityclaude-sonnet-52/5Healthcare, especially primary care delivery, is a historically slow adopter of AI due to regulatory, liability, and workflow integration hurdles, with pilots more common than production deployment.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered diagnostic support, treatment guideline integration, and literature retrieval substantially assist NPs in managing routine primary care conditions by reducing cognitive load and improving guideline adherence. These tools can speed diagnosis and treatment decisions while keeping the clinician in full control of patient care decisions.
Augmentation potentialclaude-sonnet-54/5AI tools significantly assist NPs via differential diagnosis support, dermatological image analysis, clinical documentation, and treatment guideline lookup, improving efficiency while the clinician remains responsible for decisions.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in differential diagnosis and treatment recommendations for straightforward cases, treating patients requires physical examination, direct patient interaction, clinical judgment under uncertainty, and legal accountability that AI cannot fully substitute. Current systems can support decision-making but cannot replace the NP's clinical reasoning or obtain informed consent.
Task automatabilityclaude-sonnet-52/5Diagnosis and treatment decisions require hands-on examination, patient interaction, and clinical judgment with legal accountability that current AI cannot fully replicate end-to-end.assistance is possible but full task substitution is not near the 50% time-saving bar at equal quality.
Adoption barriersclaude-haiku-4-5-202510015/5Regulatory and legal barriers are absolute: state licensing boards require a licensed healthcare provider (MD, DO, NP, PA) to evaluate, treat, and be liable for patient outcomes. Malpractice liability, informed consent requirements, and prescriptive authority are tied to individual licensed practitioners and cannot be transferred to AI systems.
Adoption barriersclaude-sonnet-55/5Diagnosing and treating patients legally requires a licensed practitioner; scope-of-practice laws and malpractice liability create hard barriers to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5High-quality clinical decision support systems require significant infrastructure, medical data integration, and ongoing human oversight. The cost of implementing and maintaining such systems across primary care settings, plus required human review, approaches or exceeds the cost of direct NP labor.
Cost vs. human wageclaude-sonnet-52/5While AI inference is cheap, the requirement for licensed clinician oversight, liability, and physical exam components means overall cost savings are modest, not order-of-magnitude.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI-powered diagnostic decision support tools exist in production (e.g., symptom checkers, clinical decision support), but they operate only as assistants, not autonomous providers. No deployed system reliably performs end-to-end primary care treatment independently, and regulatory barriers prevent AI from functioning as the treating provider.
Technical feasibility todayclaude-sonnet-52/5AI diagnostic support tools (e.g., symptom checkers, dermatology image classifiers) exist but are used as adjuncts, not as autonomous primary care providers in production settings.

Diagnose or treat acute health care problems, such as illnesses, infections, or injuries.

20

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

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare has adopted AI-assisted diagnosis in select, controlled settings (radiology, pathology) but remains cautious about autonomous decision-making in acute care. Adoption of AI for direct diagnosis or treatment ordering remains limited by regulatory scrutiny, liability concerns, and institutional conservatism, despite high digitization.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI cautiously due to regulatory, liability, and safety concerns; clinical decision-support pilots are more common than deep production deployment for autonomous diagnosis.
Augmentation potentialclaude-haiku-4-5-202510014/5AI tools demonstrably assist nurse practitioners by rapidly suggesting differential diagnoses, flagging drug interactions, and supporting evidence-based protocol selection, meaningfully raising their diagnostic and treatment planning productivity while keeping the clinician accountable in the loop.
Augmentation potentialclaude-sonnet-54/5AI tools (e.g., symptom checkers, clinical decision support, documentation assistants) meaningfully speed up differential diagnosis and treatment planning while the NP remains responsible for final decisions.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in narrowing diagnostic possibilities and suggesting treatments for clearly-defined acute conditions, but the task requires integration with patient examination, real-time clinical judgment, and adaptation to unexpected presentations. Current systems lack the multimodal assessment and liability-critical decision-making autonomy to perform end-to-end diagnosis or treatment at 50% time savings without significant human oversight.
Task automatabilityclaude-sonnet-52/5Diagnosis and treatment of acute conditions require physical examination, real-time patient interaction, and accountable clinical judgment that current AI cannot perform end-to-end; AI can support differential diagnosis but not fully replace the task.
Adoption barriersclaude-haiku-4-5-202510015/5Diagnosis and treatment of acute conditions is legally reserved to licensed healthcare professionals in most jurisdictions; liability for adverse outcomes falls on the clinician, not the system vendor. Regulatory bodies (FDA, state medical boards) mandate human accountability, and patients expect a qualified human to assess and take responsibility for their care.
Adoption barriersclaude-sonnet-55/5Diagnosing and treating patients is legally restricted to licensed practitioners under scope-of-practice laws, with high liability exposure, making this a hard-barrier task.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI diagnostic tools and infrastructure require substantial upfront investment, continuous validation, and integration with EHR systems. When accounting for liability coverage, regulatory compliance, and required human oversight, the per-task cost is not yet substantially cheaper than employing nurse practitioners.
Cost vs. human wageclaude-sonnet-52/5AI diagnostic aids are cheap per query, but the overall cost of care still requires a licensed provider for exam, treatment, and liability, so total cost savings versus the clinician's wage are modest.
Technical feasibility todayclaude-haiku-4-5-202510012/5Clinical decision support tools and AI-assisted diagnosis systems exist in medical institutions but primarily serve as suggestions rather than autonomous performers. These tools have material error rates on rare or atypical presentations and rely on human clinicians to validate, order imaging, and bear legal responsibility for treatment decisions.
Technical feasibility todayclaude-sonnet-52/5Diagnostic decision-support tools exist and are used in clinical settings, but no deployed product independently diagnoses and treats acute conditions without a licensed clinician driving and finalizing care.

Analyze and interpret patients' histories, symptoms, physical findings, or diagnostic information to develop appropriate diagnoses.

18

CI 16–20 · exposure 25 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare sectors show cautious, slow adoption of AI in core diagnostic tasks—mostly in pilots and narrow specialties (radiology reading assistance) rather than widespread NP-level diagnosis automation. Regulatory conservatism, liability concerns, and patient safety culture constrain rapid deployment.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall is a slower-adopting, highly regulated sector; AI diagnostic aids are in limited pilot or adjunct use rather than widespread production deployment for autonomous diagnosis.
Augmentation potentialclaude-haiku-4-5-202510014/5AI excels at augmenting NP diagnostic work: rapid review of medical literature, flagging relevant patient data, generating differential lists, and highlighting anomalous findings. NPs using AI-assisted tools demonstrably improve diagnostic speed and comprehensiveness while retaining final judgment authority.
Augmentation potentialclaude-sonnet-54/5AI can meaningfully assist by summarizing histories, flagging possible diagnoses, and surfacing relevant literature or lab correlations, improving NP efficiency while the clinician retains final diagnostic responsibility.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can assist in pattern recognition across medical data and suggest differential diagnoses from symptom sets, the task requires synthesizing complex patient narratives, contextual judgment, and integration of subtle clinical findings that current systems handle incompletely. End-to-end autonomous diagnosis replacing the NP at ≥50% time savings with equal quality remains undemonstrated in production.
Task automatabilityclaude-sonnet-52/5AI can support differential diagnosis generation from structured data, but synthesizing full patient history, physical exam findings, and clinical judgment into a final diagnosis still requires human clinical reasoning and accountability, so end-to-end automation with equal quality is not yet achieved.
Adoption barriersclaude-haiku-4-5-202510015/5Strong regulatory and legal barriers exist: diagnosis is a core licensed professional function in most jurisdictions, clinical liability falls on the diagnosing practitioner, and malpractice law creates error-cost asymmetry that makes autonomous systems untenable. Human clinical judgment and legal responsibility cannot be delegated.
Adoption barriersclaude-sonnet-55/5Diagnosis is a licensed clinical act requiring a credentialed provider to legally render and sign off on it, with high liability exposure, making this one of the most protected task types.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure, integration, validation, and continuous oversight required to deploy AI diagnosis in clinical settings remains expensive; combined with liability and malpractice concerns, the all-in cost per diagnostic act likely exceeds an NP's loaded wage.
Cost vs. human wageclaude-sonnet-52/5AI tools can cheaply generate diagnostic suggestions, but the cost of required clinician oversight, verification, and liability management keeps overall cost per reliable diagnosis close to or above human-only costs.
Technical feasibility todayclaude-haiku-4-5-202510012/5Clinical decision support tools (e.g., diagnostic assistants) exist but operate narrowly, often requiring curated inputs and producing outputs that still require substantial NP review. No deployed system reliably performs full diagnostic synthesis across diverse presentations without material error rates or human gatekeeping.
Technical feasibility todayclaude-sonnet-52/5Clinical decision-support and diagnostic-suggestion tools exist (e.g., symptom checkers, AI diagnostic aids) but are used as adjuncts with significant error rates and narrow validated scope, not as autonomous diagnosticians in production.

Diagnose or treat chronic health care problems, such as high blood pressure and diabetes.

18

CI 16–20 · exposure 25 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare sectors are early-stage in AI adoption for clinical decision-making; pilots of diagnostic aids exist but production deployment of autonomous systems remains rare due to regulatory caution, liability concerns, and clinical conservatism. Adoption velocity is slower than in information or finance sectors.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI slowly for diagnosis/treatment due to regulatory, liability, and workflow integration hurdles, with most current use being decision support rather than autonomous diagnosis or treatment.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can effectively augment nurse practitioners by surfacing relevant clinical evidence, highlighting risk factors in patient data, suggesting differential diagnoses, and monitoring lab trends—tools that measurably improve clinical workflow and decision quality while the practitioner retains final judgment and accountability.
Augmentation potentialclaude-sonnet-54/5AI tools (risk calculators, EHR summarization, guideline retrieval, predictive analytics) meaningfully assist NPs in monitoring and managing chronic conditions, improving efficiency while the clinician retains final decision authority.
Task automatabilityclaude-haiku-4-5-202510012/5Diagnosis of chronic conditions requires integration of patient history, physical examination findings, lab results, and clinical judgment. While AI can assist with data analysis and evidence retrieval, current systems cannot independently perform the full diagnostic and treatment-planning process with safety and the required 50% time savings at equal quality; human clinical judgment remains essential.
Task automatabilityclaude-sonnet-52/5Diagnosis and treatment of chronic conditions requires physical exam, longitudinal judgment, patient interaction, and accountability that current AI cannot perform end-to-end; AI can support but not replace this clinical workflow at the 50% time-saving-at-equal-quality bar.
Adoption barriersclaude-haiku-4-5-202510015/5Nurse practitioners hold state licensure and are legally required to provide clinical diagnosis and treatment; unauthorized diagnosis by non-licensed entities carries significant malpractice and regulatory liability. State medical boards and healthcare law create hard barriers requiring a licensed clinician to make these decisions.
Adoption barriersclaude-sonnet-55/5Diagnosing and treating chronic disease is legally restricted to licensed practitioners (NPs, physicians) under scope-of-practice and prescribing regulations, creating a hard legal barrier to full automation.
Cost vs. human wageclaude-haiku-4-5-202510011/5Nurse practitioners diagnose and treat patients in ongoing relationships requiring accountability, malpractice liability, and integration with care systems. The cost of AI systems, oversight, validation, and liability management currently exceeds the loaded hourly wage of a nurse practitioner for this task.
Cost vs. human wageclaude-sonnet-52/5AI inference is cheap, but the task requires licensed oversight, liability coverage, and clinical integration, making the all-in cost of a safe AI-supported workflow still comparable to or above human-only costs in most settings.
Technical feasibility todayclaude-haiku-4-5-202510012/5Decision-support tools and diagnostic aids exist in some EHR systems, but no deployed product reliably performs independent diagnosis and treatment of chronic diseases at clinical-grade quality. Products show promise in narrowly-scoped scenarios but lack the reliability and integration needed for production use across the complexity of chronic disease management.
Technical feasibility todayclaude-sonnet-52/5Clinical decision-support tools exist and are used for suggestions (e.g., risk scoring, guideline recall), but no deployed product independently diagnoses or manages chronic disease in production without a licensed clinician performing the core task.

Prescribe medication dosages, routes, and frequencies, based on such patient characteristics as age and gender.

17

CI 11–23 · exposure 17 · augmentation 75 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains heavily regulated and conservative in automation of clinical decision-making. While EHR-integrated dosage calculators are common, they function as assistive tools only. Meaningful autonomous prescription adoption is minimal; regulatory and liability concerns keep the human in the loop.
Sector adoption velocityclaude-sonnet-53/5Healthcare has moderate EHR-integrated decision support and dosing calculators in production, but full AI-driven prescribing pilots remain rare due to regulatory and safety constraints.
Augmentation potentialclaude-haiku-4-5-202510014/5Current AI tools (drug databases, interaction checkers, dosage calculators integrated into EHRs) meaningfully assist nurse practitioners by reducing manual lookup time and catching contraindications. These systems raise practitioner productivity by automating routine verification while the clinician retains judgment and accountability.
Augmentation potentialclaude-sonnet-54/5AI-based dosing calculators, drug interaction checkers, and clinical decision support meaningfully speed up and improve accuracy of prescribing decisions while the NP retains final authority.
Task automatabilityclaude-haiku-4-5-202510011/5While AI can assist with dosage calculations, prescribing requires synthesis of complex patient histories, contraindication checking, and clinical judgment that goes far beyond a lookup table. Current AI cannot reliably handle the full end-to-end decision—including safety verification, patient-specific adjustments, and medicolegal responsibility—without a licensed human maintaining control.
Task automatabilityclaude-sonnet-52/5Prescribing requires synthesizing patient history, contraindications, and clinical judgment with legal accountability; AI can suggest dosing but cannot autonomously complete this task end-to-end at equal quality today.
Adoption barriersclaude-haiku-4-5-202510015/5Prescribing is a regulated, licensed clinical activity. State boards of nursing and federal law require a licensed nurse practitioner (with appropriate authority) to take personal, legal responsibility for each prescription. No fully autonomous AI system can legally sign or authorize a prescription; human licensure is a hard barrier.
Adoption barriersclaude-sonnet-55/5Prescribing is tightly regulated; only licensed prescribers (NPs, physicians) can legally authorize medication orders, making this a hard legal and liability barrier to full automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI decision-support tools are relatively cheap to operate, but they do not eliminate the nurse practitioner's labor—oversight and final decision-making remain required. The cost of the technology plus required human review likely approaches or exceeds the cost of direct human prescribing without AI assistance.
Cost vs. human wageclaude-sonnet-52/5AI decision-support tools are cheap to run, but the mandatory human prescriber review, liability insurance, and oversight keep overall cost comparable to or only modestly below the human-only workflow.
Technical feasibility todayclaude-haiku-4-5-202510012/5Decision-support tools exist that suggest dosages based on patient demographics, but no deployed product autonomously prescribes medications. Clinical validation systems and drug interaction checkers are only aids; the licensed practitioner must review and authorize every prescription, making true autonomous performance infeasible today.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools offer dosing recommendations and drug interaction checks, but no deployed product independently prescribes medications without a licensed prescriber's review and sign-off.

Perform routine or annual physical examinations.

16

CI 11–20 · exposure 20 · augmentation 63 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains slower to adopt automation than information sectors due to regulatory constraints, liability concerns, and patient safety emphasis. Pilot AI tools for documentation and decision support exist, but actual displacement of the examination task itself is minimal in production.
Sector adoption velocityclaude-sonnet-52/5Healthcare overall adopts AI slowly for direct clinical care due to regulation and safety concerns, though administrative/documentation AI adoption is increasing faster.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist by automating documentation (medical scribe tools), flagging abnormal vital signs or findings, suggesting differential diagnoses, and extracting key data—all within the nurse practitioner's clinical workflow. This augmentation improves efficiency and completeness without removing the human from direct patient contact.
Augmentation potentialclaude-sonnet-53/5AI can assist via ambient scribing, pre-visit symptom summarization, and decision support to prepare for or follow up on the exam, improving efficiency without replacing hands-on assessment.
Task automatabilityclaude-haiku-4-5-202510012/5While AI can support some components (vital sign analysis, preliminary symptom screening), routine physical examinations require hands-on assessment (palpation, auscultation, neurological checks) that current AI cannot perform autonomously. The task fundamentally requires physical manipulation and real-time clinical judgment in person.
Task automatabilityclaude-sonnet-52/5Physical examinations require hands-on assessment (palpation, auscultation, physical maneuvers) that current AI cannot perform; AI can support documentation and pre-visit intake but not the physical exam itself.,
Adoption barriersclaude-haiku-4-5-202510015/5Physical examinations by nurse practitioners are subject to strict licensure and scope-of-practice regulations; only licensed healthcare professionals can legally perform or sign off on patient examinations. Patient contact, direct observation, and clinical accountability are legally mandated and cannot be delegated to AI.
Adoption barriersclaude-sonnet-55/5Physical exams legally must be performed by a licensed practitioner with direct patient contact; regulatory, licensure, and liability requirements make this a hard barrier to automation.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI systems for routine examination support (documentation, vitals integration) are relatively expensive to implement and maintain, while they reduce only administrative burden, not the high-wage clinical examination itself. The cost advantage is minimal because the human clinician remains essential.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot perform the physical component, there is no substitutive cost comparison—any AI use is an add-on cost atop the human provider's exam.
Technical feasibility todayclaude-haiku-4-5-202510012/5Deployed products exist for isolated subtasks (ECG interpretation, image analysis of skin lesions) but no end-to-end system reliably performs comprehensive physical examinations. Clinical decision support tools aid but do not replace the hands-on examination component that defines the core task.
Technical feasibility todayclaude-sonnet-51/5No deployed product performs the actual hands-on physical examination; AI tools are limited to charting, symptom-checkers, or ambient documentation adjacent to the visit, not the exam itself.

Detect and respond to adverse drug reactions, with special attention to vulnerable populations such as infants, children, pregnant and lactating women, or older adults.

14

CI 7–20 · exposure 17 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare adoption of AI remains slower than tech or finance sectors due to regulatory, liability, and patient-safety constraints. While drug interaction checkers are common, autonomous AI-driven adverse reaction detection and response in vulnerable populations is not yet a production reality in most health systems; adoption remains at pilot and exploratory stages.
Sector adoption velocityclaude-sonnet-52/5Healthcare is a highly regulated, human-contact-intensive sector with slower AI adoption for direct clinical decision-making tasks compared to administrative or information-sector tasks.
Augmentation potentialclaude-haiku-4-5-202510014/5AI can meaningfully assist nurse practitioners by automating literature review on drug interactions, flagging signals in patient data, and alerting clinicians to high-risk drug combinations or patient populations—substantially raising speed and comprehensiveness of detection work while the clinician retains judgment on clinical significance and response. This strong augmentation support does not yet enable full replacement.
Augmentation potentialclaude-sonnet-54/5AI-based clinical decision support and drug interaction checkers can meaningfully assist nurse practitioners in identifying risks and flagging potential adverse reactions, improving vigilance and speed of recognition.
Task automatabilityclaude-haiku-4-5-202510012/5AI can assist in flagging potential drug interactions and adverse reactions through pattern matching against knowledge bases, but detecting and responding to adverse reactions in vulnerable populations requires real-time clinical assessment, contextual judgment, and direct patient observation that current AI cannot reliably replace end-to-end. The task involves synthesizing patient history, physical findings, and population-specific pharmacokinetics—areas where AI lacks the safety margin needed for autonomous decision-making.
Task automatabilityclaude-sonnet-51/5This requires real-time clinical assessment, physical examination, and rapid decision-making at the bedside for vulnerable patients, which current AI cannot perform end-to-end without a licensed clinician.
Adoption barriersclaude-haiku-4-5-202510015/5Nurse practitioners are licensed clinicians with legal accountability for patient safety and drug administration decisions. Regulatory frameworks (FDA, state nursing boards) require a human clinician to assess and respond to adverse reactions, particularly in vulnerable populations; AI cannot legally substitute for this judgment without explicit clinical supervision and sign-off, creating a hard legal and professional barrier.
Adoption barriersclaude-sonnet-55/5Detecting and responding to adverse drug reactions is a core clinical function requiring licensure, scope-of-practice authority, and legal accountability, especially for vulnerable populations.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI inference for drug interaction screening is inexpensive, but the task requires integration with patient monitoring systems, clinician oversight, and liability coverage for adverse outcomes. The all-in cost of an AI system that must be supervised by a nurse practitioner to remain safe is unlikely to be substantially cheaper than the clinician performing the task directly, especially given malpractice and error-cost asymmetry.
Cost vs. human wageclaude-sonnet-51/5The task requires licensed human judgment, physical assessment, and legal accountability, so AI cannot substitute for the human cost regardless of inference expense.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI systems exist for drug interaction checking and signal detection in pharmacovigilance databases, no deployed AI product reliably performs the full task of detecting and clinically responding to adverse reactions in vulnerable populations without human oversight. Current tools are narrow (e.g., interaction checkers) and require clinician interpretation; they do not constitute autonomous feasible solutions for this safety-critical task.
Technical feasibility todayclaude-sonnet-52/5Clinical decision support tools can flag drug interactions and contraindications, but no deployed product independently detects and responds to evolving adverse reactions in vulnerable patients in production.

Supervise or coordinate patient care or support staff activities.

11

CI 3–20 · exposure 13 · augmentation 50 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare sectors adopt AI assistively (decision support, documentation aids) but supervisory roles remain staffed by licensed humans due to regulatory and liability constraints. Actual displacement of supervisory functions is minimal; adoption remains at the pilot or support-tool stage rather than production replacement.
Sector adoption velocityclaude-sonnet-52/5Healthcare adoption of AI for administrative coordination is growing (e.g., scheduling tools) but direct AI supervision of staff/patient care is essentially absent in production settings.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist with real-time dashboards showing patient acuity, staff workload alerts, and suggested task prioritization, which could help a nurse practitioner supervisor make faster decisions. However, the augmentation is moderate—core judgment and accountability remain squarely with the human supervisor.
Augmentation potentialclaude-sonnet-53/5AI tools can help with scheduling, documentation, and flagging care coordination issues, offering moderate support without replacing the supervisory judgment role.
Task automatabilityclaude-haiku-4-5-202510012/5This task requires real-time judgment about patient conditions, staff workload, and dynamic prioritization across multiple individuals. While AI can assist with scheduling and flagging high-risk patients, the supervisory coordination function—adapting to clinical changes, managing interpersonal dynamics, and making care decisions—requires human presence and accountability that current AI systems cannot fully replace.
Task automatabilityclaude-sonnet-51/5Supervising staff and coordinating patient care requires real-time interpersonal leadership, accountability, and situational judgment that current AI cannot execute end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Healthcare licensing and scope-of-practice laws require a licensed clinician to supervise patient care and clinical staff. Legal liability, patient safety standards, and regulatory oversight (state licensing boards, accreditation bodies) create hard barriers: AI cannot sign off on or legally assume responsibility for care coordination or staff supervision.
Adoption barriersclaude-sonnet-55/5Supervision of clinical staff and patient care coordination is tied to licensure, legal accountability, and scope-of-practice regulations requiring a credentialed human in charge.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI tools for scheduling and flagging are relatively inexpensive, but the loaded cost of a nurse practitioner supervisor far exceeds the savings from partial automation. Any AI system must still run alongside human supervision for clinical and legal accountability, negating cost-per-task advantage.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute performing this supervisory role, so no meaningful cost comparison favors AI; human oversight remains necessary and thus costlier to replace.
Technical feasibility todayclaude-haiku-4-5-202510012/5Limited products reliably handle autonomous supervision of patient care or staff coordination at scale. EHR systems and scheduling tools offer support, but they do not independently supervise care or coordinate staff activities; they require human review and decision-making throughout. No production system performs end-to-end supervisory coordination without substantial human oversight.
Technical feasibility todayclaude-sonnet-51/5No deployed product manages or supervises clinical support staff or coordinates live patient care operations; this remains a human management function.

Diagnose or treat complex, unstable, comorbid, episodic, or emergency conditions in collaboration with other health care providers as necessary.

8

CI 0–16 · exposure 13 · augmentation 63 · importance 4.8/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare remains one of the slowest-adopting sectors for clinical automation due to regulatory scrutiny, malpractice risk, patient safety mandates, and the requirement for human sign-off; deployment of AI in emergency/complex diagnosis is nascent and heavily constrained.
Sector adoption velocityclaude-sonnet-52/5Healthcare broadly adopts AI slowly for clinical decision-making in acute/complex care due to regulatory and safety constraints, though documentation tools spread faster.
Augmentation potentialclaude-haiku-4-5-202510013/5AI-powered diagnostic decision support, differential generation, and EHR integration can meaningfully assist NPs in complex cases, raising their speed and confidence; however, augmentation is limited to parts of the workflow (e.g., evidence synthesis, checklist generation) rather than transforming the full task while the NP remains in control.
Augmentation potentialclaude-sonnet-54/5AI clinical decision support, diagnostic aids, and rapid literature/guideline retrieval can meaningfully speed differential diagnosis and treatment planning while the NP retains responsibility.
Task automatabilityclaude-haiku-4-5-202510012/5While AI systems can assist with diagnostic support (symptom checking, differential diagnosis suggestions), the task explicitly requires collaboration with other healthcare providers and managing complex, unstable conditions—requiring real-time clinical judgment, physical examination, and accountability that current AI cannot fully handle end-to-end with 50% time savings at equal quality.
Task automatabilityclaude-sonnet-51/5Diagnosing and treating unstable, comorbid, or emergency conditions requires real-time physical assessment, procedural intervention, and dynamic clinical judgment that current AI cannot execute end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Hard legal barriers: nursing practice acts and malpractice law require a licensed NP to diagnose and treat; liability for misdiagnosis in emergency/unstable conditions falls squarely on the provider, creating non-delegable accountability that prevents autonomous AI substitution.
Adoption barriersclaude-sonnet-55/5Licensure, scope-of-practice law, and malpractice liability require a credentialed provider to diagnose and treat complex or emergency conditions.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of AI infrastructure (models, integration, real-time monitoring, liability insurance, human oversight) combined with the requirement for clinician validation and malpractice exposure makes autonomous AI more expensive than the human NP performing the task directly.
Cost vs. human wageclaude-sonnet-51/5AI cannot substitute for the clinical labor involved, so any deployment adds cost on top of the required NP oversight rather than replacing it.
Technical feasibility todayclaude-haiku-4-5-202510012/5Clinical decision support tools exist in production (e.g., diagnostic aids in EHRs), but no deployed AI system reliably diagnoses and treats complex, unstable, comorbid, or emergency conditions independently; products remain narrow, error-prone, and require physician oversight rather than functioning as autonomous performers.
Technical feasibility todayclaude-sonnet-51/5No deployed product autonomously diagnoses and treats complex or emergency cases; AI decision-support tools exist only as adjuncts to clinician judgment in narrow contexts.

Advocate for accessible health care that minimizes environmental health risks.

8

CI 5–11 · exposure 0 · augmentation 50 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare organizations have not adopted AI to perform advocacy functions; advocacy remains a distinctly human professional responsibility tied to institutional leadership and accountability, with no measurable displacement by AI systems.
Sector adoption velocityclaude-sonnet-52/5Healthcare policy and advocacy work sees minimal AI agent deployment; adoption in this specific advocacy niche is essentially nonexistent despite broader clinical AI adoption trends.
Augmentation potentialclaude-haiku-4-5-202510013/5AI can usefully assist advocates by synthesizing health equity research, modeling environmental health impacts, or drafting policy briefs, meaningfully supporting a nurse practitioner's advocacy work without replacing their core persuasion and leadership role.
Augmentation potentialclaude-sonnet-53/5AI can help draft position statements, summarize environmental health research, and prepare talking points, meaningfully aiding preparation even though the advocacy itself remains human-driven.
Task automatabilityclaude-haiku-4-5-202510011/5Advocacy requires persuasion, negotiation, and influence over policy and stakeholder behavior—activities that demand human authority, credibility, and relational trust. Current AI cannot independently advocate or effect systemic change in healthcare access and environmental policy.
Task automatabilityclaude-sonnet-51/5Advocacy involves persuasion, stakeholder relationships, policy engagement, and value-based judgment that AI cannot execute end-to-end; no off-the-shelf system performs this autonomously.
Adoption barriersclaude-haiku-4-5-202510014/5Advocacy on healthcare policy and environmental health involves regulatory influence, community trust, and professional credibility tied to a licensed human practitioner. Legal and ethical frameworks require a human professional to represent institutional or patient positions.
Adoption barriersclaude-sonnet-54/5Advocacy tied to a licensed clinician's professional credibility, community trust, and policy standing creates strong non-regulatory but substantive barriers to substitution.
Cost vs. human wageclaude-haiku-4-5-202510012/5AI can assist with background research and document drafting at low cost, but the core advocacy function—representation before stakeholders, regulators, or communities—must remain human and cannot be replaced, making true cost substitution infeasible.
Cost vs. human wageclaude-sonnet-51/5There is no AI substitute delivering equivalent advocacy outcomes, so cost comparison favors the human entirely.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed AI product performs advocacy work autonomously; advocacy is inherently a human communicative and political act. AI can support research or draft materials, but cannot serve as the advocate itself in any production healthcare system.
Technical feasibility todayclaude-sonnet-51/5No deployed products perform health policy or environmental health advocacy on behalf of clinicians; this remains a human relational and civic activity.

Prescribe medications based on efficacy, safety, and cost as legally authorized.

5

CI 3–7 · exposure 5 · augmentation 75 · importance 4.7/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510012/5Healthcare remains a heavily regulated, risk-averse sector with slow AI adoption in clinical workflows; prescribing decisions are among the most cautiously automated tasks, with adoption limited to narrow decision-support roles rather than autonomous prescription.
Sector adoption velocityclaude-sonnet-52/5Healthcare adoption of AI for clinical support is growing but remains cautious and pilot-heavy due to regulatory, safety, and liability concerns around prescribing specifically.
Augmentation potentialclaude-haiku-4-5-202510014/5AI-powered clinical decision support tools (drug interaction checkers, evidence-based treatment suggestions, cost-efficacy filtering) substantially enhance prescriber productivity and safety without removing human authority, representing meaningful augmentation in real practice.
Augmentation potentialclaude-sonnet-54/5AI tools can meaningfully assist by flagging drug interactions, suggesting cost-effective alternatives, and summarizing efficacy data, improving the practitioner's decision-making speed and accuracy.
Task automatabilityclaude-haiku-4-5-202510011/5Prescribing medications requires nuanced clinical judgment integrating patient history, comorbidities, drug interactions, and individualized risk-benefit assessment that current AI cannot reliably perform end-to-end. No AI system today can legally or safely assume full prescribing authority without human clinician oversight.
Task automatabilityclaude-sonnet-51/5Prescribing requires legal authorization, clinical judgment integrating patient history, contraindications, and liability that current AI cannot autonomously assume end-to-end.'
Adoption barriersclaude-haiku-4-5-202510015/5Prescribing is legally restricted to licensed healthcare providers (nurse practitioners, physicians, etc.) in all jurisdictions; regulations explicitly require human clinical judgment and legal accountability, creating insurmountable barriers to full automation.
Adoption barriersclaude-sonnet-55/5Prescribing is tightly regulated; only licensed practitioners with prescriptive authority can legally prescribe, creating a hard legal barrier to substitution.
Cost vs. human wageclaude-haiku-4-5-202510011/5The infrastructure, validation, liability management, and required human oversight for AI prescription assistance cost significantly more than the time savings from automation, making AI more expensive than the loaded human cost per prescription.
Cost vs. human wageclaude-sonnet-51/5Since AI cannot legally perform this task at all, there is no valid cost comparison—the human prescriber is mandatory.
Technical feasibility todayclaude-haiku-4-5-202510012/5AI can assist with drug interaction checking and cost-efficacy comparisons in deployed systems, but no AI product independently performs the full prescribing decision at scale in production. Clinical decision support tools exist but require human verification and remain far from autonomous prescribing.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently prescribes medications; clinical decision support tools exist only as advisory aids requiring a licensed prescriber's sign-off.

Consult with, or refer patients to, appropriate specialists when conditions exceed the scope of practice or expertise.

4

CI 0–7 · exposure 5 · augmentation 50 · importance 4.5/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare remains heavily regulated with slow AI adoption for autonomous clinical decision-making; referral decisions are not yet being displaced by AI in production settings due to legal, liability, and quality assurance requirements.
Sector adoption velocityclaude-sonnet-52/5Healthcare adoption of AI for clinical decision-making is cautious and heavily regulated, with slow uptake for tasks involving liability and specialist referral decisions.
Augmentation potentialclaude-haiku-4-5-202510013/5AI tools can usefully assist nurse practitioners by suggesting relevant specialists, flagging red-flag symptoms, or surfacing guideline-based referral criteria, but the final clinical judgment and referral decision remains with the human provider.
Augmentation potentialclaude-sonnet-53/5AI can help by flagging red-flag symptoms, suggesting relevant specialists, or summarizing patient history to support the referral decision, but the decision and communication remain human-driven.
Task automatabilityclaude-haiku-4-5-202510011/5This task requires nuanced clinical judgment to assess when a patient's condition exceeds scope of practice and selection of the most appropriate specialist—decisions that depend on complex medical knowledge, patient context, and real-time clinical reasoning that current AI cannot reliably perform end-to-end.
Task automatabilityclaude-sonnet-51/5This requires clinical judgment about scope-of-practice limits, professional relationship management, and accountability that current AI cannot autonomously execute or be trusted to perform end-to-end.
Adoption barriersclaude-haiku-4-5-202510015/5Scope-of-practice determinations and specialist referrals are core elements of licensed clinical practice; regulatory frameworks (state licensing boards, malpractice liability) and organizational protocols require a licensed healthcare provider to make and sign off on these decisions.
Adoption barriersclaude-sonnet-55/5Referral and consultation decisions are core clinical judgments tied to licensure, liability, and scope-of-practice regulations, requiring a legally accountable practitioner.
Cost vs. human wageclaude-haiku-4-5-202510011/5The cost of building, validating, and maintaining an AI system for specialist referral decisions—including necessary clinical oversight and liability mitigation—exceeds the cost of a nurse practitioner making these decisions, especially given the high stakes of incorrect referrals.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI-only substitute performing this task, so cost comparison favors the human who bears legal and clinical responsibility for the decision.
Technical feasibility todayclaude-haiku-4-5-202510012/5While AI can assist in identifying potential specialist candidates based on diagnosis codes or symptoms, no deployed system reliably makes the clinical judgment to refer or selects the appropriate specialist with sufficient accuracy for independent use in production healthcare settings.
Technical feasibility todayclaude-sonnet-51/5No deployed product independently makes referral decisions or initiates specialist consultations as a substitute for the NP's judgment; this remains a clinical decision-making act performed by licensed professionals.

Perform primary care procedures such as suturing, splinting, administering immunizations, taking cultures, and debriding wounds.

0

CI 0–0 · exposure 0 · augmentation 25 · importance 4.2/5 · click for rater detail

Panel ratings per dimension for this task
DimensionModelRatingRationale
Sector adoption velocityclaude-haiku-4-5-202510011/5Healthcare remains a laggard sector for autonomous task substitution due to regulatory, liability, and human-contact requirements. Adoption of AI for performing primary care procedures is negligible, with organizations relying on trained clinicians.
Sector adoption velocityclaude-sonnet-51/5Physical, hands-on medical procedures show essentially no AI/robotic displacement in current healthcare delivery despite adoption of AI in documentation and diagnostics elsewhere in the field.
Augmentation potentialclaude-haiku-4-5-202510012/5AI offers minimal assistance during the execution of these procedures themselves. Pre-procedure support (e.g., diagnostic imaging analysis) and post-procedure documentation might receive modest AI support, but the core clinical acts remain largely manual and unaugmented.
Augmentation potentialclaude-sonnet-52/5AI offers minimal direct assistance during the physical act of suturing or debriding, though some digital tools (wound imaging analysis, documentation aids) provide marginal peripheral support.
Task automatabilityclaude-haiku-4-5-202510011/5These are hands-on clinical procedures requiring real-time physical manipulation, sterile technique, and direct patient contact. Current AI systems cannot perform suturing, splinting, wound debridement, or other manual interventions; the task is fundamentally incompatible with automation today.
Task automatabilityclaude-sonnet-51/5This is hands-on physical procedural work (suturing, splinting, wound debridement, immunizations) requiring fine motor manipulation of a patient's body; no current AI system can perform these physical interventions.
Adoption barriersclaude-haiku-4-5-202510015/5These procedures are strictly regulated; they must be performed by or directly supervised by a licensed healthcare provider. Legal and liability requirements create hard barriers to substitution—no AI system can legally perform or independently sign off on these clinical interventions.
Adoption barriersclaude-sonnet-55/5These are licensed clinical procedures requiring hands-on scope-of-practice authorization, direct liability for patient harm, and physical presence, making substitution legally and practically barred.
Cost vs. human wageclaude-haiku-4-5-202510011/5AI has no cost advantage when the task cannot be performed at all. The human nurse practitioner's labor remains the only viable option, making any comparison moot.
Cost vs. human wageclaude-sonnet-51/5There is no viable AI substitute performing these physical procedures, so any comparison favors the human provider entirely; robotic alternatives, where they exist, are far more costly than clinician time.
Technical feasibility todayclaude-haiku-4-5-202510011/5No deployed product performs these clinical procedures end-to-end. While AI assists with image analysis or protocol guidance, no autonomous system reliably executes suturing, immunization administration, or wound care in production environments.
Technical feasibility todayclaude-sonnet-51/5No deployed AI or robotic product autonomously performs suturing, splinting, or wound care in clinical practice; surgical robots exist but are human-operated tools, not autonomous performers of these tasks.

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.