Environmental Science and Protection Technicians, Including Health

19-4042.00

Median wage $55,090/yr34,670 employed (US)Rank #464 of 923 scored · top 50% by substitution

Perform laboratory and field tests to monitor the environment and investigate sources of pollution, including those that affect health, under the direction of an environmental scientist, engineer, or other specialist. May collect samples of gases, soil, water, and other materials for testing.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution28

Exposure27

Augmentation58

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

25 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%26

panel mean rating 2.0/5 → substitution pressure 26/100

Cost vs. human wagew 15%30

panel mean rating 2.2/5 → substitution pressure 30/100

Adoption barriersw 20%inverted — strong barriers lower the score32

panel mean rating 3.7/5 (barrier strength) → substitution pressure 32/100

Sector adoption velocityw 10%23

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

Task breakdown (25 tasks)

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

Perform statistical analysis of environmental data.

CI 65–84 · exposure 75 · augmentation 100 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Environmental science and consulting sectors show rapid adoption of automated statistical platforms and AI-assisted analytics; government environmental agencies and private environmental firms routinely deploy such tools in production workflows.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and government/technician-heavy sectors are slower adopters of AI tools compared to finance or software, though data analysis workflows are increasingly digitized.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI statistical tools dramatically augment technician productivity by automating data preparation, generating multiple analytical approaches, and producing exploratory visualizations while the technician handles interpretation, validation, and domain-specific methodology decisions.
Augmentation potential	claude-sonnet-5	5/5	AI substantially speeds up exploratory analysis, code generation, visualization, and interpretation drafting, while the technician retains responsibility for validity and regulatory compliance.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI and statistical tools can automate the majority of statistical analysis workflows—data cleaning, descriptive statistics, hypothesis testing, regression, and visualization—with significant time savings. However, interpretation of results and determining appropriate statistical methods for novel environmental contexts may still require expert judgment.
Task automatability	claude-sonnet-5	4/5	Statistical analysis of structured environmental data (regression, trend analysis, summary stats) is well within current AI/statistical tool capabilities, especially with code-generation assistants handling data cleaning and analysis scripts.
Adoption barriers	claude-haiku-4-5-20251001	2/5	No licensing requirement mandates human performance of statistical analysis, though organizations may prefer human oversight for interpretation and methodology selection. Regulatory requirements typically specify analysis be conducted, not by whom, and there is no liability asymmetry that blocks automation.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement specifically for running statistics, though regulatory reporting may require sign-off by qualified personnel, creating moderate friction rather than a hard barrier.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Software-based statistical analysis is orders of magnitude cheaper per task than a human technician's time once tools are licensed; computational infrastructure and inference costs are negligible compared to loaded labor costs.
Cost vs. human wage	claude-sonnet-5	4/5	AI-assisted statistical analysis tools are inexpensive relative to technician time for routine analyses, though oversight and validation by a qualified analyst still add cost.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature statistical software (R, Python libraries, SPSS, SAS) and AI-assisted analytics platforms are deployed at scale in environmental agencies, consulting firms, and research institutions, reliably performing standard analyses with well-documented error rates.
Technical feasibility today	claude-sonnet-5	3/5	Products like ChatGPT with code interpreter, R/Python-integrated AI assistants, and specialized environmental data platforms can perform statistical analyses, but reliability depends on data quality, domain-specific validation, and human review of methodology and interpretation.

Calculate amount of pollutant in samples or compute air pollution or gas flow in industrial processes, using chemical and mathematical formulas.

CI 51–79 · exposure 67 · augmentation 88 · 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	Environmental and industrial monitoring sectors have digitized extensively and increasingly use automated data analysis pipelines, lab information management systems, and AI-assisted interpretation. Adoption of computational automation in this domain is fairly rapid among regulated labs and industrial operators.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental technician work is a slower-adopting sector with significant physical sampling components and legacy lab equipment, limiting deep AI integration despite the calculation subtask's high automatability.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI significantly assists technicians by automating routine calculations, validating formulas, flagging anomalies, and generating reports, allowing humans to focus on sample preparation, method selection, and interpretation of results. This creates substantial productivity gains while the technician remains in control.
Augmentation potential	claude-sonnet-5	4/5	Software and AI tools significantly speed up and reduce errors in pollutant calculations and flow computations, letting technicians focus more on sampling and interpretation.
Task automatability	claude-haiku-4-5-20251001	4/5	Chemical and mathematical calculations for pollutant quantification and gas flow computation are straightforward algorithmic tasks that AI can perform end-to-end with high accuracy. Given standardized formulas and numerical inputs, current systems can reliably execute these calculations and likely achieve well over 50% time savings compared to manual calculation.
Task automatability	claude-sonnet-5	3/5	The calculation portion (applying formulas to sample data) is easily automatable, but the task also implies sample handling, instrument reading, and validation that require physical/field work not automatable by current AI alone.'
Adoption barriers	claude-haiku-4-5-20251001	3/5	While there are no strict licensing barriers preventing automated calculation, quality assurance and regulatory compliance (e.g., EPA method compliance, chain-of-custody requirements, certified lab protocols) typically require human review and sign-off, creating moderate organizational friction around full automation.
Adoption barriers	claude-sonnet-5	3/5	Environmental compliance reporting often requires certified technician sign-off or accredited lab procedures, creating moderate regulatory and liability barriers even though the math itself is simple to automate.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI computational inference for formula application costs orders of magnitude less than a technician's hourly labor once integrated into laboratory information systems or analysis workflows. The cost per calculation is negligible compared to human wage rates.
Cost vs. human wage	claude-sonnet-5	4/5	Once formulas and data pipelines are set up, computational cost is trivial compared to technician time spent on manual calculations, though initial data capture and integration still require human effort.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature software products and scientific computing libraries (including AI-powered data analysis tools) demonstrably perform chemical calculations, formula-based quantification, and pollutant analysis reliably in production across environmental labs and industrial settings. These capabilities are well-established and widely deployed.
Technical feasibility today	claude-sonnet-5	3/5	Lab information systems and environmental monitoring software already automate concentration and flow calculations, but full integration with automated reading and quality assurance across labs is inconsistent and often still requires technician verification.

Record test data and prepare reports, summaries, or charts that interpret test results.

CI 60–67 · exposure 70 · augmentation 88 · 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	3/5	Environmental labs and agencies show moderate AI adoption: automated data logging and basic report generation are common, but adoption of full AI-driven interpretation remains uneven. Highly regulated sectors adopt more slowly than information-intensive ones.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental and lab-based government/quasi-government sectors tend to adopt digital tools slowly, with many still relying on legacy LIMS systems and manual reporting workflows.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists technicians by auto-generating drafts, flagging outliers, suggesting interpretations, and producing polished charts and summaries. A human can review and validate AI output much faster than creating reports from scratch, substantially raising productivity.
Augmentation potential	claude-sonnet-5	5/5	AI tools significantly speed up drafting summaries, generating charts, and structuring reports from raw data while the technician still reviews and validates results, providing substantial augmentation.
Task automatability	claude-haiku-4-5-20251001	4/5	Current AI systems can automatically extract numerical test data, generate summaries, create charts, and interpret routine results with minimal human setup. The task is largely data-to-document transformation, which LLMs and visualization tools handle well, though complex anomalies or novel findings may require human review.
Task automatability	claude-sonnet-5	4/5	Recording structured test data and generating reports/summaries/charts from that data is a well-defined data-to-text/document task that LLMs and reporting tools can largely automate, though final interpretation may need human validation for accuracy and regulatory compliance.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Regulatory frameworks (EPA, OSHA, ISO standards) often require documented chain of custody and human sign-off on environmental test reports, creating oversight friction. Liability concerns and organizational preference for human verification prevent full substitution, though AI can handle the mechanical work.
Adoption barriers	claude-sonnet-5	3/5	Reports often require technician or supervisor sign-off for regulatory submissions (e.g., EPA compliance), creating a moderate barrier even though the drafting itself isn't restricted to licensed personnel.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI inference, data integration, and light oversight cost substantially less than a technician's loaded wage (likely $45–65k annually). Amortized across many reports, the per-task cost is orders of magnitude lower than human labor.
Cost vs. human wage	claude-sonnet-5	4/5	Automated data entry, chart generation, and templated report writing are cheap to run via software/AI compared to the hourly cost of a technician manually compiling reports.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Mature products (data pipeline tools, LLM-based report generation, charting libraries) perform data recording, summarization, and basic chart generation reliably in production. Environmental labs increasingly use automated data capture and interpretation systems, though full end-to-end deployment varies by organization.
Technical feasibility today	claude-sonnet-5	3/5	Products like AI-assisted lab information management systems (LIMS) and business intelligence tools with generative report features exist, but many environmental technicians still manually compile reports due to variability in test formats and regulatory specificity.

Maintain files, such as hazardous waste databases, chemical usage data, personnel exposure information, or diagrams showing equipment locations.

CI 43–65 · exposure 58 · 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	Environmental and safety-regulated sectors have historically lagged in automation adoption due to compliance caution and organizational risk aversion; while digital tools are standard, AI-driven autonomous file management remains in pilot stages rather than production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental health and safety compliance functions in smaller organizations and industrial/field settings tend to be slower adopters of AI compared to information-sector professional services.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist technicians by automating data extraction from reports, flagging inconsistencies, auto-populating routine fields, and organizing files, substantially raising technician productivity while human oversight remains for compliance-critical decisions.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully speed up data entry, flagging inconsistencies, auto-populating diagrams or logs, and organizing records, significantly aiding technicians while they retain oversight for accuracy and compliance.
Task automatability	claude-haiku-4-5-20251001	3/5	File maintenance involving data entry, organization, and updates can be partially automated through RPA and AI systems (document processing, database updates), but requires human judgment for data validation, compliance verification, and handling exceptions. Roughly half the routine work could be automated with setup.
Task automatability	claude-sonnet-5	4/5	Database/file maintenance, data entry, and updating records are highly structured tasks that current AI and automation tools can handle with substantial time savings, though initial data extraction from varied sources may need human verification.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Hazardous waste databases and personnel exposure records are heavily regulated (EPA, OSHA compliance); legal and liability requirements typically mandate human accountability and sign-off on data accuracy, creating regulatory barriers to full automation.
Adoption barriers	claude-sonnet-5	2/5	Some regulatory recordkeeping requirements exist for hazardous waste and exposure data, but the recordkeeping task itself (as opposed to analysis/certification) isn't restricted to licensed personnel to perform system entry.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI-driven data entry and file organization costs are becoming comparable to technician labor for routine tasks, but human oversight for accuracy and compliance remains essential, keeping the combined cost roughly equal to human-only approaches.
Cost vs. human wage	claude-sonnet-5	4/5	Automated data entry, categorization, and file organization via software is substantially cheaper than dedicated technician time once systems are configured, though setup and validation costs are non-trivial.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Database management and document organization tools exist in production (including AI-assisted data extraction and categorization), but error rates remain material for sensitive compliance records like hazardous waste tracking, and scope is limited to well-structured data sources.
Technical feasibility today	claude-sonnet-5	3/5	Products like database management systems with AI-assisted data entry, OCR, and document management exist and are used in EHS software, but full autonomous maintenance of specialized hazardous waste/exposure databases with regulatory accuracy still requires human oversight.

Distribute permits, closure plans, or cleanup plans.

CI 25–74 · exposure 50 · augmentation 50 · importance 2.9/5 · 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	Environmental agencies and regulated firms move cautiously on automation of permitting and compliance documentation due to legal and liability concerns. Adoption remains in pilot or support-tool phases rather than end-to-end deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental technician roles are in a moderately digitized but often government/regulatory-adjacent sector with slower modernization of back-office workflows compared to finance or tech.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by auto-populating templates, cross-checking data against regulations, and flagging inconsistencies, meaningfully speeding up drafting and review. However, the human expert must retain final judgment and authority over permit terms.
Augmentation potential	claude-sonnet-5	3/5	AI-assisted document management and automated notification systems can meaningfully speed up distribution tasks, though the underlying task is simple enough that gains are moderate rather than transformative.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can draft and format permits or plans, the task involves legal authorization and binding document issuance that requires human signature and legal accountability. Distribution itself is automatable, but the gatekeeping and validation steps limit end-to-end automation to well under 50% time savings.
Task automatability	claude-sonnet-5	4/5	Distributing documents like permits or plans is a routine administrative task that can be handled via automated email/document management systems with minimal human oversight, meeting the time-saving threshold easily.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental permits and closure/cleanup plans are typically issued under regulatory authority and often require a licensed or authorized human to sign off. Legal liability, regulatory coverage (Clean Air Act, RCRA, CERCLA, etc.), and the requirement for professional accountability create significant barriers to full automation.
Adoption barriers	claude-sonnet-5	2/5	Some regulatory recordkeeping and chain-of-custody requirements may apply, but simple distribution of already-approved documents carries low liability and no licensing requirement to perform the distribution itself.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI systems still require substantial human oversight to ensure regulatory compliance and accuracy, making the integrated cost comparable to or only moderately lower than a technician handling the task directly.
Cost vs. human wage	claude-sonnet-5	5/5	Automated distribution via existing software (email, cloud document systems) costs pennies compared to a technician's time spent manually distributing paperwork.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature product reliably generates, validates, and distributes legally compliant environmental permits autonomously. Drafting tools exist, but deployment in production requires human environmental scientists to verify technical correctness and legal sufficiency before issuance.
Technical feasibility today	claude-sonnet-5	4/5	Document management and distribution systems (e-signature platforms, workflow automation, government portals) are widely deployed and reliably handle routine document distribution today.

Provide information or technical or program assistance to government representatives, employers, or the general public on the issues of public health, environmental protection, or workplace safety.

CI 34–34 · 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	Government environmental and occupational safety agencies move slowly on automation due to regulatory accountability requirements and the need for human expertise in enforcement and official guidance. Most sectors using these technicians are not early adopters of AI for core advisory roles.
Sector adoption velocity	claude-sonnet-5	2/5	Government and environmental/public health sectors are historically slow adopters of AI tools relative to information/finance sectors, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can substantially assist technicians by synthesizing regulatory documents, drafting routine responses to common questions, and summarizing scientific literature on health or environmental topics, enabling faster and broader information provision while the technician maintains judgment and accountability.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly help technicians draft explanatory materials, answer common questions, and summarize regulations, meaningfully boosting productivity while humans retain responsibility for accuracy and judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate factual information about public health, environmental, or safety topics, this task requires contextual judgment, understanding of specific local regulations, and responsiveness to varied stakeholder concerns. Providing appropriate technical assistance to government officials or employers demands nuanced interpretation of regulations and customized guidance that current AI systems cannot reliably deliver end-to-end at equal quality.
Task automatability	claude-sonnet-5	2/5	AI can draft informational responses and FAQs, but this task requires site-specific technical judgment, verification of regulatory compliance, and interactive discussion that current systems cannot fully replace end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	3/5	While not formally licensed, public health and environmental protection advice carries liability exposure if incorrect; government agencies and employers typically require human credentials and signoff. This creates meaningful friction and organizational risk that slows substitution.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement for this specific task, but liability concerns around inaccurate environmental/safety guidance and preference for human accountability create moderate friction.
Cost vs. human wage	claude-haiku-4-5-20251001	3/5	AI inference costs for generating information and guidance are modest, but the need for human review, fact-checking, and customization for specific regulatory contexts makes the all-in cost roughly comparable to hiring a technician for straightforward advisory tasks.
Cost vs. human wage	claude-sonnet-5	3/5	AI can cheaply handle routine informational queries, but complex technical assistance still requires human expert review, keeping blended costs roughly comparable to human technicians for substantive work.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI systems can draft informational content and answer routine questions about environmental or safety standards, but deployed products lack the contextual awareness, regulatory specificity, and liability tolerance needed for government representatives or employers to rely on them for official technical assistance without substantial human oversight.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI assistants exist for general public health/environmental Q&A, but no deployed product reliably provides authoritative technical/program assistance across the variety of contexts this task covers.

Discuss test results and analyses with customers.

CI 25–30 · exposure 25 · augmentation 63 · 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	2/5	Environmental and health testing sectors are moderately digitized but remain conservative on customer-facing automation; adoption is limited to internal use of AI-assisted report generation, not substitution of the discussion itself. Real production displacement in this customer-interaction layer is minimal.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and technical field services are a moderately low-digitization sector with slow AI adoption for direct client interactions compared to finance or professional services.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by drafting summaries, flagging key findings, suggesting explanations, and preparing technicians with context before calls, meaningfully improving their efficiency and confidence. However, the human remains essential for the interpersonal and judgment components of the discussion.
Augmentation potential	claude-sonnet-5	4/5	AI can help technicians prepare clear summaries, anticipate customer questions, and draft explanatory materials, meaningfully boosting productivity while the human still leads the actual conversation.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can draft initial summaries of test results, this task fundamentally requires two-way dialogue, understanding customer context, answering dynamic questions, and building trust—capabilities that current AI struggles with at production scale. A technician must explain nuanced findings, calibrate communication to audience, and respond to follow-up questions in real time.
Task automatability	claude-sonnet-5	2/5	Discussing results with customers requires real-time interpersonal communication, answering follow-up questions, and tailoring explanations to customer concerns, which AI cannot fully replace end-to-end today.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Customer-facing discussions carry legal/liability risk if test results are misrepresented; many organizations and regulations expect a qualified human to discuss results affecting health or environmental decisions. Professional expectations and customer preference for human credibility add organizational friction to full automation.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing mandate for this specific conversation, but liability concerns around miscommunicating environmental/health risk findings and customer expectations for human accountability create meaningful friction.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Integration costs (maintaining accurate test data, ensuring regulatory compliance in communication, human oversight for liability) plus the need for human fallback for any substantive customer interaction makes the all-in AI cost comparable to or higher than direct technician labor.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can cheaply draft summaries, the actual customer-facing discussion still requires a human technician's time and judgment, limiting cost savings for the full task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably handle complex, context-dependent customer discussions about technical results at scale. Chatbots can answer basic FAQ-style questions about reports, but cannot substitute for a technician navigating customer concerns, objections, and custom explanations with minimal error.
Technical feasibility today	claude-sonnet-5	2/5	Chatbots and AI assistants can summarize technical data, but no deployed product reliably conducts substantive client discussions about environmental test results in production without human involvement.

Inspect workplaces to ensure the absence of health and safety hazards, such as high noise levels, radiation, or potential lighting hazards.

CI 25–25 · exposure 25 · augmentation 63 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI-driven inspection remains limited and largely in pilot phases within larger industrial organizations. Most workplaces still rely on manual technician inspections; digitalization of occupational safety is slower than IT or finance sectors, constrained by fragmented regulatory requirements and the physical nature of the work.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental health and safety inspection is a physically-grounded, low-digitization field where AI adoption is mostly limited to sensor-assisted monitoring rather than full inspection automation.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist technicians by automating data logging, flagging anomalies in sensor readings, and generating preliminary reports, which can streamline inspection workflows and reduce manual note-taking. However, the core judgment, visual assessment, and regulatory sign-off remain human responsibilities.
Augmentation potential	claude-sonnet-5	4/5	AI-powered sensors, data logging, and analytics tools significantly help technicians detect and document hazards faster, though the human must still verify and act on findings.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze sensor data for noise, radiation, and lighting levels, the task requires physical inspection of complex workplace environments, assessment of contextual hazards, and judgment about compliance with varied regulatory standards. Current AI cannot reliably perform the full end-to-end inspection without substantial human oversight and physical presence.
Task automatability	claude-sonnet-5	2/5	Physical inspection requires being on-site with sensors and human judgment about context-specific hazards; AI cannot yet fully replace the physical walkthrough and situational assessment.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Workplace safety inspections are heavily regulated; compliance documentation often requires certification and sign-off by licensed professionals. Liability for missed hazards and potential injury creates strong legal and organizational barriers to full automation, and many regulatory frameworks explicitly require human inspection and professional judgment.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks (OSHA, EPA) generally require certified professionals to conduct and certify safety inspections, creating strong compliance and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI-assisted inspection tools (sensors, data analysis platforms) require significant capital investment and integration costs, while human technicians remain necessary for interpretation, follow-up investigation, and remediation guidance. The all-in cost is comparable to or exceeds a technician's loaded wage for most applications.
Cost vs. human wage	claude-sonnet-5	2/5	Sensor deployment, calibration, and human oversight for interpreting results still require significant investment comparable to or exceeding technician wages for equivalent coverage.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some specialized tools exist (noise meters with IoT connectivity, radiation detectors with data logging) that provide AI-analyzable measurements, but no integrated deployed product reliably conducts comprehensive workplace safety inspections autonomously. Most systems are narrow, confined to single hazard types, and still require human technicians on-site.
Technical feasibility today	claude-sonnet-5	2/5	Some IoT sensors and AI-based monitoring tools exist for noise/radiation/lighting detection, but no product performs full autonomous workplace inspection and hazard judgment reliably in production today.

Weigh, analyze, or measure collected sample particles, such as lead, coal dust, or rock, to determine concentration of pollutants.

CI 25–25 · exposure 25 · augmentation 50 · 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	Environmental labs are slow to adopt full automation due to regulatory constraints, the need for specialized equipment validation, and limited digitization relative to information sectors. Adoption remains mostly in pilot phases or partial workflow automation.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental testing and lab sciences are a moderately digitized but physically grounded sector with slower uptake of full AI-driven automation compared to information-sector tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-assisted data interpretation and automated calculation of pollutant concentrations from instrument readings can meaningfully speed analysis and reduce transcription errors, but the technician remains essential for sample validation, quality control, and regulatory compliance.
Augmentation potential	claude-sonnet-5	3/5	AI can help analyze concentration data, flag anomalies, and support report generation, providing moderate assistance to technicians without replacing physical sample analysis steps.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in analyzing spectroscopic or chromatographic data from instruments, the hands-on weighing and physical sample handling requires human technicians. Only the analytical data interpretation portion is partially automatable, falling short of the 50% time-saving bar for the full task.
Task automatability	claude-sonnet-5	2/5	The physical sample handling, weighing, and lab measurement steps require manual manipulation and calibrated instruments; AI can assist data analysis but cannot perform the physical assay end-to-end today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong regulatory barriers exist: EPA, OSHA, and ISO protocols mandate that qualified technicians perform or validate sample analysis; chain-of-custody requirements and certification standards require human sign-off on results used in environmental compliance and health decisions.
Adoption barriers	claude-sonnet-5	4/5	Environmental monitoring for pollutants like lead often falls under regulatory chain-of-custody and certification requirements, requiring accredited procedures and often human sign-off for legal/regulatory compliance.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Laboratory instrumentation and AI analysis software, plus required calibration and oversight, represent significant capital and operational costs. The loaded wage for a skilled technician is often lower when amortized across multiple samples, making automation cost-prohibitive.
Cost vs. human wage	claude-sonnet-5	2/5	Specialized lab equipment and calibrated instruments are costly and still require skilled technician oversight, so AI/automation cost savings versus a trained technician are modest, not order-of-magnitude cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI-powered laboratory analysis software exists but requires human operators to prepare samples, position them in instruments, and validate results. No end-to-end deployed system reliably performs the complete weighing, analysis, and measurement workflow without technician intervention.
Technical feasibility today	claude-sonnet-5	2/5	Some automated lab instruments (e.g., gravimetric analyzers, spectrometers) exist and are widely used, but they are automation of measurement instruments rather than AI systems performing the full analytical judgment and quality control task.

Monitor emission control devices to ensure they are operating properly and comply with state and federal regulations.

CI 25–25 · exposure 25 · 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	2/5	Environmental agencies and regulated facilities move slowly on automation due to compliance risk aversion, legacy equipment heterogeneity, and regulatory conservatism; most adoption remains in large industrial facilities with mature IT, not across the sector.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental compliance and industrial monitoring sectors adopt automation slowly due to regulatory conservatism, safety requirements, and reliance on physical infrastructure rather than pure information work.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist technicians by automating data collection, flagging sensor anomalies, and generating compliance reports, improving their productivity and reducing manual record-keeping; however, the core diagnostic and hands-on work remains human-dependent.
Augmentation potential	claude-sonnet-5	3/5	AI-powered sensor analytics and anomaly detection can help technicians flag potential issues faster and prioritize inspections, meaningfully aiding but not replacing the monitoring task.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze sensor data and flag anomalies, monitoring emission control devices requires physical inspection, troubleshooting failed equipment, and real-time intervention that cannot be fully automated today. Current systems can assist with data analysis but cannot replace the hands-on diagnostic and corrective work.
Task automatability	claude-sonnet-5	2/5	Physical monitoring and inspection of emission control devices requires on-site sensor reading, equipment checks, and judgment about anomalies that current AI cannot fully perform end-to-end without human presence and action.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory requirements mandate qualified technicians to inspect and certify compliance with state and federal environmental regulations; liability for false negatives (missed violations) creates strong legal barriers, and equipment manufacturer certifications often require licensed personnel sign-off.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks often require certified technicians to inspect, verify, and report compliance, and liability for environmental violations creates strong incentives to keep humans accountable for sign-off.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI-based monitoring systems require significant infrastructure investment, integration with proprietary equipment, and ongoing human oversight for equipment failures and regulatory sign-off, making total cost roughly comparable to or higher than current technician-based monitoring.
Cost vs. human wage	claude-sonnet-5	2/5	While automated sensor systems reduce some labor costs, they still require deployed hardware, calibration, and human oversight, keeping costs comparable to or only modestly below a technician's wage for the full task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product today reliably performs end-to-end monitoring and compliance verification of emission control systems without human technician oversight. Sensor platforms and data analytics exist but lack the integration, physical access, and real-world reliability needed for independent operation.
Technical feasibility today	claude-sonnet-5	2/5	IoT sensors and analytics dashboards exist for continuous emissions monitoring, but interpreting readings, performing physical inspections, and ensuring regulatory compliance still require human technicians in production settings.

Develop or implement programs for monitoring of environmental pollution or radiation.

CI 25–25 · exposure 25 · augmentation 63 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Environmental agencies and labs are traditionally conservative, budget-constrained, and slow to adopt automation; most still rely on established protocols and human expertise. Production deployment of AI-driven program development is rare; adoption remains in early pilot phases if present at all.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental and public health sectors are generally slower adopters of AI agents compared to information/finance industries, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist technicians by automating routine data processing, generating monitoring reports, flagging anomalies in sensor streams, and recommending protocol adjustments—but the human technician remains essential for program design decisions, regulatory interpretation, and validation of outputs.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist with data analysis, trend detection, regulatory research, and drafting monitoring protocols, significantly boosting technician productivity while humans retain design and field responsibilities.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data analysis and visualization of pollution/radiation monitoring data, the task requires developing programs that integrate hardware sensors, calibration protocols, quality assurance, and field validation—activities requiring domain expertise, stakeholder engagement, and real-world testing that current AI cannot orchestrate end-to-end at 50% time savings.
Task automatability	claude-sonnet-5	2/5	Developing and implementing monitoring programs involves site assessment, sensor placement, regulatory compliance, and field verification that require physical presence and judgment AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental monitoring programs are often regulated by EPA, state agencies, or local ordinances; technicians frequently must be certified or hold specific credentials (e.g., Radiation Safety Officer). Liability for monitoring accuracy, public health consequences of equipment failure, and accountability requirements create strong legal and organizational barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks often require certified professionals to design and sign off on environmental and radiation monitoring programs, and liability for public health/safety errors is high.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI can reduce costs on specific subtasks (data processing, report generation) but cannot replace the wages of technicians who design, validate, and oversee monitoring programs. Oversight and quality control remain labor-intensive, keeping overall cost comparable to or higher than unaugmented human work.
Cost vs. human wage	claude-sonnet-5	2/5	The physical fieldwork, equipment deployment, and calibration components still require paid human labor and specialized equipment, limiting overall cost savings despite AI aiding planning documentation.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products autonomously develop or implement environmental monitoring programs end-to-end. AI tools exist for data analysis and sensor integration (narrow, research-stage), but the full program development cycle—requirements gathering, hardware selection, protocol design, regulatory compliance, staff training, field deployment—remains human-driven.
Technical feasibility today	claude-sonnet-5	2/5	AI tools can assist with data analysis and report drafting but no deployed product autonomously designs and implements physical pollution/radiation monitoring programs in production.

Make recommendations to control or eliminate unsafe conditions at workplaces or public facilities.

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	Workplace safety remains highly regulated and risk-averse; organizations are slow to adopt AI-driven safety decisions without human certification. Most adoption is in narrow, high-volume domains (e.g., factory floor monitoring), not in the end-to-end recommendation function.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental health and safety fields are physically grounded and have low digitization and AI adoption compared to information-sector work; pilots exist but production deployment is rare.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist by flagging hazards from sensor data, summarizing regulatory requirements, or generating initial draft control recommendations for expert review. However, the final judgment and accountability still rest with the human technician, making augmentation meaningful but not transformative.
Augmentation potential	claude-sonnet-5	3/5	AI can help technicians research regulations, draft reports, and analyze collected data faster, meaningfully assisting but not replacing the on-site judgment core to the task.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze data patterns and identify common hazards from records or sensor data, making sound recommendations requires contextual judgment about workplace-specific conditions, feasibility, cost-benefit tradeoffs, and regulatory compliance. Current systems lack the situated understanding and accountability needed to recommend safety interventions end-to-end.
Task automatability	claude-sonnet-5	2/5	Generating generic safety recommendations from text descriptions is feasible, but the task requires on-site inspection, judgment about specific hazards, and contextual assessment that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Safety recommendations often trigger compliance reporting, regulatory liability, and insurance implications. Many jurisdictions require a licensed or certified professional to sign off on workplace safety assessments, and organizations face legal risk if algorithmic recommendations contribute to inadequate controls or worker harm.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks (OSHA, EPA) often require certified professionals to make and sign off on safety determinations, creating liability and licensing barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI hazard-analysis tools have meaningful setup and integration costs, and human oversight of recommendations remains essential for liability and accuracy. The all-in cost of an AI system with necessary human review and validation is comparable to or exceeds the loaded wage of a technician for routine cases.
Cost vs. human wage	claude-sonnet-5	2/5	AI could cheaply draft report language, but the physical inspection, sampling, and liability-bearing judgment still require a human technician, so overall cost savings are limited.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools exist to assist hazard detection (e.g., computer vision for PPE compliance, sensor anomaly detection), but no deployed product reliably makes independent safety recommendations across diverse workplace contexts. Existing solutions are narrow, require significant human validation, and lack the legal accountability expected of formal safety recommendations.
Technical feasibility today	claude-sonnet-5	2/5	Some AI tools assist with hazard checklists or drafting compliance reports, but no deployed product autonomously inspects facilities and issues reliable safety recommendations in production.

Develop testing procedures.

CI 25–25 · exposure 25 · augmentation 63 · importance 3.3/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Environmental science and health technician roles remain in regulated, compliance-heavy sectors where procedure development is tightly controlled; adoption of AI automation in this domain is slow, with most organizations still relying on human expertise and established regulatory templates.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and technician roles are in a sector with low-to-moderate digitization and slow AI adoption for regulated technical documentation compared to information/finance sectors.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist technicians by suggesting procedure templates, identifying relevant standards, and drafting documentation, moderately raising productivity during the development phase. However, the need for domain validation and regulatory compliance limits the depth of augmentation.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by drafting initial procedure outlines, summarizing regulatory requirements, and suggesting standard methodologies, significantly speeding up the human-led development process.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can draft procedure templates and suggest testing methodologies based on existing standards, developing novel testing procedures requires domain expertise, regulatory knowledge, and validation that current systems struggle with end-to-end. Significant human oversight and iteration would be needed, falling short of the 50% time-saving threshold at equal quality.
Task automatability	claude-sonnet-5	2/5	Developing testing procedures requires domain expertise, regulatory knowledge, and site-specific judgment that current AI cannot fully replicate end-to-end, though it can draft templates or suggest protocols based on precedent.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Testing procedures in environmental and health contexts are often subject to regulatory requirements (EPA, OSHA, etc.) and professional standards, and a qualified human expert typically must validate and sign off on procedures before deployment. Liability for flawed procedures creates a strong legal barrier to full automation.
Adoption barriers	claude-sonnet-5	4/5	Testing procedures in environmental and health contexts are often governed by regulatory standards (EPA, OSHA, state agencies) requiring qualified personnel to develop and certify methods, creating substantial compliance and liability barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Integration costs, domain expert oversight, and the need for human validation and refinement mean that AI-assisted procedure development still requires significant human labor, keeping total costs close to or above employing a technician directly.
Cost vs. human wage	claude-sonnet-5	2/5	While AI drafting is cheap, the necessary expert validation, regulatory compliance checks, and liability review largely preserve the cost of skilled technician/scientist labor, so overall savings are modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature production systems reliably develop testing procedures independently; existing products may assist with documentation or suggest frameworks but require substantial human expertise and regulatory validation. Deployed solutions exist only as narrow assistants, not autonomous performers of this task.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously develops validated environmental/health testing procedures; LLMs can assist drafting but outputs require expert review and are not production-ready without human authorship.

Determine amounts and kinds of chemicals to use in destroying harmful organisms or removing impurities from purification systems.

CI 25–25 · exposure 25 · augmentation 63 · 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	Environmental and water treatment sectors are moderately digitized but adoption of autonomous chemical-selection AI is still in pilot phases. Most organizations rely on established protocols and licensed technician judgment rather than AI agents in production.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental and water treatment sectors are historically slow adopters of AI due to regulatory caution, legacy infrastructure, and safety-critical operations.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist technicians by recommending chemical options, dosages, and flagging hazards based on historical data, which raises productivity in research and planning phases. However, the final determination remains a human responsibility, limiting augmentation impact on the core task.
Augmentation potential	claude-sonnet-5	4/5	AI-driven modeling and predictive analytics can meaningfully assist technicians in calculating chemical needs and optimizing treatment processes, improving efficiency while humans retain final decision authority.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help identify chemicals and dosages based on stored protocols and data, the task requires real-time assessment of specific environmental conditions, organism types, and system states that demand domain expertise and judgment. Current AI cannot reliably end-to-end automate this at 50% time savings without substantial human verification.
Task automatability	claude-sonnet-5	2/5	This requires site-specific measurements, chemical calculations tied to real-time conditions, and regulatory compliance judgment; AI can assist calculations but cannot autonomously determine dosing without validated sensor data and human verification.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental regulations, safety standards (EPA, OSHA), and liability for incorrect chemical use create significant legal and compliance barriers. A licensed or credentialed technician typically must verify and approve chemical selections, and some decisions require documented human accountability.
Adoption barriers	claude-sonnet-5	4/5	Water/wastewater treatment and chemical handling are heavily regulated with certification requirements and liability for public health outcomes, creating strong barriers to full automation of dosage decisions.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI inference for chemical recommendation is cheap, but the task requires context-specific data gathering, system inspection, and regulatory documentation that still demand human labor. The all-in cost is comparable to or higher than human technician cost when integration and oversight are included.
Cost vs. human wage	claude-sonnet-5	2/5	Because errors in chemical dosing carry safety and regulatory consequences, human verification remains mandatory, keeping the effective all-in cost of AI-assisted determination close to or above the human-only cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some decision-support tools and databases exist to recommend chemicals, but no deployed product reliably makes autonomous chemical-selection decisions for field conditions without expert human sign-off. Products are narrow in scope and accuracy is not production-grade independent.
Technical feasibility today	claude-sonnet-5	2/5	No deployed AI product independently determines chemical treatment dosages for purification systems in production without human oversight; existing tools are decision-support calculators, not autonomous deciders.

Analyze potential environmental impacts of production process changes, and recommend steps to mitigate negative impacts.

CI 25–25 · exposure 25 · augmentation 63 · importance 3.0/5 · 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	Environmental compliance remains highly regulated and human-dependent. Adoption is slow in practice; most firms use consultants or in-house experts for process change reviews. Digital-first, agent-driven automation has not penetrated this sector meaningfully.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental technician roles are in a moderately digitized but physically-grounded, regulation-heavy sector where AI adoption for core technical judgment tasks remains slow and pilot-stage.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by retrieving regulatory data, summarizing research on similar processes, modeling preliminary emissions scenarios, and drafting mitigation checklists. These augment the technician's productivity without replacing the core judgment and recommendation role.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by synthesizing regulations, prior impact studies, and drafting reports, significantly speeding research and documentation while the technician retains final analytical responsibility.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data analysis and literature review on environmental impacts, the task requires domain judgment about process-specific risks, stakeholder context, and trade-offs that demand human expertise. End-to-end automation with 50% time savings at equal quality is not demonstrated by current systems.
Task automatability	claude-sonnet-5	2/5	This requires site-specific technical judgment, integration of regulatory knowledge, and field data interpretation that current AI cannot reliably perform end-to-end; AI can assist with parts (literature review, drafting) but not the full analytical and recommendation process.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory frameworks (Clean Air Act, Clean Water Act, state environmental codes) often require a qualified professional to sign off on impact assessments and mitigation plans; many jurisdictions mandate licensed environmental scientists or engineers. Legal liability and mandate-driven oversight create strong barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Environmental compliance often requires certified professionals to sign off on assessments, and regulatory frameworks (e.g., EIA requirements) mandate qualified human judgment and accountability.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Environmental impact analysis and mitigation recommendations carry high liability cost if wrong, requiring expert review and sign-off regardless. The all-in cost (AI inference plus required human oversight and validation) remains comparable to or higher than direct human analysis.
Cost vs. human wage	claude-sonnet-5	2/5	Human expertise, site knowledge, and legal accountability make AI a supplement rather than a substitute; cost savings from AI assistance are modest given required expert review and liability exposure.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs the full analysis-to-recommendation cycle for novel production process changes. AI tools can support fragments (emissions modeling, regulatory lookup) but integration into a validated, defensible recommendation system is not mature or production-standard in environmental compliance.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously conducts environmental impact analyses and mitigation recommendations for production changes; existing tools are decision-support aids requiring expert oversight.

Examine and analyze material for presence and concentration of contaminants, such as asbestos, using variety of microscopes.

CI 23–25 · exposure 25 · augmentation 50 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Environmental testing remains heavily concentrated in specialized laboratories and field settings with high regulatory oversight, low digital-first culture, and strong professional credentialing requirements. Adoption of AI automation in this sector has been minimal, with human technicians still the norm across most institutions.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental testing labs are a physical, regulated, moderately-digitized sector with slow AI adoption for core analytical work, though software for data logging is common.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI image-analysis tools can assist by flagging suspect areas, suggesting contaminant types, and reducing manual scanning time, allowing human technicians to focus verification and decision-making. However, the augmentation is moderate because the human must ultimately validate and authorize all findings in a regulated context.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with image pattern recognition, fiber counting suggestions, and report drafting, improving throughput while a certified technician still verifies and performs the analysis.
Task automatability	claude-haiku-4-5-20251001	2/5	While image recognition and microscopy image analysis have advanced significantly, the task requires identifying subtle morphological features and concentration assessment that still depend heavily on expert interpretation and handling of physical samples. Current AI can assist in some detection but cannot reliably replace the end-to-end workflow of sample preparation, microscope operation, and authoritative analysis without substantial human oversight.
Task automatability	claude-sonnet-5	2/5	This requires physical sample handling and hands-on microscopy (e.g., PLM/PCM for asbestos) that current AI cannot perform end-to-end; only the image-analysis/classification sub-step could be assisted.HD placephyseholder
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental and occupational health testing for contaminants like asbestos is heavily regulated by agencies (EPA, OSHA); results often require human certification and professional sign-off for legal liability and insurance purposes. Regulatory frameworks mandate qualified human analysis and documentation, creating substantial legal and compliance barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Asbestos and contaminant testing is often subject to accreditation (e.g., NVLAP, AHERA) requiring qualified human analysts and lab certification, creating strong regulatory and liability barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The infrastructure cost (specialized microscopy equipment, sample preparation, integration with lab workflows) combined with required expert oversight makes AI automation more expensive than employing a trained technician, especially when factoring in liability and quality assurance for regulatory compliance.
Cost vs. human wage	claude-sonnet-5	2/5	Physical sample prep, microscope operation, and certified analyst sign-off still require human labor and calibrated equipment, so AI only marginally reduces cost versus the human-driven lab workflow.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI-based microscopy image analysis products exist in research and early deployment phases, but they lack the regulatory validation and consistent performance required for authoritative contaminant identification in compliance-critical settings. Deployed systems have material limitations in discriminating contaminant types and quantifying concentration with the accuracy demanded by environmental and health standards.
Technical feasibility today	claude-sonnet-5	2/5	Some AI-assisted image analysis tools exist in research settings for fiber counting, but no widely deployed product autonomously performs full contaminant identification and certified reporting in production labs.

Prepare samples or photomicrographs for testing and analysis.

CI 16–30 · exposure 20 · augmentation 50 · 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	Environmental testing labs and field technician roles tend to be dispersed, budget-constrained, and slower to adopt automation; most labs still rely on trained human technicians rather than automated sample preparation systems in routine operations.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental testing labs are moderate adopters of digital tools (LIMS, imaging software) but physical sample prep automation remains rare and sector-wide AI adoption for this specific task is slow.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-assisted image analysis and documentation tools can help technicians review and catalog photomicrographs more efficiently, and automated sample tracking systems can reduce manual logging, offering useful but not transformative productivity gains.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with image analysis of photomicrographs, automated labeling, and data logging, improving throughput even though the physical preparation itself remains manual.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with image analysis and documentation of photomicrographs, the hands-on preparation of physical samples—handling materials, calibrating equipment, ensuring sterility—requires dexterous manipulation and real-time problem-solving that current systems cannot reliably perform end-to-end without significant human oversight.
Task automatability	claude-sonnet-5	2/5	Sample preparation is largely a physical, hands-on lab task involving handling of environmental media (soil, water, air filters) and specimen mounting, which current AI cannot perform without robotics; only the imaging/documentation portion could be assisted by software.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Laboratory and environmental analysis work is often subject to strict regulatory compliance, chain-of-custody requirements, and quality assurance protocols that mandate documented human responsibility and sign-off on sample integrity and handling.
Adoption barriers	claude-sonnet-5	3/5	While not strictly licensed work, chain-of-custody, quality-control protocols, and regulatory requirements (e.g., EPA method compliance) impose procedural rigor that discourages unsupervised automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of setting up robotic sample preparation systems, maintaining imaging hardware, and providing oversight would likely exceed or match the loaded cost of a trained technician performing the task, especially for varied sample types and small-batch work.
Cost vs. human wage	claude-sonnet-5	1/5	Physical sample handling still requires human technicians and lab equipment; AI offers no direct substitute so cost comparison favors human labor as the only viable option currently.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Limited production deployment exists; computer vision systems can analyze photomicrographs once captured, but no mature end-to-end product reliably handles the full workflow from sample preparation through photomicrography without human intervention at multiple steps.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product autonomously prepares physical environmental samples or performs microscopy slide preparation; this remains a manual lab bench task requiring human dexterity.

Develop or implement site recycling or hazardous waste stream programs.

CI 11–25 · 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	Environmental technician roles remain largely in-person, site-specific, and embedded in regulated industries where adoption of AI for high-stakes compliance tasks has been slow; most organizations still rely on trained human technicians rather than algorithmic decision-making for program design and implementation.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental compliance and technician-level fieldwork sectors show slow, limited AI adoption compared to information/professional services sectors.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist technicians by automating waste stream analysis, generating regulatory compliance checklists, drafting program documentation, and identifying recycling best practices, but the technician must retain judgment on site-specific factors and legal responsibility.
Augmentation potential	claude-sonnet-5	3/5	AI tools can help draft waste management plans, summarize regulatory requirements, and organize data, providing moderate assistance while humans handle site-specific implementation and judgment.
Task automatability	claude-haiku-4-5-20251001	1/5	Developing and implementing site recycling or hazardous waste programs requires navigating complex regulatory frameworks, stakeholder engagement, site-specific risk assessment, and designing custom procedural systems—activities fundamentally dependent on human judgment, legal accountability, and contextual problem-solving that current AI cannot execute end-to-end.
Task automatability	claude-sonnet-5	2/5	This task requires site-specific assessment, regulatory knowledge, physical inspection, and stakeholder coordination that current AI cannot execute end-to-end; AI can assist with documentation and planning templates but not design or implement the actual program.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental and hazardous waste programs are heavily regulated by EPA, state agencies, and occupational safety standards; responsible implementation typically requires a licensed or certified technician signature and often formal organizational authorization, creating legal and compliance barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Hazardous waste programs are governed by RCRA and other environmental regulations requiring qualified personnel to certify compliance, creating strong liability and regulatory barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Implementing hazardous waste programs carries substantial liability and regulatory costs; AI tools for supporting analysis are cheap but cannot replace the technician's professional judgment and accountability, making the all-in cost of human oversight still lower than attempting autonomous implementation.
Cost vs. human wage	claude-sonnet-5	2/5	AI could cheaply generate draft plans or summarize regulations, but the bulk of cost is in site assessment, compliance verification, and implementation oversight that still requires paid human technicians.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	While AI can assist with research, draft documentation, and analyze waste streams, no deployed product reliably develops or implements entire waste programs independently; existing systems handle only narrow subtasks like data analysis or document generation, requiring significant human oversight and legal review.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product autonomously develops or implements hazardous waste or recycling programs at a site level; this remains a human-led technical and regulatory task.

Calibrate microscopes or test instruments.

CI 5–30 · exposure 13 · augmentation 38 · 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	Laboratory environments adopting AI are moving slowly on hands-on calibration tasks; most labs still rely on technician-performed procedures with manual logbooks or simple software assistants rather than autonomous systems. Adoption remains limited to large, well-resourced research institutions.
Sector adoption velocity	claude-sonnet-5	1/5	Environmental science field technician work is physical, lab-based, and has seen minimal AI-driven automation of instrument calibration tasks specifically.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-powered measurement verification, automated data logging, and visual anomaly detection can assist technicians by reducing manual inspection time and flagging out-of-spec conditions. However, the task remains primarily technician-driven, so augmentation is helpful but not transformative.
Augmentation potential	claude-sonnet-5	2/5	AI can assist with tracking calibration schedules, logging results, or flagging drift in readings, but it does not meaningfully speed up the physical calibration process itself.
Task automatability	claude-haiku-4-5-20251001	2/5	Microscope and instrument calibration involves precise mechanical and optical adjustments that typically require physical manipulation, visual inspection, and judgment about whether settings meet acceptance criteria. While AI can potentially guide the process via image recognition or measurement data, the end-to-end task including hands-on setup and validation remains largely manual; current systems cannot reliably achieve 50% time savings at equal quality.
Task automatability	claude-sonnet-5	1/5	Calibration requires physical manipulation of hardware, reference standards, and hands-on adjustment that current AI cannot perform without robotic embodiment, which is not standard in this field.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory and quality assurance frameworks (ISO, GLP, FDA guidance in some contexts) often mandate documented human verification and sign-off on calibration procedures to ensure traceability and legal defensibility. Additionally, accuracy failures in calibration can directly compromise downstream experimental validity, raising error costs and liability concerns.
Adoption barriers	claude-sonnet-5	3/5	While no formal licensing mandates a human specifically for calibration, quality control, accreditation standards (e.g., ISO/EPA methods), and instrument-specific certification protocols create meaningful procedural friction against automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The cost of AI systems (including specialized computer vision, calibration software, and integration) does not yet undercut the loaded wage of a trained laboratory technician for this task, especially when accounting for the equipment, integration complexity, and oversight required.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute performing this physical calibration task, so any AI-based approach would require added robotic hardware costing far more than a technician's time.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Some laboratory information systems and instrument software include automated calibration routines, but these are narrow, instrument-specific, and still require human oversight and physical setup. No general-purpose deployed AI system reliably performs full microscope or test instrument calibration independently; human technicians remain essential for hands-on verification.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product autonomously calibrates lab microscopes or environmental test instruments today; this remains a manual technician task.

Investigate hazardous conditions or spills or outbreaks of disease or food poisoning, collecting samples 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	2/5	Environmental and public health sectors show slow digital adoption overall, with most hazard investigation and outbreak response still performed by field technicians in traditional ways. Pilot automation remains limited to lab analysis phases, not field investigation.
Sector adoption velocity	claude-sonnet-5	1/5	Environmental field inspection and sampling remains a physically-dominated, low-digitization sector with minimal AI-driven automation of the core collection task itself.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist technicians by suggesting sampling locations based on historical data, analyzing preliminary field measurements, or flagging anomalies in collected samples—but the human technician remains central to safe, compliant fieldwork execution.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with route planning, hazard prediction, sample data logging, and analyzing lab results afterward, but it does not materially transform the physical collection process itself.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with sample data analysis and categorization post-collection, the investigative fieldwork—identifying hazardous sites, safely collecting samples from unpredictable environments, and making real-time safety decisions—requires human presence and judgment. Current AI systems cannot reliably perform on-site investigation and sample collection autonomously.
Task automatability	claude-sonnet-5	1/5	This requires physical presence at hazardous sites, sample collection using specialized equipment, and hands-on fieldwork that current AI systems cannot perform, as they lack physical embodiment.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory frameworks (EPA, OSHA, CLIA for food/disease investigations) often require licensed or certified personnel to conduct fieldwork and collect samples for legal validity and chain-of-custody compliance. Liability for incorrect sample collection or missed hazards creates strong legal and organizational friction.
Adoption barriers	claude-sonnet-5	4/5	Handling hazardous materials, chain-of-custody requirements for forensic/legal samples, and safety certifications create strong regulatory and liability barriers requiring trained, often certified personnel physically on-site.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The loaded cost of a trained technician in the field remains lower than the combination of AI systems, robotics, data integration, and human oversight needed to approach comparable output for hazard investigation and sample collection.
Cost vs. human wage	claude-sonnet-5	1/5	AI cannot substitute for the physical labor, specialized sampling equipment, and safety protocols required, so there is no AI cost basis for comparison—human labor is the only option.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products perform end-to-end hazard investigation and sample collection at scale. AI tools exist for lab analysis of samples and data synthesis, but the critical on-site investigation and safe sampling collection phase remains primarily manual in production environments.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product can physically travel to contamination sites, collect environmental or biological samples, or handle hazardous materials; this remains purely a human field task.

Inspect sanitary conditions at public facilities.

CI 5–25 · 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	2/5	Public sector health and safety inspection has lagged AI adoption due to regulatory conservatism, small-scale operations, decentralized facilities management, and organizational inertia; pilots are rare and production use minimal.
Sector adoption velocity	claude-sonnet-5	1/5	Public health and environmental inspection is a slow-moving, physically-grounded government sector with minimal AI-driven displacement of on-site inspection duties.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI-powered facility monitoring (e.g., computer vision dashboards flagging problem areas, trend detection) can assist human inspectors by prioritizing areas of concern and reducing routine surveillance overhead, though the human must ultimately validate and certify compliance.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with generating checklists, analyzing photos or sensor data, flagging anomalies, and drafting inspection reports, meaningfully aiding but not replacing the inspector's physical presence and judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	Visual inspection of sanitary conditions can be partially automated with computer vision systems (detecting visible dirt, debris, clutter), but identifying subtle contamination, assessing compliance with nuanced regulatory standards, and making judgment calls about facility safety requires human contextual knowledge and authority that current AI cannot reliably replace end-to-end.
Task automatability	claude-sonnet-5	1/5	Physical on-site inspection of sanitary conditions requires being present, observing, smelling, touching surfaces, and testing equipment in real facilities—no current AI system can perform this end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Health and safety inspections at public facilities carry liability exposure and regulatory requirements; many jurisdictions legally mandate a qualified, credentialed human inspector to document and certify compliance, creating hard adoption barriers.
Adoption barriers	claude-sonnet-5	4/5	Sanitary inspections are typically government-mandated and require certified/authorized inspectors to physically verify and sign off on compliance, creating strong legal and licensing barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Computer vision hardware, integration, and continuous human oversight for validation and enforcement remain expensive relative to the loaded wage of a technician performing routine inspections, particularly at scale across multiple facilities.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing the physical inspection itself, so AI cost is not comparable—human presence is mandatory, making AI more expensive in effect (zero capability at any cost) for the physical component.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Prototype CV systems exist for facility monitoring and hygiene detection, but deployed products lack sufficient accuracy and domain expertise to perform regulatory inspections independently; human inspectors remain the standard in production environments.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously conducts physical sanitary inspections of public facilities; sensors and checklists exist but require a human inspector on-site.

Collect samples of gases, soils, water, industrial wastewater, or asbestos products to conduct tests on pollutant levels or identify sources of pollution.

CI 5–16 · exposure 8 · augmentation 38 · 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	Environmental protection is moderately digitized but remains dependent on physical field presence; adoption of AI for planning and analysis is growing, but the core collection activity is lagging in automation due to its embedded physicality and regulatory constraints.
Sector adoption velocity	claude-sonnet-5	1/5	Environmental field technician work is a physical, low-digitization sector with minimal AI/robotic adoption for sample collection tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can augment technicians by optimizing sampling locations via data analysis, automating lab result interpretation, and streamlining compliance documentation, but the physical collection itself remains human-driven with limited AI enhancement.
Augmentation potential	claude-sonnet-5	2/5	AI can help with planning sampling routes, predicting contamination sources from data, or analyzing lab results afterward, but offers little assistance to the physical collection act itself.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with data logging and some lab analysis of results, the core task of physically collecting samples in the field—determining locations, using specialized equipment, handling hazardous materials safely—requires human presence and judgment that AI cannot perform end-to-end today.
Task automatability	claude-sonnet-5	1/5	This is a physical fieldwork task requiring travel to sites, physical sample collection with specialized equipment, and chain-of-custody handling that current AI cannot perform.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Sample collection for environmental and health compliance typically requires chain-of-custody documentation and is often subject to EPA/OSHA standards; samples must be collected by authorized personnel following strict protocols, creating regulatory and legal barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Regulatory chain-of-custody, certification requirements for sampling protocols, and legal admissibility of samples in environmental compliance cases create strong barriers to non-human execution.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Current technician wages are modest (median ~$46k), while the capital cost of sampling equipment, field vehicles, and any specialized robots or drones for real-world data collection would far exceed the labor cost for actual sample collection and transport.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute for physical sample collection, so the comparison is moot—human labor (or robotics, not AI per se) is the only option today, making AI more expensive/impossible.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI system can autonomously perform multi-step sample collection in real environments (asbestos, industrial wastewater, soil at variable sites) from end to end; field robotics exist for narrow use cases but not at production scale for this occupational task.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product performs physical environmental sampling; this remains entirely a manual field task performed by technicians.

Set up equipment or stations to monitor and collect pollutants from sites, such as smoke stacks, manufacturing plants, or mechanical equipment.

CI 5–16 · exposure 8 · augmentation 38 · importance 3.5/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI in this domain is slow; monitoring stations themselves are increasingly automated and networked, but the initial physical setup and installation remains a bottleneck requiring human expertise with limited incentive for full automation given liability and compliance requirements.
Sector adoption velocity	claude-sonnet-5	1/5	Environmental field technician work is a physically-demanding, low-digitization sector with minimal AI/robotic adoption for equipment installation tasks.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist technicians by recommending optimal sensor placement based on facility maps, automating QA checklists, predicting maintenance schedules, and organizing data workflows, but the human remains essential for actual setup and calibration.
Augmentation potential	claude-sonnet-5	2/5	AI could help plan monitoring station placement or optimize data collection schedules, but offers little assistance for the physical act of setting up equipment on-site.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI could assist with planning sensor placement and data logging workflows, the physical setup of monitoring equipment at diverse industrial sites requires judgment about site conditions, safety hazards, equipment placement specifics, and real-time adjustments that current AI cannot perform end-to-end with 50% time savings.
Task automatability	claude-sonnet-5	1/5	This is a hands-on physical task requiring transporting, installing, calibrating, and securing monitoring equipment at industrial sites; no current AI system can perform this physical setup work.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong barriers exist: workplace safety regulations require qualified technicians to physically enter hazardous industrial sites, OSHA and EPA compliance mandate sign-off by authorized personnel, and equipment calibration often requires licensed oversight or certification.
Adoption barriers	claude-sonnet-5	4/5	Regulatory compliance monitoring often requires certified technicians following specific protocols, site access and safety authorization, and physical presence, creating strong practical and sometimes regulatory barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The loaded cost of a technician performing this task includes their wage, travel, safety equipment, and liability; AI systems cannot replace this labor and any deployed monitoring software still requires human setup and maintenance, making total cost comparable or higher than human labor.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this physical installation task, so AI cost is not applicable/comparable and the human remains the only viable option.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed product reliably performs the physical installation and calibration of pollutant monitoring equipment at industrial sites; this remains a task requiring human technicians with site-specific expertise and hands-on capability.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI or robotic product autonomously sets up environmental monitoring equipment at smoke stacks or plants; this remains entirely a manual field technician task.

Direct activities of workers in laboratory.

CI 0–7 · exposure 0 · augmentation 25 · importance 3.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	1/5	Laboratory management and worker supervision remain highly human-centric functions with no measurable AI adoption in production. Regulatory and liability constraints make autonomous AI direction of human workers essentially absent even in digitized lab environments.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental and lab science sectors show slower AI adoption for supervisory/management tasks compared to information-heavy fields, with pilots focused on data analysis rather than staff direction.
Augmentation potential	claude-haiku-4-5-20251001	2/5	AI could provide modest assistance with scheduling, performance analytics, or protocol recommendations, but cannot meaningfully augment the core function of directly supervising and directing workers, which depends on human authority and presence.
Augmentation potential	claude-sonnet-5	2/5	AI can help schedule tasks, track lab workflows, or flag anomalies, but offers limited direct enhancement to the interpersonal act of directing workers.
Task automatability	claude-haiku-4-5-20251001	1/5	Directing workers requires real-time human judgment, interpersonal communication, motivation, and accountability that current AI cannot reliably perform. The task fundamentally depends on understanding worker capabilities, resolving conflicts, and making dynamic decisions based on lab conditions—all deeply human responsibilities.
Task automatability	claude-sonnet-5	1/5	Directing and supervising human lab workers requires real-time judgment, physical presence, hands-on demonstration, and interpersonal management that current AI cannot perform end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	5/5	This task carries explicit legal and organizational barriers: a human supervisor must be present to legally direct laboratory workers, assume responsibility for safety compliance, and maintain accountability for worker performance and welfare under employment and occupational health laws.
Adoption barriers	claude-sonnet-5	4/5	Supervisory responsibility often carries accountability, safety oversight, and sometimes certification requirements, creating strong organizational and liability barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	Replacing a technician supervisor would require an AI system with supervisory liability and real-time oversight capabilities that do not exist commercially; the cost of building and maintaining such a system would far exceed the loaded wage of the human being replaced.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute for this supervisory role, so AI cost comparison is not meaningful—human labor remains the only functional option.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI systems can autonomously direct human laboratory workers at any scale. While AI can assist with scheduling or data analysis, the supervisory authority and accountability required here remain non-delegable to current AI systems.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product manages or directs laboratory technician staff; this remains a human supervisory function.

Initiate procedures to close down or fine establishments violating environmental or health regulations.

CI 0–0 · exposure 0 · augmentation 50 · 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	1/5	Regulatory and government agencies responsible for environmental enforcement move slowly on automation and maintain strict human oversight requirements. No significant displacement of this task by AI is evident in practice.
Sector adoption velocity	claude-sonnet-5	1/5	Government regulatory/enforcement functions are slow-adopting, highly bureaucratic, and legally constrained, with minimal AI deployment in this specific enforcement role.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by analyzing inspection data, flagging violations, and drafting violation reports, helping technicians prepare stronger cases. However, the final decision and initiation of enforcement remain human responsibilities.
Augmentation potential	claude-sonnet-5	3/5	AI can help draft violation reports, compile evidence, or track regulatory compliance data to support the decision, but the initiation of enforcement remains a human judgment call.
Task automatability	claude-haiku-4-5-20251001	1/5	This task requires legal authority, judgment about regulatory violations, and interpersonal discretion in enforcing compliance. Current AI cannot issue fines, legally initiate enforcement procedures, or make binding regulatory decisions—it lacks legal standing and accountability.
Task automatability	claude-sonnet-5	1/5	This involves formal legal/regulatory enforcement action requiring official authority, judgment, and accountability that AI cannot exercise; no meaningful portion meets the automation bar.
Adoption barriers	claude-haiku-4-5-20251001	5/5	Hard regulatory and legal barriers exist: only authorized human inspectors/technicians can legally initiate enforcement procedures, issue citations, or recommend facility closure. Administrative and liability requirements mandate human judgment and accountability.
Adoption barriers	claude-sonnet-5	5/5	This is a government enforcement action requiring statutory authority, due process, and legal accountability—only an authorized official can legally initiate such action.
Cost vs. human wage	claude-haiku-4-5-20251001	1/5	The task's cost lies heavily in legal liability, regulatory sign-off, and human accountability—elements that cannot be outsourced to AI. A human technician remains necessary, so AI offers minimal cost advantage.
Cost vs. human wage	claude-sonnet-5	1/5	There is no AI substitute performing this function, so cost comparison favors the human entirely; AI cannot legally execute enforcement actions.
Technical feasibility today	claude-haiku-4-5-20251001	1/5	No deployed AI system can independently initiate enforcement actions, fines, or facility closures; these require human authority and legal accountability. Assisting in violation detection is feasible, but the core task of formally initiating regulatory procedures remains out of reach for automation.
Technical feasibility today	claude-sonnet-5	1/5	No deployed AI product initiates enforcement closures or fines against businesses; this remains a government official function performed by authorized personnel.

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.