Occupational Health and Safety Specialists

19-5011.00

Median wage $90,150/yr140,610 employed (US)Rank #520 of 923 scored · top 56% by substitution

Review, evaluate, and analyze work environments and design programs and procedures to control, eliminate, and prevent disease or injury caused by chemical, physical, and biological agents or ergonomic factors. May conduct inspections and enforce adherence to laws and regulations governing the health and safety of individuals. May be employed in the public or private sector.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution26

Exposure26

Augmentation57

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

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

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

Tasks on the substitution scale

22 rated tasks, binned by substitution score.

Position among all scored occupations

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

Tasks with substitution ≥ 70

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%28

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

Technical feasibility todayw 20%24

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

Cost vs. human wagew 15%26

panel mean rating 2.1/5 → substitution pressure 26/100

Adoption barriersw 20%inverted — strong barriers lower the score28

panel mean rating 3.9/5 (barrier strength) → substitution pressure 28/100

Sector adoption velocityw 10%23

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

Task breakdown (22 tasks)

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

Analyze incident data to identify trends in injuries, illnesses, accidents, or other hazards.

CI 52–76 · exposure 62 · augmentation 88 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Mid-stage adoption: larger manufacturing and healthcare organizations actively deploy incident analytics platforms, but adoption is uneven; smaller and non-digitized firms lag, and AI-driven incident trend analysis is still in early-to-middling production penetration rather than ubiquitous deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety functions are often embedded in manufacturing, construction, and industrial sectors with lower digitization and slower AI tool adoption compared to finance or professional services.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI dramatically augments specialist productivity by rapidly surfacing trends, correlations, and anomalies in large incident datasets that would take humans days or weeks to identify manually, while the specialist retains judgment on causation, remediation priority, and organizational context.
Augmentation potential	claude-sonnet-5	4/5	AI-powered analytics can significantly speed up pattern detection across large incident datasets, helping specialists focus attention on high-risk trends while they retain interpretive and corrective-action responsibilities.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI systems can automate most of the incident data analysis pipeline—extracting patterns, identifying trends, and flagging anomalies in structured and semi-structured incident reports at speeds far exceeding manual review, likely achieving >50% time savings while maintaining quality. However, some domain judgment about contextual relevance and causal interpretation may still require human validation.
Task automatability	claude-sonnet-5	3/5	AI can process structured incident data, identify statistical trends, and generate summaries, but requires human framing of hazard categories, causal interpretation, and validation against real-world context specific to the workplace.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Light barriers exist: some organizations require human sign-off on safety conclusions, and occupational health roles carry modest regulatory oversight, but the analysis task itself is not legally restricted to licensed professionals and can be performed and deployed with minimal friction.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement mandates human-only trend analysis, though liability for missed hazard patterns and regulatory reporting obligations (e.g., OSHA) create moderate incentive for human review and sign-off.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI-powered analytics inference and integration costs are orders of magnitude lower than the loaded salary of a full-time occupational health specialist performing manual incident analysis and report generation.
Cost vs. human wage	claude-sonnet-5	3/5	Analytics software plus AI features reduce time spent on data crunching, but data cleaning, integration with incident reporting systems, and specialist oversight keep costs from being dramatically lower than a human analyst's marginal cost.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products (analytics platforms, ML-powered incident management systems, and data visualization tools) reliably perform trend detection and anomaly flagging in production at scale across manufacturing and healthcare sectors. Minor limitations exist in handling highly unstructured narrative data or novel hazard types.
Technical feasibility today	claude-sonnet-5	3/5	BI/analytics tools with AI features (e.g., anomaly detection, trend dashboards) are deployed in EHS software today, but full autonomous trend interpretation with domain-appropriate judgment still requires specialist review.

Maintain inventories of hazardous materials or hazardous wastes, using waste tracking systems to ensure that materials are handled properly.

CI 48–79 · exposure 70 · augmentation 75 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Manufacturing, chemical, healthcare, and facilities management sectors have actively adopted digital waste tracking and automated inventory systems over the past decade, with substantial production deployment in mid-to-large organizations; smaller firms lag but adoption is accelerating.
Sector adoption velocity	claude-sonnet-5	2/5	EHS and industrial safety functions are historically slower to digitize fully, though waste tracking software adoption is increasing among larger regulated firms; sector-wide AI agent adoption remains limited.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI inventory and tracking systems significantly assist specialists by automating data entry, flag anomalies, generate compliance reports, and free time for investigation and remediation of non-standard situations, directly raising their productivity while they remain responsible for oversight and policy decisions.
Augmentation potential	claude-sonnet-5	4/5	AI-enabled inventory and waste tracking systems significantly reduce manual record-keeping burden, flag discrepancies, and improve compliance monitoring, meaningfully boosting specialist productivity.
Task automatability	claude-haiku-4-5-20251001	5/5	Inventory maintenance and waste tracking are well-structured, data-entry and rule-based tasks that lend themselves to full automation; AI systems can monitor stock levels, generate alerts for compliance thresholds, update tracking systems, and produce reports with minimal manual intervention, easily meeting the 50% time-saving threshold at equal quality.
Task automatability	claude-sonnet-5	3/5	Data entry, tracking, and reconciliation of hazardous materials inventories can largely be automated with existing EHS software and database systems, but verification of physical handling and compliance still requires human oversight.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While regulatory compliance frameworks (EPA, DOT, OSHA) mandate proper waste tracking, the actual automation of inventory maintenance is not legally prohibited and requires only supervisory oversight rather than human sign-off on every entry; organizations often add manual verification steps for risk aversion, creating moderate friction.
Adoption barriers	claude-sonnet-5	3/5	Regulatory frameworks (e.g., EPA, OSHA) require accurate documentation and often specify responsible personnel, creating moderate compliance and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI-driven waste tracking systems cost a fraction of a full-time specialist's loaded wage once deployed, with ongoing inference and maintenance costs negligible; the one-time integration and training effort is modest relative to annual salary savings.
Cost vs. human wage	claude-sonnet-5	3/5	Software-based tracking systems reduce labor costs for record-keeping, but ongoing data entry, verification, and compliance oversight still require paid specialist time, keeping costs roughly comparable to partial automation gains.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature ERP and specialized waste management software (e.g., ChemWatch, Veritiv) with built-in tracking and automated alerts are deployed in production across many organizations; however, integration with varied legacy systems and requirement for occasional human verification of physical counts slightly limit the perfect 5 rating.
Technical feasibility today	claude-sonnet-5	3/5	Commercial waste tracking and EHS management systems (e.g., chemical inventory software) are widely deployed and reliably track materials, though they require human data input and periodic audits for accuracy.

Recommend measures to help protect workers from potentially hazardous work methods, processes, or materials.

CI 29–71 · exposure 53 · augmentation 75 · importance 4.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Large corporations and regulated industries are piloting AI-assisted safety recommendation systems, but most organizations still rely on traditional expert review; adoption remains in early-to-middle stage rather than deep production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety is a slower-adopting, compliance-heavy, physically-grounded field with limited AI agent deployment compared to information-sector professions.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments specialist productivity by rapidly compiling relevant hazard data, regulations, and precedent into structured drafts that specialists then refine and contextualize, substantially reducing research and synthesis time while the human expert retains final judgment.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly speed up literature review, hazard identification checklists, and drafting of safety recommendations, meaningfully boosting specialist productivity while the human retains final judgment.
Task automatability	claude-haiku-4-5-20251001	5/5	Current AI systems can rapidly synthesize hazard data, regulatory standards, and best practices to generate protective measures across a wide range of work methods and materials, meeting or exceeding time-saving thresholds for comprehensive safety recommendations at equal quality.
Task automatability	claude-sonnet-5	2/5	AI can draft generic hazard-control recommendations from documented standards, but tailoring to specific worksites, materials, and processes requires site inspection, contextual judgment, and physical assessment that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While recommendations often require human expert judgment and organizational sign-off (and liability may fall on the organization, not the AI), there are no strict legal requirements barring AI from assisting or even drafting recommendations; adoption friction is moderate, centered on trust and oversight.
Adoption barriers	claude-sonnet-5	4/5	OSHA and similar regulatory frameworks often require a qualified professional to certify hazard assessments and control recommendations, creating liability and credentialing barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI inference and integration cost for generating safety recommendations is modest relative to the loaded wage of an occupational health specialist, especially when used for initial draft recommendations that accelerate expert review.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply generate draft recommendations from regulatory text, but the human specialist's site visits, judgment, and liability review remain necessary, keeping blended costs roughly comparable to a human-only approach.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	AI products (generative and specialized safety/regulatory databases) can produce plausible recommendations, but they lack real-time understanding of site-specific conditions and regulatory nuance; deployment remains mostly in advisory/drafting roles with human expert sign-off rather than autonomous decision-making at scale.
Technical feasibility today	claude-sonnet-5	2/5	Some compliance-support and EHS software tools use AI to suggest controls based on hazard databases, but no deployed product reliably performs full hazard-specific recommendation generation without expert review at scale.

Write reports.

CI 34–60 · exposure 53 · augmentation 88 · importance 3.9/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational health and safety is a highly regulated field with strong compliance cultures and reliance on human expertise; adoption of generative AI for report writing remains early and cautious, with most firms using AI only in pilot or assistive modes rather than autonomous production systems.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational health and safety is a compliance-heavy, often industrial/physical-sector field with slower digitization and AI adoption compared to finance or software.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly boost the productivity of a specialist by auto-generating report sections, organizing evidence, and flagging missing compliance elements, allowing the human to focus on interpretation, judgment, and regulatory sign-off. This assistive role is well-suited to the task's requirements and is already seeing adoption.
Augmentation potential	claude-sonnet-5	5/5	AI is highly effective at drafting, formatting, and summarizing incident data into report language, substantially boosting specialist productivity while they retain responsibility for accuracy and compliance.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can draft significant portions of safety reports by synthesizing inspection data, incident summaries, and compliance checklists, but requires human subject-matter expertise and legal judgment to finalize recommendations and ensure regulatory accuracy. Time savings of 30–50% are plausible with setup, though full end-to-end automation without review falls short of the equal-quality threshold.
Task automatability	claude-sonnet-5	4/5	LLMs can draft structured safety reports from provided data, incident notes, and templates, achieving significant time savings though a human must verify facts and site-specific compliance details.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Occupational health and safety reports must meet OSHA and other regulatory standards, and the specialist's professional judgment and (in many jurisdictions) licensure or legal accountability for recommendations create strong barriers to full automation. Liability asymmetry is high: errors in safety guidance create direct human-harm consequences.
Adoption barriers	claude-sonnet-5	3/5	Reports often require a certified specialist's professional judgment and signature for regulatory submissions, creating moderate liability and authorization friction even if drafting is AI-assisted.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	API costs for AI-assisted drafting are low, but the task requires specialized domain knowledge and legal oversight that limits human wage replacement; full cost ratio remains unfavorable when oversight labor is factored in.
Cost vs. human wage	claude-sonnet-5	4/5	Drafting narrative reports via AI is far cheaper than a specialist's time per page, though integration with inspection data and review overhead reduce the full order-of-magnitude gain.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	LLMs can generate report templates and initial drafts, but deployed products lack reliable domain validation of safety-critical claims and do not yet perform this task at production quality in regulated environments. Pilot tools exist but material error rates and compliance gaps prevent routine unsupervised deployment.
Technical feasibility today	claude-sonnet-5	3/5	General-purpose AI writing tools and some compliance-specific platforms are used to draft reports today, but domain-specific accuracy, formatting standards, and regulatory citations still require human review, limiting reliability at scale.

Provide new-employee health and safety orientations and develop materials for these presentations.

CI 30–48 · exposure 42 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Health and safety functions are slower to digitize and automate than information-intensive roles. Adoption remains largely pilot-stage (using AI for content drafts) with minimal displacement of the specialist role. Organizational conservatism and regulatory caution keep velocity low.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational health and safety functions are typically embedded in manufacturing, construction, and industrial sectors with slower AI adoption patterns compared to information-heavy industries.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can substantially assist by generating draft slides, suggesting relevant content, checking compliance checklists, and personalizing materials for different employee groups. A specialist using these tools can prepare and customize orientations faster while maintaining final control and judgment, significantly raising productivity.
Augmentation potential	claude-sonnet-5	4/5	AI substantially speeds up drafting of orientation slides, handouts, and scripts, letting specialists focus on tailoring content and delivering the in-person or interactive components.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can draft presentation materials and generate standard content templates, but delivering live orientations requires adaptive communication, answering unexpected questions, and assessing comprehension in real-time—tasks that demand human presence and judgment. The end-to-end task cannot achieve 50% time savings at equal quality today.
Task automatability	claude-sonnet-5	3/5	AI can draft orientation materials and slide content effectively, but delivering live orientation and adapting to site-specific hazards still requires human involvement, capping full end-to-end automation at roughly half the task.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory and liability barriers are significant: occupational health and safety orientations are often mandated by OSHA or equivalent bodies, and the employer bears legal responsibility for ensuring comprehension and compliance. A human health and safety specialist typically must oversee and sign off on the training, creating hard barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	OSHA and similar regulations often require documented, competent safety training tailored to the workplace, creating moderate compliance and liability friction even though no license is strictly required to develop materials.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI can reduce content-creation labor, but the specialist's expertise, liability oversight, and live delivery remain necessary. Overall cost ratio is unfavorable because the human time saved on drafting is offset by the specialist's ongoing role in delivery, customization, and legal accountability.
Cost vs. human wage	claude-sonnet-5	3/5	Material development costs drop significantly with AI drafting, but delivery, facilitation, and hazard-specific customization still require paid staff time, keeping overall costs roughly comparable to fully human-run programs.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Products exist for generating training materials and slide decks (GenAI tools, LMS authoring aids), but reliable delivery of live orientations with safety compliance remains primarily human-performed. Narrow automation of content generation is feasible; full task automation is not production-ready.
Technical feasibility today	claude-sonnet-5	3/5	Generative AI tools are routinely used to produce training content and presentations, but no deployed product autonomously runs new-employee safety orientations reliably at scale in real workplaces.

Investigate the adequacy of ventilation, exhaust equipment, lighting, or other conditions that could affect employee health, comfort, or performance.

CI 25–30 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational safety remains a traditionally human-centric, compliance-driven field with slow AI adoption. Most organizations rely on certified safety professionals and external inspectors rather than automated systems, reflecting both regulatory requirements and organizational caution around health and safety delegation.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety is a moderately digitized field with growing sensor and IoT adoption, but this remains a physically-grounded inspection role with slower AI integration than office-based professional services.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist specialists by analyzing sensor data, flagging anomalies in lighting or air quality readings, or organizing inspection documentation, but the human specialist must retain responsibility for judgment calls on adequacy and compliance. Tools for data synthesis offer modest productivity gains while the specialist stays in the loop.
Augmentation potential	claude-sonnet-5	3/5	AI-powered sensors and data analytics tools can help specialists monitor conditions continuously and flag anomalies, improving efficiency while the specialist still performs judgment-based investigation and interpretation.
Task automatability	claude-haiku-4-5-20251001	2/5	Physical inspection of ventilation, exhaust equipment, and lighting requires on-site measurement and sensory assessment that current AI cannot perform autonomously. While AI could analyze data from sensors or photos, the core task demands hands-on investigation of equipment adequacy and environmental conditions that goes beyond what today's systems can do end-to-end.
Task automatability	claude-sonnet-5	2/5	Requires physical on-site inspection, sensor readings, and observational judgment about workplace conditions that current AI cannot perform end-to-end without human presence., though AI can assist with analysis of collected data.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory and liability barriers exist: OSHA and similar bodies require qualified safety professionals to certify workplace conditions, and legal responsibility for workplace health falls on designated human specialists. Employer liability for health failures creates high error-cost asymmetry that favors human accountability.
Adoption barriers	claude-sonnet-5	3/5	OSHA and similar regulations often require qualified professionals to assess and certify workplace conditions, and liability for missed hazards creates disincentive to fully automate, though not always requiring formal licensure.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The labor cost of specialized occupational health and safety specialists (typically $60k–$90k+ annually) remains substantially lower than the combined cost of sensor arrays, imaging equipment, AI analysis, and required human oversight to validate findings on-site.
Cost vs. human wage	claude-sonnet-5	2/5	Sensor deployment and monitoring can reduce some costs, but human inspection, judgment, and physical presence remain necessary, keeping overall costs comparable to or only modestly below human-only approaches.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs comprehensive workplace health investigations. AI tools can assist with photo analysis or data interpretation, but production systems do not independently assess ventilation adequacy, exhaust functionality, or integrated workplace conditions at the standard required for occupational safety compliance.
Technical feasibility today	claude-sonnet-5	2/5	IoT sensors and monitoring products exist for measuring air quality, lighting, and noise, but no deployed AI system autonomously conducts full workplace investigations without human inspectors.

Conduct safety training or education programs and demonstrate the use of safety equipment.

CI 25–30 · exposure 25 · augmentation 63 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While online safety training platforms are increasingly used, actual adoption of AI-driven autonomous training delivery remains limited. Most organizations still rely on human-led instructor training sessions, especially for high-risk equipment, reflecting slower digitization and regulatory conservatism in this domain.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety functions are in physically-oriented industrial sectors with lower digitization and slower AI adoption compared to information-based professions.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist by generating training curricula, creating interactive modules, producing safety videos, and administering quizzes, enabling human trainers to focus on live demonstrations and competency verification. This represents meaningful but bounded productivity gain while the human expert remains essential.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by generating training curricula, quizzes, videos, and multilingual materials, and by tracking compliance, significantly boosting the specialist's productivity while they still deliver hands-on components.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate training materials and deliver video-based safety instruction, demonstrating physical safety equipment use requires hands-on interaction, real-time adjustment to learner questions, and verification of competency that current systems cannot reliably perform end-to-end. This falls well short of the 50% time-saving threshold for a complete training delivery.
Task automatability	claude-sonnet-5	2/5	Content generation for training materials can be automated, but live delivery, hands-on demonstration of equipment, and interactive facilitation require physical presence and cannot be fully automated today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Occupational health and safety training is heavily regulated (OSHA, industry-specific standards) with legal liability for inadequate training, and many jurisdictions require documented certification by qualified human instructors. Liability asymmetry—errors in safety training can cause serious injury or death—creates strong legal and organizational barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	Many safety trainings are regulatorily mandated (e.g., OSHA) and often require qualified trainers or documented certification, creating moderate compliance and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-generated training content can reduce some preparation costs, but the need for human trainers to deliver demonstrations, ensure hands-on competency, and provide personalized feedback means the all-in cost per trainee remains comparable to or higher than traditional human-led training.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply generate slides or training scripts, but the physical demonstration and in-person instruction still require human labor, keeping overall costs comparable to human-led training.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed products can assist with training content creation and online delivery modules, but no mature product reliably conducts live safety equipment demonstrations or adapts to learner confusion in real time. Most production systems still require human trainers to handle the critical demonstration and competency assessment components.
Technical feasibility today	claude-sonnet-5	2/5	AI tools (e-learning platforms, video generation, chatbots) exist to support training content creation, but no deployed product reliably conducts full safety training including physical equipment demonstration.

Develop or maintain hygiene programs, such as noise surveys, continuous atmosphere monitoring, ventilation surveys, or asbestos management plans.

CI 25–25 · exposure 25 · augmentation 50 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption is limited to monitoring automation and data dashboards in larger organizations; autonomous program development and maintenance remain rare. Smaller firms and non-digitized workplaces dominate this field, limiting tech adoption velocity.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety is a compliance-heavy, moderately digitized field with slow AI adoption for physical monitoring tasks, though administrative aspects see some pilot tool use.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI tools can meaningfully assist with continuous data collection, trend analysis, report drafting, and regulatory checklist generation, improving a specialist's productivity in documentation and monitoring review. However, the core tasks of survey design and remediation decisions require human expertise.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with analyzing sensor data trends, drafting compliance documentation, and flagging anomalies, improving productivity while the specialist still performs surveys and makes judgment calls.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data analysis from monitoring systems and report generation, the task requires in-situ surveying, engineering judgment on ventilation design, and regulatory interpretation that demand human expertise and on-site assessment. No current system performs the full cycle end-to-end with 50% time savings at equal quality.
Task automatability	claude-sonnet-5	2/5	This involves physical site surveys, sensor deployment, and program management requiring in-person inspection and judgment that current AI cannot perform end-to-end; AI can only assist with data analysis and documentation portions.mail
Adoption barriers	claude-haiku-4-5-20251001	4/5	Occupational health programs are heavily regulated; many jurisdictions require a licensed occupational health specialist or certified industrial hygienist to develop, sign off on, and take liability for programs. Asbestos management plans typically require professional certification and legal responsibility.
Adoption barriers	claude-sonnet-5	4/5	OSHA and other regulatory frameworks often require certified professionals to sign off on hygiene programs and safety compliance, creating liability and credentialing barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI infrastructure (sensors, platforms, analysis tools) plus human oversight still costs substantially more than hiring specialists directly, and the human remains essential for judgment calls, site-specific planning, and regulatory documentation.
Cost vs. human wage	claude-sonnet-5	2/5	Physical surveys, equipment calibration, and site-specific judgment still require paid specialist labor and equipment costs, so AI only reduces a portion of costs like report drafting rather than the full task cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Products exist for data logging and analysis (e.g., monitoring platforms with dashboards), but end-to-end program development—including survey design, hazard interpretation, compliance decisions, and asbestos management planning—remains primarily human-driven. AI cannot reliably perform the integrated design and decision-making required.
Technical feasibility today	claude-sonnet-5	2/5	Some monitoring hardware includes software analytics, and AI tools can help draft plans, but no deployed product autonomously develops or maintains full industrial hygiene programs in production today.

Inspect or evaluate workplace environments, equipment, or practices to ensure compliance with safety standards and government regulations.

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

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational health remains a compliance-driven, heavily regulated domain where human expertise is entrenched. Adoption of AI for core inspection tasks is slow; most use is still pilot-stage or limited to ancillary reporting support.
Sector adoption velocity	claude-sonnet-5	2/5	Workplace safety inspection is a physical, compliance-heavy field with slow AI adoption; some large industrial firms pilot sensor-based monitoring but widespread production deployment for full inspections is rare.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist specialists by automating hazard documentation, analyzing historical data, generating compliance reports, and flagging anomalies in images—tools already in use at some organizations—but the specialist remains central to judgment and sign-off.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully augment specialists via automated hazard flagging from images/sensors, generating inspection checklists, analyzing regulatory text, and drafting compliance reports, improving efficiency while the human remains responsible for judgment and sign-off.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze documentation and flag known hazards from images/data, physical workplace inspection—assessing equipment condition, worker behavior, environmental factors in real-time—requires human presence and judgment. Current systems cannot perform the full task end-to-end at 50% time savings.
Task automatability	claude-sonnet-5	2/5	Physical inspection of workplaces, equipment, and practices requires on-site sensory observation and judgment that current AI cannot perform end-to-end; AI can assist with checklist generation and report drafting but not the core inspection.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory and liability barriers exist: government regulations often mandate licensed occupational health professionals conduct inspections, and liability for missed hazards creates high error costs. Legal attestation by a qualified human is typically required.
Adoption barriers	claude-sonnet-5	4/5	Many jurisdictions require inspections and compliance certifications to be conducted or attested by qualified/licensed safety professionals, creating legal liability and regulatory barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI can reduce data collection and report generation costs, but the loaded human specialist wage is relatively modest compared to the infrastructure and oversight cost of AI-driven inspection systems with sufficient accuracy for liability.
Cost vs. human wage	claude-sonnet-5	2/5	AI sensor/camera systems can be cheaper for narrow continuous monitoring, but full inspection requiring physical presence, regulatory interpretation, and sign-off still requires costly human labor, keeping overall cost comparable or higher when AI integration and oversight are counted.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Vision models and document analysis tools exist but are narrow; no production system reliably conducts complete workplace compliance inspections. Deployed products handle isolated components (e.g., document review), not the integrated human-led inspection workflow.
Technical feasibility today	claude-sonnet-5	2/5	Some computer vision tools exist for hazard detection from images/video in narrow contexts (e.g., PPE compliance cameras), but no deployed product performs comprehensive workplace safety inspections autonomously and reliably.

Investigate health-related complaints and inspect facilities to ensure that they comply with public health legislation and regulations.

CI 25–25 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Government health and safety agencies, small-to-mid-size workplaces, and regulated industries adopt AI cautiously and slowly for this task. Adoption remains in pilot phase (compliance tracking, document flagging) rather than production displacement; cultural and regulatory inertia limits uptake.
Sector adoption velocity	claude-sonnet-5	2/5	Public health and safety inspection sectors show slow AI adoption due to reliance on physical presence, regulatory certification, and limited digitization of on-site processes.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist by pre-screening complaints, cross-referencing regulations, generating compliance checklists, and organizing facility documentation—tasks that reduce specialist time on paperwork. However, augmentation is limited to preparatory and analytical work; core investigation and judgment remain human-led.
Augmentation potential	claude-sonnet-5	3/5	AI can help specialists by analyzing complaint data, drafting reports, flagging patterns, and suggesting regulatory citations, improving efficiency without replacing the on-site inspection itself.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with document review, regulatory cross-referencing, and initial data analysis, the core task requires on-site facility inspection, professional judgment on hazard severity, and investigation of contextual health complaints that demand human presence and domain expertise. No end-to-end automation achieves 50% time savings today.
Task automatability	claude-sonnet-5	2/5	The physical inspection of facilities and in-person investigation of complaints requires direct sensory observation and judgment that current AI cannot perform end-to-end; AI can only assist with documentation and analysis portions.“},
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory and legal barriers apply: occupational health inspection typically requires licensed or certified professionals; liability for missed hazards or incorrect compliance verdicts falls on the certifying human; and regulations often explicitly mandate professional judgment and on-site verification by qualified inspectors, creating a hard requirement for human authority.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks typically require certified or licensed specialists to conduct official inspections and sign off on compliance findings, creating strong legal and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for document review and compliance checking have modest cost, but the human specialist wage remains the dominant cost driver because the physical inspection, investigation judgment, and decision-making cannot be fully displaced. Integration and oversight overhead adds cost without proportional savings.
Cost vs. human wage	claude-sonnet-5	2/5	Because the physical presence and judgment of a trained specialist is required, AI cannot substitute for the bulk of labor cost, so cost savings are limited to peripheral documentation tasks.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed products exist for compliance document analysis and regulatory tracking, but no production system reliably performs the full investigative and inspection task—physical facility walkthroughs, complaint assessment, and enforcement decisions still require human specialists. Current AI cannot substitute for in-person hazard identification.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously conducts facility inspections or investigates complaints; some AI-assisted checklist and reporting tools exist but the core physical inspection is not automatable.

Maintain or update emergency response plans or procedures.

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

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI for emergency planning remains minimal in production settings. Most organizations still rely on human safety professionals and external consultants; pilot AI-assisted drafting is emerging but not widespread, particularly in regulated industries where fiduciary and legal accountability is high.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety functions in many industries (manufacturing, construction) show slower AI adoption compared to information/finance sectors, with pilots more common than production use.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist safety specialists by auto-generating plan templates, flagging regulatory updates, drafting procedure language, and organizing hazard scenario documents. These augmentations can substantially raise productivity while the specialist retains full control over final content and approval.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly speed up drafting, formatting, and cross-referencing regulatory requirements, meaningfully boosting specialist productivity while they retain final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate draft language, compliance templates, and procedural text, maintaining and updating emergency response plans requires human judgment on organizational context, regulatory interpretation, and scenario-specific hazard assessment. AI can assist significantly but cannot reliably handle the full responsibility of plan validation and sign-off.
Task automatability	claude-sonnet-5	2/5	Drafting and updating text can be assisted by AI, but the task requires site-specific hazard knowledge, regulatory judgment, and physical inspection that current systems cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory frameworks (OSHA, EPA, industry-specific codes) often require that qualified safety professionals develop, certify, or formally approve emergency response plans. Liability exposure for errors makes organizations risk-averse about delegating plan authority to automated systems without human responsibility.
Adoption barriers	claude-sonnet-5	4/5	Emergency plans often require sign-off by certified safety professionals and must meet OSHA/regulatory standards, creating liability and compliance barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-assisted document generation may reduce drafting time, but the specialized domain expertise, regulatory compliance verification, and human sign-off required mean total cost remains high. AI cost savings are offset by mandatory human review and validation labor.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cut drafting time but the specialist still must validate compliance, conduct site assessments, and coordinate stakeholders, keeping overall cost savings modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products currently perform end-to-end emergency plan maintenance in production. Some AI tools draft safety documents and checklists, but none reliably update plans independently while ensuring regulatory compliance and organizational appropriateness without substantial human oversight.
Technical feasibility today	claude-sonnet-5	2/5	Generative AI tools can help draft or revise plan language, but no deployed product reliably maintains emergency response plans without significant expert review and site knowledge.

Coordinate "right-to-know" programs regarding hazardous chemicals or other substances.

CI 25–25 · exposure 25 · augmentation 50 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI in this domain is slow; organizations still rely primarily on human specialists and basic compliance software. Right-to-know program coordination remains a human-centric function in most workplaces, with only incremental tool-use adoption (document management, reporting).
Sector adoption velocity	claude-sonnet-5	2/5	EHS/safety compliance is a sector with slow, cautious tech adoption; most organizations still rely on manual or semi-digitized program coordination with limited AI agent deployment.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist meaningfully by drafting hazard communication materials, organizing chemical databases, flagging regulatory updates, and generating compliance reports. These tools reduce the specialist's workload on routine tasks, though the specialist remains responsible for final review, worker communication, and program oversight.
Augmentation potential	claude-sonnet-5	3/5	AI can help organize SDS databases, draft training materials, and flag compliance gaps, meaningfully assisting the specialist without replacing the coordination role.
Task automatability	claude-haiku-4-5-20251001	2/5	This task requires understanding regulatory requirements, organizational policies, and ongoing communication with multiple stakeholders. While AI could draft informational materials or organize chemical inventory data, the core coordination function—managing stakeholder communication, ensuring compliance, and adapting to workplace-specific hazards—requires sustained human judgment and accountability that current systems cannot reliably handle end-to-end.
Task automatability	claude-sonnet-5	2/5	Coordinating right-to-know programs involves compiling SDS libraries, tracking regulatory compliance, training coordination, and interfacing with employees and regulators—AI can assist with documentation but the coordination, verification, and on-site compliance judgment resist full automation.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory barriers exist: OSHA and EPA rules often require that a qualified human occupational health and safety specialist oversee right-to-know compliance, ensure worker notification, and maintain accountability for hazard communication. Liability for non-compliance or injury rests with the organization and its agents, creating legal friction against full automation.
Adoption barriers	claude-sonnet-5	4/5	OSHA Hazard Communication Standard and similar regulations require designated qualified personnel to manage these programs, and liability for chemical exposure incidents keeps human accountability central.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for document generation or data organization are relatively cheap, but the coordination, compliance verification, and human oversight required mean total cost remains comparable to or exceeds hiring a specialist, especially when liability exposure is factored in.
Cost vs. human wage	claude-sonnet-5	2/5	Software can reduce documentation costs but a human specialist is still needed for site coordination, employee training, and regulatory liaison, so overall cost savings are modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No production systems reliably manage the full coordination of right-to-know programs; existing tools are narrow in scope (inventory management, document generation). Current AI lacks the integrated capability to monitor regulatory changes, update programs dynamically, and coordinate with workers and management as required by law.
Technical feasibility today	claude-sonnet-5	2/5	Some compliance software helps manage SDS inventories and generate labels/training materials, but no deployed AI product autonomously runs a full right-to-know program including audits and employee communication.

Develop or maintain medical monitoring programs for employees.

CI 25–25 · exposure 25 · augmentation 50 · importance 3.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI in occupational health remains slow and limited to data management and reporting aids. The sectors that employ occupational health specialists (manufacturing, chemicals, healthcare) show cautious adoption patterns, with most programs still designed and maintained by human specialists with traditional tools.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational health and safety is a sector with historically slower AI adoption, relying heavily on physical inspections, regulatory compliance, and human judgment, with AI use mostly limited to pilot data-tracking tools.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by analyzing employee health data, identifying patterns in workplace hazard exposures, drafting program components, and automating routine documentation and compliance checks. However, specialists retain decision-making authority over program structure and medical interpretation.
Augmentation potential	claude-sonnet-5	3/5	AI can help track exposure data, generate monitoring schedules, flag anomalies, and draft compliance documentation, meaningfully supporting the specialist without replacing their judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with program design, data analysis, and documentation, the task requires interpreting complex medical and regulatory contexts, making clinical judgments about monitoring thresholds, and coordinating with occupational health providers. No current AI system handles the full end-to-end program development and maintenance with the required depth and accountability.
Task automatability	claude-sonnet-5	2/5	Developing and maintaining medical monitoring programs requires judgment about regulatory compliance, workplace-specific hazards, and clinical protocols that AI cannot fully execute end-to-end today, though AI can assist with documentation and data tracking.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Medical monitoring program development involves regulatory compliance under OSHA and other health standards, interpretation of medical data that may trigger legal obligations, and documented responsibility for program validity. Organizational and regulatory friction strongly protects human specialist sign-off on program design and medical determinations.
Adoption barriers	claude-sonnet-5	4/5	Medical monitoring programs are often mandated by OSHA and other regulations requiring qualified professional oversight (e.g., physician review), creating strong legal and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for health program support remain relatively specialized and require integration with existing occupational health infrastructure, HR systems, and compliance frameworks. The cost of these systems plus required expert oversight remains comparable to or higher than employing specialists.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools can reduce some administrative overhead but the core program design and compliance oversight still require specialist human labor, keeping costs comparable to or only modestly below human-only approaches.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Products exist for occupational health data management and hazard analysis, but they require substantial human oversight to interpret medical findings, ensure regulatory compliance, and customize programs to workplace-specific conditions. No deployed system independently develops or maintains compliant medical monitoring programs at scale.
Technical feasibility today	claude-sonnet-5	2/5	Some software products exist for tracking employee health surveillance data, but no deployed AI product independently designs or maintains full medical monitoring programs reliably in production.

Investigate accidents to identify causes or to determine how such accidents might be prevented in the future.

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

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational health and safety is a regulated, compliance-driven field with strong human accountability norms. While some organizations pilot AI-assisted incident analytics, production adoption of AI-driven investigation remains limited; most sectors continue to rely on human specialists.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety is a specialized, moderately digitized field with slow AI adoption for hands-on investigative work, though software for incident tracking and reporting is common.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by analyzing historical incident data, identifying patterns, suggesting hypotheses, and automating preliminary report generation. However, the creative and judgment-intensive aspects of root-cause investigation (context, witness credibility, site conditions) remain best served by human specialists augmented by analytical tools rather than transformed by them.
Augmentation potential	claude-sonnet-5	3/5	AI can help specialists organize evidence, draft reports, cross-reference safety databases, and identify patterns across past incidents, meaningfully aiding but not replacing the investigative process.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data analysis, report summarization, and pattern identification from accident records, the task fundamentally requires on-site investigation, witness interviews, physical evidence assessment, and contextual judgment that current AI systems cannot perform end-to-end. AI tools lack the embodied investigation capability and the integrative reasoning needed to meet the 50% time-saving bar reliably.
Task automatability	claude-sonnet-5	2/5	Accident investigation requires physical site inspection, interviewing witnesses, evaluating equipment condition, and contextual judgment that AI cannot perform directly; AI can assist with report drafting and pattern analysis but not the core investigative work.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory requirements (OSHA, state/provincial labor laws) typically mandate that licensed or certified safety professionals conduct formal accident investigations, and liability for preventative measures often rests with a human expert. Legal accountability and worker-protection statutes create hard barriers to full automation or substitution.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks (e.g., OSHA) typically require qualified professionals to investigate and certify findings, and liability for incorrect causal determinations creates strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-assisted analysis tools (e.g., for report parsing or clustering similar incidents) are available but require skilled human investigators to direct, interpret, and act on findings. The loaded cost of a qualified occupational health specialist remains substantially lower than the combined cost of AI infrastructure plus mandatory human oversight and verification.
Cost vs. human wage	claude-sonnet-5	2/5	Since AI cannot perform the physical investigation, human labor costs dominate; any AI cost savings apply only to peripheral documentation tasks, not the full task cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products perform full accident investigation reliably. AI can support specific sub-tasks (document analysis, trend detection in past incidents) but production systems do not independently conduct or validate investigations; human specialists remain essential and legally accountable.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously conducts accident investigations; this remains a human-led field activity with AI tools only supporting documentation or data analysis afterward.

Inspect specified areas to ensure the presence of fire prevention equipment, safety equipment, or first-aid supplies.

CI 9–30 · exposure 13 · augmentation 50 · importance 3.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Occupational health and safety is a regulated, conservative sector with strong human-contact requirements and slow digitization in many small-to-mid-size firms. Adoption of autonomous inspection technology remains minimal despite pilot projects, with organizations preferring certified human inspectors for compliance.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety functions are in a moderately digitizing but still physically-grounded sector, with slow adoption of full inspection automation despite growth in safety-management software.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-assisted inspection tools (e.g., computer vision dashboards showing equipment locations, automated checklists, drone feeds for high or hard-to-reach areas) can meaningfully assist human inspectors by organizing data and reducing travel time, though the final judgment and certification remain human responsibilities.
Augmentation potential	claude-sonnet-5	3/5	AI can assist via checklist generation, scheduling, computer-vision-assisted photo review, or IoT sensor alerts that flag missing equipment, improving efficiency while the human still performs the physical inspection.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI could potentially identify missing or misplaced equipment via computer vision in well-defined environments, the task requires physical inspection of 'specified areas' with judgment about equipment condition, accessibility, and compliance. Current systems lack the embodied ability to move through workspaces and perform the tactile, spatial assessment that inspection demands.
Task automatability	claude-sonnet-5	1/5	This requires physical presence to walk through facilities and visually verify equipment is present, functional, and correctly located, which current AI cannot perform end-to-end without robotics.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Safety and fire code compliance inspections are subject to regulatory requirements that typically mandate a qualified human inspector or third-party auditor sign off on findings. Liability for missed equipment in case of incident creates strong incentive to retain human accountability and formal authorization barriers.
Adoption barriers	claude-sonnet-5	3/5	Safety inspections are often tied to compliance frameworks (OSHA, insurance requirements) mandating qualified personnel to verify and attest to conditions, creating moderate regulatory and liability barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	A single AI system capable of autonomous facility inspection would require expensive hardware (mobile robot or drone), integration, and continuous oversight to verify findings. The loaded cost per inspection would likely exceed that of a trained human inspector visiting a facility, especially for smaller organizations.
Cost vs. human wage	claude-sonnet-5	2/5	Without a robotic or sensor-based substitute for physical inspection, AI cannot replace the human cost driver; any AI-assisted checklist software adds marginal cost savings only in documentation, not the core inspection.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Computer vision systems can detect objects in images and video feeds, and robotic platforms exist for facility monitoring, but no mature deployed product reliably performs end-to-end fire/safety equipment inspection at scale in varied occupational settings with acceptable error rates for liability-sensitive domains.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously conducts physical safety equipment inspections; this remains a human physical task, though some IoT sensors track specific equipment status separately.

Collect samples of hazardous materials or arrange for sample collection.

CI 7–25 · exposure 13 · augmentation 38 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational health and safety remains a traditional, compliance-driven sector with moderate digitization. While some larger organizations use scheduling software, the core task of hazmat sampling arrangement has seen slow AI adoption due to regulatory requirements and the critical nature of error prevention.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational health and safety work involves significant field/physical components with historically slower AI adoption compared to purely digital information work.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist with scheduling, routing logistics, compliance checklist generation, and documentation preparation, which would improve the productivity of an occupational health specialist coordinating sample collection across multiple sites or vendors.
Augmentation potential	claude-sonnet-5	2/5	AI can assist with arranging sample collection logistics, generating documentation, or analyzing lab results, but offers little assistance for the physical collection itself.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI systems can assist with scheduling and coordinating sample collection logistics, the physical act of collecting hazardous material samples in potentially dangerous environments remains fundamentally human work requiring real-time judgment, safety protocols, and direct handling. Current AI cannot perform fieldwork or operate in hazardous environments, limiting time savings to scheduling and documentation portions.
Task automatability	claude-sonnet-5	1/5	This is a physical task requiring on-site presence, handling of hazardous materials, and use of specialized sampling equipment; current AI systems cannot perform physical sample collection.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Significant regulatory and liability barriers exist: OSHA, EPA, and other regulations typically require that hazardous material samples be collected by certified or trained personnel, and liability for improper sampling falls on the organization. These requirements create legal and procedural protection for human involvement.
Adoption barriers	claude-sonnet-5	4/5	Handling hazardous materials often requires certified personnel, safety training, chain-of-custody protocols, and regulatory compliance (e.g., OSHA, EPA), creating strong barriers to non-human execution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of deploying AI scheduling and coordination tools is likely comparable to or exceeds the cost of a person arranging sample collection, given the task's relatively straightforward coordination requirements and the low-cost nature of administrative arrangement work by qualified staff.
Cost vs. human wage	claude-sonnet-5	1/5	AI has no capability to substitute for physical sample collection, so cost comparison favors the human doing the task; any AI role would only be in scheduling/arranging, a minor fraction of the task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some AI tools exist for scheduling appointments and coordinating logistics, but no deployed product reliably performs the full task of hazardous material sampling or comprehensive arrangement thereof. The specialized nature of hazmat sample collection and the need for qualified personnel involvement means production systems are limited to ancillary functions.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product physically collects hazardous material samples; this remains entirely a human/robotic-specialist physical task outside current AI product scope.

Perform laboratory analyses or physical inspections of samples to detect disease or to assess purity or cleanliness.

CI 7–25 · exposure 13 · augmentation 50 · importance 3.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	While some sectors (clinical labs, quality assurance) use automated analyzers, adoption of AI-driven autonomous inspection remains limited to pilots and research settings. Most occupational health labs continue to rely on human technicians for sample handling and final assessment, reflecting slow real-world displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational health and safety is a moderately digitized field but physical inspection and lab work remain slow to adopt AI due to hands-on requirements and regulatory oversight.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by automating preliminary image analysis, flagging anomalies, and reducing manual screening workload, enabling technicians to focus on interpretation and edge cases. However, the human remains essential for final judgment, particularly in disease detection and complex sample assessment.
Augmentation potential	claude-sonnet-5	3/5	AI can assist by analyzing lab data, flagging anomalies in imaging or sensor readings, and helping interpret results, but the physical sampling and inspection steps still require human execution.
Task automatability	claude-haiku-4-5-20251001	2/5	Laboratory analyses can be partially automated (e.g., spectroscopy, image classification), but physical inspection of samples requires human judgment, contextual interpretation, and equipment handling that current AI cannot reliably execute end-to-end. The task combines wet-lab work with decision-making that falls short of the 50% time-saving threshold for full automation.
Task automatability	claude-sonnet-5	1/5	This requires physical manipulation of samples, use of lab instruments, and hands-on physical inspection which current AI cannot perform end-to-end without robotics integration far beyond off-the-shelf availability.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Laboratory analyses for disease detection and cleanliness assessment are often regulated by standards (ISO, GLP, clinical lab rules) and may require certified personnel to perform or sign off on results. Liability for false negatives in disease detection creates strong error-cost asymmetry and organizational friction against full automation.
Adoption barriers	claude-sonnet-5	4/5	Health and safety inspections often require certified specialists, regulatory compliance, and legal accountability for findings, creating strong professional and liability barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI tools for laboratory work (automated analyzers, image recognition) still require integration with existing lab infrastructure, human technician oversight, and validation. The total cost-per-analysis remains comparable to or higher than human labor when accounting for equipment, maintenance, and error-checking costs.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot replace the physical handling, instrument operation, and judgment involved, so there is no meaningful AI cost basis to compare against human labor for this physical task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI-assisted image analysis and some automated lab assays exist, but they require human setup, validation, and interpretation. No deployed end-to-end system reliably performs both the physical inspection and the disease/purity assessment autonomously; products remain narrow and require significant human oversight.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously performs physical sample collection and laboratory/physical inspection for disease or purity assessment; this remains a manual, instrument-based human task.

Collect samples of dust, gases, vapors, or other potentially toxic materials for analysis.

CI 5–25 · exposure 13 · augmentation 50 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Occupational health and safety is a heavily regulated, risk-averse sector with strong human certification requirements. Adoption of autonomous sampling remains minimal; organizations continue relying on trained human specialists because regulatory and liability frameworks mandate human accountability.
Sector adoption velocity	claude-sonnet-5	2/5	Industrial hygiene and safety fields adopt digital tools slowly and sampling remains manual; some automated sensors exist but broad AI-driven agentic replacement is rare.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by recommending optimal sampling locations, analyzing historical patterns to suggest when sampling is needed, or automating post-collection data analysis and reporting. However, augmentation is secondary to the primary physical collection task that requires human presence.
Augmentation potential	claude-sonnet-5	3/5	AI-enabled sensors, data logging, and analysis tools can help specialists plan sampling strategies, interpret readings, and manage records, improving efficiency around the core physical task.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with sampling protocols and analysis of results, the physical act of collecting samples in hazardous environments requires human presence, dexterity, and real-time judgment. Current robots cannot reliably navigate complex industrial sites or safely deploy sampling equipment in variable conditions.
Task automatability	claude-sonnet-5	1/5	This is a physical sampling task requiring on-site presence, handling of equipment, and navigating hazardous environments; no AI system can perform physical collection of environmental samples.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Occupational health and safety regulations (OSHA, EPA, ISO standards) typically mandate that samples be collected by certified occupational health and safety specialists or trained technicians who must document chain of custody and legal accountability. Liability for improper sampling and regulatory compliance creates high barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Sampling protocols often require certified specialists following regulatory chain-of-custody and safety procedures, creating strong procedural and legal barriers to full automation, though robotic/sensor assistance is not legally prohibited.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Specialized sampling equipment and trained technicians cost significant amounts; autonomous robotic samplers, where they exist, are expensive capital investments with high integration costs. The per-sample cost via humans remains competitive with available AI/robotic alternatives.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical labor involved, so there is no viable AI cost comparison—human presence is mandatory.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed production system autonomously collects toxic material samples end-to-end. Specialized robotic samplers exist only in narrow research contexts and require extensive human oversight; deployed solutions remain manual or semi-automated with human operators.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs physical sample collection; this remains entirely a human field task, though sensors can assist with monitoring.

Collaborate with engineers or physicians to institute control or remedial measures for hazardous or potentially hazardous conditions or equipment.

CI 7–21 · exposure 13 · augmentation 50 · importance 4.1/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Occupational health and safety remains a heavily regulated, human-intensive domain with slow digital transformation. Adoption of AI is largely in monitoring and reporting rather than decision-making, and organizational friction from compliance requirements and professional licensing is high.
Sector adoption velocity	claude-sonnet-5	2/5	Occupational safety and industrial/physical-plant environments show slower AI adoption than information-sector work, with pilots mostly limited to monitoring/detection tools rather than remediation decisions.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist with hazard literature searches, regulatory requirement synthesis, control-measure databases, and documentation, meaningfully supporting specialist productivity in research and proposal stages. However, collaboration with engineers/physicians and final remedial decisions remain human-driven.
Augmentation potential	claude-sonnet-5	3/5	AI can assist by analyzing hazard data, suggesting control options, or summarizing regulations, but the core collaborative decision-making and implementation remain human-driven.
Task automatability	claude-haiku-4-5-20251001	2/5	This task requires domain expertise, judgment, and interpersonal collaboration with licensed professionals (engineers/physicians) to evaluate complex hazards and design contextual solutions. While AI can assist with hazard identification and literature review, the core collaborative decision-making and remedial design remain firmly in human hands.
Task automatability	claude-sonnet-5	1/5	This is a collaborative, judgment-heavy task requiring physical inspection, cross-disciplinary negotiation, and engineering trade-offs that AI cannot execute end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Liability, regulatory oversight (OSHA, industry standards), and the legal requirement for qualified professionals to sign off on hazard controls create strong barriers. Many jurisdictions require licensed engineers or occupational health professionals to design and approve control measures, preventing pure automation.
Adoption barriers	claude-sonnet-5	4/5	Liability, regulatory compliance (e.g., OSHA), and the need for professional judgment and sign-off by qualified specialists/engineers create strong barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The specialized expertise required (occupational health certifications, engineering or medical knowledge, liability responsibility) means human specialists command high loaded wages; AI assistance is augmentative rather than substitutive, offering no cost advantage over human-led collaboration.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this collaborative remediation work, so cost comparison favors the human specialist entirely.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs end-to-end hazard assessment and remedial design collaboration at production scale. AI tools can support document analysis and suggest generic controls, but cannot independently institute measures or coordinate with licensed professionals as required by the task.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously collaborates with engineers or physicians to design and institute hazard controls; this remains fundamentally a human coordination activity.

Conduct audits at hazardous waste sites or industrial sites or participate in hazardous waste site investigations.

CI 0–14 · exposure 13 · augmentation 50 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Adoption of AI for independent audit/investigation work in hazardous waste is negligible; these sectors are heavily regulated with strong human-certification requirements, low digitization in field work, and organizational resistance to removing the human expert from the critical path.
Sector adoption velocity	claude-sonnet-5	1/5	Environmental health and safety fieldwork is a physical, low-digitization sector with minimal AI agent deployment for on-site hazardous investigations.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist with data organization, historical record analysis, compliance checklist generation, and reporting, helping specialists work faster; however, the core physical assessment and hazard judgment remain human-dependent, limiting augmentation scope.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with pre-audit research, checklist generation, report drafting, data analysis of samples, and compliance documentation, but the core site investigation remains human-led.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data analysis, compliance checking, and preliminary site documentation review, the task requires physical on-site presence, assessment of unpredictable environmental conditions, and real-time judgment of complex hazards that current AI cannot perform end-to-end. Significant manual field work and human decision-making remain essential.
Task automatability	claude-sonnet-5	1/5	This requires physical presence at hazardous sites, sensory inspection, and hands-on evaluation of physical hazards that current AI cannot perform end-to-end without a human on site.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Strong regulatory and legal barriers exist: OSHA, EPA, and environmental liability laws typically require a licensed occupational health and safety specialist or qualified professional to conduct and sign off on hazardous waste site audits and investigations. Liability exposure and certification requirements create hard barriers to full automation.
Adoption barriers	claude-sonnet-5	5/5	Hazardous site audits typically require certified/licensed professionals (e.g., OSHA, EPA compliance), with significant liability, safety, and regulatory requirements mandating human sign-off and physical presence.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	AI tools for supporting audit work (document analysis, data management) remain relatively expensive compared to the modest cost of human specialists doing field work, especially when accounting for liability, oversight, and integration into compliance workflows.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical inspection and legally-required credentialed judgment involved, so there is no meaningful cost substitution today; humans remain the only viable performer.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed AI products reliably conduct independent hazardous waste site audits or investigations at scale. AI tools exist for documentation and analysis support, but production systems handling the full audit and investigation workflow with legal/regulatory accountability do not exist in practice.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product conducts physical hazardous waste site audits or investigations autonomously; this remains a physical, judgment-intensive field task performed by certified specialists.

Order suspension of activities that pose threats to workers' health or safety.

CI 0–0 · exposure 0 · augmentation 38 · importance 4.4/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	This is a discretionary, judgment-heavy, legally-binding human function in a heavily regulated domain; there is no meaningful AI adoption trend for autonomous suspension of work activities.
Sector adoption velocity	claude-sonnet-5	1/5	Safety enforcement in physical workplaces (construction, manufacturing) is a low-digitization domain where AI adoption for authoritative actions is essentially nonexistent.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could assist by analyzing hazard data, flagging risks, or drafting justification reports, but the actual order issuance and responsibility must remain with the qualified specialist.
Augmentation potential	claude-sonnet-5	3/5	AI can help detect hazards via sensors, computer vision, or predictive analytics, giving specialists better information to decide when to order a suspension.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires issuing formal orders to suspend worker activities, which involves legal authority, organizational hierarchy, and accountability that only a human authority figure can exercise. AI cannot issue binding operational orders that carry legal and safety liability.
Task automatability	claude-sonnet-5	1/5	This is an authoritative, judgment-based enforcement action requiring on-site assessment and legal authority; AI cannot issue binding stop-work orders today.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Strong regulatory and organizational barriers exist: occupational health and safety specialists must be qualified personnel with legal authority to issue directives, and liability for incorrect suspension decisions rests with the human decision-maker.
Adoption barriers	claude-sonnet-5	5/5	Issuing suspension orders typically requires designated authority, legal liability, and regulatory backing (e.g., OSHA-type powers), making this a hard, non-delegable human function.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The cost of AI oversight and validation to ensure safe suspension orders would exceed the cost of having the specialist perform this task directly, given the legal and safety-critical nature.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this action, so cost comparison favors the human who retains sole authority and accountability.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI system can unilaterally issue or enforce work suspension orders in production environments; this requires human judgment, legal standing, and organizational authority that AI systems do not possess.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product independently halts workplace activities; at most AI tools flag hazards for human specialists to act on.

Prepare hazardous, radioactive, or mixed waste samples for transportation or storage by treating, compacting, packaging, and labeling them.

CI 0–0 · exposure 0 · augmentation 25 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Hazardous waste handling remains a heavily regulated, human-centric sector with minimal AI or autonomous system adoption. Organizations continue to rely on trained specialists and manual oversight due to safety, liability, and compliance constraints.
Sector adoption velocity	claude-sonnet-5	1/5	Physical hazardous waste handling is a highly regulated, low-digitization, safety-critical sector with minimal AI/robotic adoption for the physical manipulation involved.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could modestly assist through real-time guidance on regulatory compliance, documentation checklists, or monitoring alerts, but current systems offer limited assistance because the core work is physical manipulation and direct judgment of material condition under strict regulatory protocols.
Augmentation potential	claude-sonnet-5	2/5	AI could assist with labeling documentation, tracking inventory, or compliance checklists, but offers little help with the core physical treating, compacting, and packaging steps.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires direct physical manipulation of hazardous and radioactive materials, real-time sensory assessment (detecting leaks, radiation levels, proper sealing), and judgment calls dependent on material properties. Current AI systems cannot reliably handle physical preparation, treatment, or compaction of hazardous waste samples.
Task automatability	claude-sonnet-5	1/5	This is a physical hands-on task involving handling hazardous/radioactive materials, treating, compacting, and packaging them, which requires physical manipulation AI cannot perform today.
Adoption barriers	claude-haiku-4-5-20251001	5/5	This task is governed by strict regulatory frameworks (EPA, NRC, DOT, OSHA) that typically require a licensed or certified human specialist to prepare, inspect, and sign off on hazardous and radioactive waste handling. Legal liability and safety requirements create hard barriers to substitution.
Adoption barriers	claude-sonnet-5	5/5	Strict regulatory frameworks (EPA, NRC, DOT hazmat regulations) require certified/licensed personnel to handle, package, and label hazardous and radioactive waste, creating hard legal barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Specialized robotic or remote-handling equipment for hazardous waste is extremely expensive, requires radiation shielding, containment infrastructure, and integration costs that far exceed the loaded wage of a trained occupational health specialist.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing the physical handling, so any AI cost comparison is moot; humans (often specially trained/certified) remain the only cost-effective option.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI product performs the end-to-end physical preparation and handling of radioactive or hazardous waste. This task is fundamentally outside the scope of software or even current robotic systems in production safety environments.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product physically treats, compacts, or packages hazardous waste; this remains entirely a manual, physical operations task requiring robotics not yet in production for this use case.

Related occupations — Life, Physical & Social Science

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.