Environmental Scientists and Specialists, Including Health

19-2041.00

Median wage $82,220/yr89,250 employed (US)Rank #311 of 923 scored · top 34% by substitution

Conduct research or perform investigation for the purpose of identifying, abating, or eliminating sources of pollutants or hazards that affect either the environment or public health. Using knowledge of various scientific disciplines, may collect, synthesize, study, report, and recommend action based on data derived from measurements or observations of air, food, soil, water, and other sources.

Sub-scores

0–100 · band = confidence interval from rater disagreement

Substitution33

Exposure34

Augmentation71

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

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

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

Tasks on the substitution scale

22 rated tasks, binned by substitution score.

Position among all scored occupations

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

Tasks with substitution ≥ 70

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

Why this score

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

Task automatabilityw 35%35

panel mean rating 2.4/5 → substitution pressure 35/100

Technical feasibility todayw 20%32

panel mean rating 2.3/5 → substitution pressure 32/100

Cost vs. human wagew 15%33

panel mean rating 2.3/5 → substitution pressure 33/100

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%31

panel mean rating 2.2/5 → substitution pressure 31/100

Task breakdown (22 tasks)

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

Prepare charts or graphs from data samples, providing summary information on the environmental relevance of the data.

CI 72–92 · exposure 87 · augmentation 100 · 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 science and consulting firms are moderate-to-high adopters of data analytics automation and AI-assisted visualization tools; BI platforms and Python-based workflows with AI assistance are becoming standard practice in digitized environmental organizations.
Sector adoption velocity	claude-sonnet-5	3/5	Environmental science and consulting firms are moderate adopters of AI tools for reporting tasks, but sector-wide adoption lags behind finance/tech; many labs still use manual or semi-automated workflows.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI substantially augments environmental scientists by automating boilerplate chart creation and initial summaries, freeing them to focus on interpretation, validation, and high-level environmental messaging—a core productivity gain while the human retains critical oversight.
Augmentation potential	claude-sonnet-5	5/5	AI substantially speeds up drafting of charts, graphs, and summary interpretations, letting scientists focus on validation and higher-level analysis, a clear productivity multiplier.
Task automatability	claude-haiku-4-5-20251001	5/5	Creating charts and graphs from data samples is a core data visualization task that current AI systems (Python libraries, BI tools, and agents) can perform end-to-end with high reliability and significant time savings. Summarizing environmental relevance via text generation is also well within AI capability, meeting the ≥50% time-saving threshold.
Task automatability	claude-sonnet-5	4/5	Generating charts and summary narratives from structured data samples is well within current LLM and data-visualization tool capabilities (e.g., Python libraries, BI tools, AI code interpreters), especially when data is already tabulated.mportant caveats around messy raw environmental data reduce it slightly from full automation.
Adoption barriers	claude-haiku-4-5-20251001	2/5	Minimal barriers exist: charts and summaries are outputs that do not require licensing, legal sign-off, or mandatory human oversight in most environmental contexts. The task does not inherently involve liability asymmetry or regulatory gatekeeping of the automation itself.
Adoption barriers	claude-sonnet-5	2/5	No licensing requirement to produce charts/graphs, though scientific reports may require reviewer sign-off by a credentialed specialist for regulatory submissions, creating mild friction.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	AI-driven chart generation and summarization cost pennies per task (minimal API calls, no human labor), while a specialist would spend 30–60 minutes per chart at a loaded wage of $50–80/hour; AI is orders of magnitude cheaper.
Cost vs. human wage	claude-sonnet-5	4/5	Automated charting and basic summary generation is inference-cheap compared to an analyst's hourly time, though initial data cleaning/integration and oversight of correctness add some cost.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature products like Tableau, Power BI, Python (matplotlib/plotly), and AI-assisted code generation tools reliably produce charts and graphs from data in production. Summarization of environmental context is handled competently by current LLMs and deployed data analytics platforms.
Technical feasibility today	claude-sonnet-5	4/5	Tools like ChatGPT with code interpreter, Excel/Power BI AI features, and specialized environmental data platforms already generate charts and narrative summaries reliably in production settings, though domain-specific environmental relevance framing may need human review.

Collect, synthesize, analyze, manage, and report environmental data, such as pollution emission measurements, atmospheric monitoring measurements, meteorological or mineralogical information, or soil or water samples.

CI 30–92 · exposure 62 · augmentation 75 · 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	4/5	Environmental agencies and utilities are rapidly adopting automated sensor networks, data platforms, and AI-driven analysis; adoption is widespread in public and private environmental monitoring sectors, particularly for routine sampling and reporting.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and government/regulatory sectors are typically slower adopters of AI tools compared to finance or tech, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments human specialists by automating routine data handling, flagging anomalies, generating preliminary analyses, and producing draft reports, freeing scientists to focus on interpretation, hypothesis testing, and policy guidance.
Augmentation potential	claude-sonnet-5	4/5	AI significantly aids in data synthesis, statistical analysis, trend detection, and automated report drafting, meaningfully boosting scientist productivity even though physical sampling remains manual.
Task automatability	claude-haiku-4-5-20251001	5/5	AI systems can collect data from sensors and databases, synthesize disparate data sources, perform statistical and geospatial analysis, manage databases, and generate reports—all at scale and with speed savings exceeding 50%. Standard tools like Python libraries, automated ETL pipelines, and AI-driven data interpretation are in widespread production use.
Task automatability	claude-sonnet-5	2/5	Physical sample collection and field monitoring require human/robotic presence, though data synthesis, analysis, and reporting portions can be AI-assisted; the full end-to-end task doesn't meet the 50% threshold given fieldwork requirements.
Adoption barriers	claude-haiku-4-5-20251001	2/5	No legal requirement mandates human scientists perform data collection and analysis; however, some regulatory contexts may require human sign-off on interpretations or policy recommendations, and organizational inertia around validation protocols creates modest friction.
Adoption barriers	claude-sonnet-5	3/5	Regulatory reporting (e.g., EPA compliance data) often requires certified professional sign-off and chain-of-custody protocols for samples, creating moderate legal/procedural barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated data collection, analysis, and reporting infrastructure costs a fraction of employing human specialists for continuous monitoring and reporting—often orders of magnitude cheaper when operating at scale across multiple sites and parameters.
Cost vs. human wage	claude-sonnet-5	2/5	Field collection still requires paid technicians and equipment; AI reduces some analysis/reporting labor but overall cost savings are partial since sampling and instrumentation costs dominate.
Technical feasibility today	claude-haiku-4-5-20251001	5/5	Mature products for data collection (sensor networks, APIs), synthesis (data integration platforms), analysis (statistical software, AI frameworks), and report generation (automated dashboarding, LLM summaries) are deployed daily in environmental monitoring agencies and private firms.
Technical feasibility today	claude-sonnet-5	2/5	Data analysis tools and statistical software exist and are widely used, but no integrated deployed product handles the full collect-to-report pipeline including physical sampling reliably in production.

Monitor environmental impacts of development activities.

CI 25–82 · exposure 58 · augmentation 88 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	4/5	Environmental agencies, mining companies, oil and gas firms, and utilities are rapidly deploying automated monitoring systems; remote sensing and IoT adoption in environmental compliance is accelerating, though enforcement and regulatory agencies lag behind industry.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and consulting are moderate-to-slow adopters of AI, with pilots in remote sensing and GIS analytics but limited production-scale deployment across the sector.
Augmentation potential	claude-haiku-4-5-20251001	5/5	AI dramatically augments environmental scientists' productivity by automating data collection and flagging anomalies, freeing experts to focus on interpretation, policy recommendations, and complex causal analysis while remaining in decision-making authority.
Augmentation potential	claude-sonnet-5	4/5	AI substantially enhances tasks like analyzing satellite/aerial imagery, modeling pollution dispersion, and flagging anomalies, giving scientists strong productivity gains while they remain responsible for interpretation and reporting.
Task automatability	claude-haiku-4-5-20251001	5/5	Environmental impact monitoring is highly automatable through sensor networks, satellite imagery analysis, and machine learning models that detect pollution, land-use changes, and ecological degradation. Current AI systems can continuously collect, process, and flag environmental data with significant time savings and no loss of quality compared to manual field surveys.
Task automatability	claude-sonnet-5	2/5	This task involves field data collection, sensor deployment, site inspections, and professional judgment about ecological and regulatory context that current AI cannot perform end-to-end; AI can assist with data analysis but not the full monitoring workflow.
Adoption barriers	claude-haiku-4-5-20251001	3/5	Regulatory frameworks often require licensed environmental scientists to certify findings and sign off on reports; many jurisdictions mandate professional oversight of environmental assessments, creating meaningful but surmountable adoption friction.
Adoption barriers	claude-sonnet-5	4/5	Environmental monitoring for regulatory compliance often requires certified professionals, chain-of-custody documentation, and legal accountability for reported findings, creating strong licensing and liability barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	5/5	Automated monitoring via satellite imagery, drones, and sensor arrays costs orders of magnitude less per monitored area than hiring environmental scientists for continuous field work and manual data collection.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools can cheaply process satellite imagery or sensor streams, but the overall task still requires human fieldwork, sampling, and compliance judgment, keeping all-in costs comparable to or only modestly below human-only approaches.
Technical feasibility today	claude-haiku-4-5-20251001	4/5	Deployed products for environmental monitoring (remote sensing platforms, IoT sensor networks, AI-driven pollution detection) operate reliably in production across government agencies and private firms, though integration complexity and data standardization challenges occasionally require human oversight or validation.
Technical feasibility today	claude-sonnet-5	2/5	Products exist for remote sensing analysis, satellite change detection, and data logging dashboards, but no deployed system independently conducts comprehensive environmental impact monitoring in production at scale.

Conduct environmental audits or inspections or investigations of violations.

CI 20–66 · exposure 49 · augmentation 63 · importance 3.6/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Environmental compliance and audit firms have begun deploying AI-assisted document review and data extraction; however, adoption remains pilot-stage at most organizations due to regulatory caution, liability concerns, and the requirement for professional certification and human sign-off.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental compliance and field inspection work sits in a moderately digitized but physically-grounded sector with slow, cautious AI adoption compared to office-based professional services.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially assists auditors by automating literature review, prior violation searches, permit comparisons, and preliminary report drafting, reducing manual paperwork and allowing the specialist to focus on site investigation and expert judgment—transforming productivity while the licensed professional remains in control.
Augmentation potential	claude-sonnet-5	3/5	AI can help analyze sensor/satellite data, draft audit reports, and flag anomalies in records, providing useful assistance while humans still perform inspections and judgment calls.
Task automatability	claude-haiku-4-5-20251001	5/5	Environmental audits and inspections of violations involve documentation review, data analysis, checklist verification, and standardized reporting—all routine and automatable. AI can systematically review permits, environmental records, prior violation data, and generate audit reports with ≥50% time savings relative to human auditors performing the same work.
Task automatability	claude-sonnet-5	2/5	On-site inspection requires physical presence, sensory observation, and judgment about site-specific conditions that AI cannot perform end-to-end; AI can assist with report drafting and data review but not the core inspection activity.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental audits often require a licensed professional (PE, environmental consultant, or state-certified inspector) to sign off on findings and testify to regulatory compliance; liability for false clearances creates legal and financial risk if automation errors occur, preventing full substitution.
Adoption barriers	claude-sonnet-5	4/5	Regulatory frameworks often require certified inspectors or qualified professionals to conduct audits and sign off on compliance findings, creating strong legal/licensing barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	4/5	AI inference and integration (regulatory database access, automated document review, report generation) cost a fraction of the loaded hourly wage for a specialist auditor, but oversight by a licensed professional remains legally required, limiting total cost displacement to perhaps 60–80% of the original.
Cost vs. human wage	claude-sonnet-5	2/5	Physical site visits, sampling, and legal documentation still require trained personnel, so AI only reduces costs for the analysis/reporting portion, not the full task.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed products (e.g., document automation, environmental compliance software, data extraction tools) exist and perform parts of this work reliably, but no production system fully replaces the human auditor end-to-end. Material gaps remain in handling novel or complex site conditions, interpretive judgment, and legal liability for certification.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product autonomously conducts physical environmental audits or violation investigations; these remain research-stage or purely assistive tools for documentation.

Process and review environmental permits, licenses, or related materials.

CI 34–43 · exposure 45 · 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 agencies and consulting firms are slow to digitize core workflows; most still rely on manual review and filing. While some large firms pilot AI document processing, deep production adoption in the sector remains limited, reflecting low digitization maturity and regulatory conservatism.
Sector adoption velocity	claude-sonnet-5	2/5	Government and environmental regulatory sectors are typically slow adopters of AI due to legacy systems, procurement cycles, and compliance concerns, with pilots more common than production use.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly boost human productivity by pre-processing permits, extracting key terms, cross-referencing requirements, and flagging potential compliance gaps—freeing specialists to focus on substantive judgment and interpretation. This augmentation is already partly deployed in advanced firms and would materially improve throughput.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully speed up document review, flag inconsistencies, and summarize technical submissions, substantially aiding specialists while they retain final decision authority.
Task automatability	claude-haiku-4-5-20251001	3/5	AI can extract and summarize permit data, flag compliance issues, and organize documentation at scale, but permit review requires domain knowledge, regulatory interpretation, and judgment calls that currently demand human oversight. Roughly half the task—data organization and preliminary flagging—is automatable; final review and approval remain human-dependent.
Task automatability	claude-sonnet-5	3/5	AI can extract data, check permit applications against regulatory checklists, and flag missing information, but final review of technical/legal sufficiency and site-specific judgment still requires human expertise.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Permit approval and sign-off typically require a licensed environmental professional or subject-matter expert; liability and regulatory accountability flow to a human. Many jurisdictions legally require professional review and sign-off, creating hard barriers to full automation regardless of technical capability.
Adoption barriers	claude-sonnet-5	4/5	Permit approval often requires a licensed environmental scientist or regulatory authority sign-off, and errors carry legal/environmental liability, creating strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI infrastructure (OCR, classification models, integration) for permit processing has non-trivial setup and maintenance costs. Environmental scientists' salaries are moderate but the human remains necessary for final judgment, so all-in cost rivals or exceeds the human's loaded hourly rate for the mixed task.
Cost vs. human wage	claude-sonnet-5	3/5	AI-assisted document processing can cut time on routine checks, but integration with agency systems, data quality issues, and required human verification keep costs roughly comparable to human review for now.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Document processing and classification tools exist in production (contract review, compliance screening platforms), but permit review specifically remains limited to narrow scope and often as an assist layer. Material error rates persist when handling complex multi-jurisdictional or novel permit types, so deployment is not yet mature at scale.
Technical feasibility today	claude-sonnet-5	2/5	Some document automation and compliance-checking tools exist in regulatory tech, but no widely deployed product reliably performs full permit review without significant human oversight in production environmental agencies.

Communicate scientific or technical information to the public, organizations, or internal audiences through oral briefings, written documents, workshops, conferences, training sessions, or public hearings.

CI 25–50 · exposure 38 · augmentation 88 · 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 science and government agencies adopt AI tools for document assistance and analysis, but communication to the public remains a human-led function; adoption of autonomous AI for public-facing briefings and hearings is minimal, with most organizations treating AI as a drafting aid rather than a replacement.
Sector adoption velocity	claude-sonnet-5	3/5	Environmental science and government/consulting sectors are adopting AI writing and communication tools at a moderate pace, behind fast-moving sectors like finance or tech, with pilots more common than full production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can meaningfully assist environmental scientists by generating initial drafts, creating visualizations, organizing technical information, and producing multiple messaging variants, allowing humans to focus on refinement, delivery, and audience interaction. This assistant role can substantially raise productivity while the scientist retains control and credibility.
Augmentation potential	claude-sonnet-5	5/5	AI is highly effective at helping scientists draft reports, structure presentations, simplify technical jargon for lay audiences, and prepare training materials, substantially boosting productivity while humans retain final communication responsibility.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate written documents and draft presentations, the task fundamentally requires adapting complex information for diverse audiences with varying expertise and needs—something requiring contextual judgment, audience feedback, and iterative refinement that AI cannot reliably do end-to-end today. Current systems cannot independently orchestrate effective oral briefings, workshops, or public hearings at 50% time savings with equal quality.
Task automatability	claude-sonnet-5	3/5	AI can draft reports, presentations, and training materials effectively, but live oral briefings, public hearings, and audience-adapted communication require human presence and real-time judgment that current AI cannot fully replace.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory and professional norms strongly favor human accountability for communicating environmental and health information to the public; regulatory agencies, courts, and stakeholders expect a credentialed scientist to own and defend statements. Liability, accuracy expectations, and legal standing for public hearings create meaningful barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	No licensing requirement mandates a human deliver this specifically, but public trust, accountability for scientific claims, and expectations of human representation at hearings create real friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI can reduce costs for initial draft generation and formatting, but the labor savings are partial—human environmental scientists must still conduct research, customize messaging, deliver presentations, and manage audience interaction. The all-in cost of AI-assisted communication plus required human oversight remains comparable to or exceeds simple human-performed communication.
Cost vs. human wage	claude-sonnet-5	3/5	AI substantially cuts drafting time for written materials, but the live delivery components (briefings, hearings, workshops) still require paid human time, keeping overall cost savings moderate rather than dramatic.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed AI tools can assist with document drafting and presentation slides, but no product reliably performs the full task of communicating environmental science to mixed audiences across modalities (oral, written, workshop, hearing) at production quality. Tools exist for narrow components (text generation, slide design) but not the integrated, adaptive communication performance this task demands.
Technical feasibility today	claude-sonnet-5	3/5	Products like ChatGPT, Gemini, and specialized writing tools reliably assist with drafting technical documents and slide decks, but no deployed product independently conducts public hearings or live technical briefings.

Develop methods to minimize the impact of production processes on the environment, based on the study and assessment of industrial production, environmental legislation, and physical, biological, and social environments.

CI 25–47 · exposure 45 · augmentation 75 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Environmental science is moderately digitized but dominated by small consulting firms, NGOs, and government agencies with slower IT adoption cycles. Pilot projects exist, but production deployment of AI-driven methodology development is rare; regulatory conservatism slows adoption.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and industrial compliance sectors are moderate-to-slow adopters of AI, with pilots for data analysis but limited deployment for regulatory-facing strategic work.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can powerfully augment environmental scientists through automated literature synthesis, real-time regulatory tracking, rapid what-if scenario modeling, and data visualization of complex environmental systems. These assist human experts in designing more robust mitigation strategies without removing human judgment on feasibility and trade-offs.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by summarizing regulations, modeling scenarios, analyzing environmental data, and drafting reports, significantly speeding up the scientist's overall workflow.
Task automatability	claude-haiku-4-5-20251001	4/5	AI can substantially automate literature review, regulatory compliance mapping, impact modeling simulations, and synthesis of environmental data to generate draft methodologies. However, the task requires expert judgment on trade-offs between production constraints and environmental outcomes, site-specific social factors, and stakeholder engagement—elements that currently need human oversight, achieving roughly 50–70% time savings rather than full automation.
Task automatability	claude-sonnet-5	2/5	This requires synthesizing site-specific engineering data, regulatory nuance, and stakeholder context into novel mitigation strategies, which exceeds current AI's ability to reliably originate without heavy human framing and validation.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory compliance, legal defensibility of environmental impact assessments, and liability for inadequate mitigation methods create strong barriers. Permitting agencies often require sign-off by a qualified environmental professional; stakeholder trust in methodology is also critical and difficult to automate away.
Adoption barriers	claude-sonnet-5	4/5	Environmental compliance work is often tied to regulatory filings, permits, and professional certification (e.g., PE or certified environmental scientist sign-off), creating legal and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Environmental scientists command high salaries (~$70k–$90k loaded), and the task requires expert-level reasoning. AI inference and integration costs are modest, but oversight and expert validation still dominate total cost; AI barely reduces per-task cost and may not achieve cost parity, let alone savings.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can cut research and drafting time, the core analytical and consultative work still requires expert scientists, engineers, and legal review, keeping all-in costs close to human-only baselines.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Deployed AI tools exist for environmental data analysis, LCA (life cycle assessment) modeling, and regulatory database search, but no single end-to-end system reliably produces defensible, site-specific mitigation methods without substantial expert review. Products work within narrow scopes (e.g., emissions calculations) but struggle with the integrative judgment this task demands.
Technical feasibility today	claude-sonnet-5	2/5	AI tools can assist with literature review, regulatory summarization, and drafting reports, but no deployed product independently develops validated environmental impact mitigation methods for production processes.

Plan or develop research models, using knowledge of mathematical and statistical concepts.

CI 25–37 · exposure 30 · augmentation 75 · importance 2.8/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	3/5	Research institutions and environmental agencies increasingly use computational tools and statistical software, but adoption of autonomous or minimally-supervised AI model development remains in the pilot and early-production phase. Traditional peer review and human expertise remain gatekeepers.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science research is a moderately digitized but specialized academic/government sector where AI tool adoption for research design is still nascent, with pilots more common than production-scale deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI tools substantially augment productivity by automating routine statistical calculations, suggesting model forms, and accelerating exploratory data analysis. Environmental scientists using these tools can focus on research questions and interpretation while AI handles computational heavy lifting, markedly raising output quality and speed.
Augmentation potential	claude-sonnet-5	4/5	AI substantially assists with statistical method suggestions, code generation, literature synthesis, and exploratory data analysis, meaningfully speeding up model planning while the scientist retains final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist with statistical computations and suggest model structures, developing novel research models requires domain expertise, problem framing, and judgment about which mathematical approaches fit specific environmental questions. AI tools today cannot fully replace the iterative, creative work of hypothesis formation and model validation that characterizes this task.
Task automatability	claude-sonnet-5	2/5	AI can assist with statistical modeling code and suggest methodologies, but designing a novel research model requires domain judgment, hypothesis framing, and contextual knowledge of environmental systems that current AI cannot reliably originate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental research models often support regulatory decisions and public health conclusions, creating liability and error-cost asymmetry that demands human expert accountability. Professional licensing (environmental science degrees and credentials) and organizational trust in peer-reviewed, human-authored models create substantial adoption friction.
Adoption barriers	claude-sonnet-5	3/5	No strict licensing requirement mandates a human for this specific task, but organizational and scientific-integrity norms (peer review, credentialed authorship, institutional accountability) create meaningful friction against full AI autonomy.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Computational costs for AI-assisted modeling are low, but they supplement rather than replace the expert environmental scientist whose knowledge and judgment are essential. The all-in cost (including necessary human oversight and validation) remains comparable to or exceeds traditional specialist labor.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can cut some coding and literature-review time cheaply, the overall task still requires significant expert oversight and iteration, so all-in cost savings versus a scientist's time are modest rather than order-of-magnitude.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Statistical software and AI-assisted tools exist and are used in research settings, but current systems require significant human oversight for model design, validation, and interpretation. Deployment is common in academic and research organizations, but error rates and the need for expert judgment remain material.
Technical feasibility today	claude-sonnet-5	2/5	Tools like Copilot or code-generation LLMs can help draft statistical scripts or suggest model structures, but no deployed product independently plans a full research model for a novel environmental science question at production reliability.

Analyze data to determine validity, quality, and scientific significance and to interpret correlations between human activities and environmental effects.

CI 25–35 · exposure 30 · augmentation 75 · 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	While environmental organizations use data analysis software, adoption of AI for autonomous interpretation of environmental significance remains limited. Most environmental science workflows continue to rely on human specialists, with AI serving as a supplementary tool rather than a replacement.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and public health sectors are moderate adopters of AI/data tools but lag behind finance or tech in deploying AI directly into scientific validity judgments; many organizations still rely on manual expert review.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly augments human environmental scientists by automating data cleaning, statistical testing, and correlation detection, allowing experts to focus on interpretation and causal reasoning. Tools like machine learning and visualization substantially raise the productivity of skilled specialists.
Augmentation potential	claude-sonnet-5	4/5	AI substantially aids in exploratory data analysis, anomaly detection, visualization, and literature synthesis, meaningfully speeding up an analyst's workflow while the scientist retains responsibility for interpretation and significance calls.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can perform statistical analysis and detect correlations in large datasets, determining 'validity,' 'quality,' and 'scientific significance' require expert judgment about methodology, confounders, and causal inference that exceed current AI capabilities. AI cannot reliably interpret the complex human-activity-to-environmental-effect chains without substantial human oversight.
Task automatability	claude-sonnet-5	2/5	AI can assist with statistical analysis and pattern detection, but determining scientific significance and validity of environmental causal claims requires domain judgment, contextual knowledge, and accountability that current systems cannot reliably replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental policy and remediation decisions based on this analysis often require sign-off by credentialed environmental scientists or engineers, and liability for incorrect interpretations of environmental effects is substantial. Regulatory frameworks typically require human expert judgment and authorization.
Adoption barriers	claude-sonnet-5	3/5	While no license is strictly required to run analyses, findings often feed into regulatory, legal, or public health decisions where credentialed scientific judgment and accountability are expected, creating moderate institutional and liability-driven barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools lower the computational cost of analysis, but the requirement for expert human environmental scientists to validate findings, interpret results, and ensure methodological soundness means the all-in cost remains comparable to or higher than human-only analysis due to required oversight.
Cost vs. human wage	claude-sonnet-5	2/5	AI can reduce time spent on data wrangling and basic statistics, but the need for expert review, correction, and interpretation of ambiguous or noisy environmental data limits net cost savings compared to a trained scientist's judgment.
Technical feasibility today	claude-haiku-4-5-20251001	3/5	Data analysis tools and statistical software exist and are deployed, but determining validity and scientific significance in environmental contexts is typically done by human specialists using AI as an analytical aid rather than as an autonomous system. Products exist but require material human judgment and verification.
Technical feasibility today	claude-sonnet-5	2/5	Data analysis tools (R, Python, ML libraries) and AI copilots exist and are used in research pipelines, but no deployed product autonomously validates data quality and infers causal environmental correlations without expert oversight.

Research sources of pollution to determine their effects on the environment and to develop theories or methods of pollution abatement or control.

CI 25–30 · exposure 25 · augmentation 63 · 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	Environmental science remains dominated by traditional academic and government research institutions with slower digital transformation. While AI-assisted data analysis is growing, autonomous theory development and pollution-source investigation have seen minimal production deployment in the sector.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and regulatory sectors have historically been slower to adopt AI compared to finance or information sectors, though environmental data analytics tools are gradually being piloted.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI tools meaningfully assist environmental scientists with literature mining, geospatial data processing, statistical modeling, and simulation—raising productivity on analysis phases. However, augmentation is partial; field investigation design, causal hypothesis formation, and method innovation still require strong human leadership and judgment.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly enhance literature reviews, data mining of pollution datasets, statistical modeling, and hypothesis screening, meaningfully boosting researcher productivity while humans retain control over research design and conclusions.
Task automatability	claude-haiku-4-5-20251001	2/5	Environmental pollution research involves hypothesis formation, field investigation, complex data interpretation, and novel theory development—capabilities where current AI falls short of 50% time savings at equal quality. While AI can assist with literature review and some data analysis, the core investigative and creative work remains heavily human-dependent.
Task automatability	claude-sonnet-5	2/5	This task requires original research, field investigation, hypothesis generation, and novel theoretical development that current AI cannot perform end-to-end; AI can assist with literature review and data analysis but not conduct the core research and theory-building autonomously.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental research outcomes often inform regulatory policy and environmental remediation decisions; liability for incorrect pollution theories is substantial. Professional licensure (e.g., PG, PE in some jurisdictions), peer review requirements, and stakeholder trust in human expert judgment create meaningful legal and organizational barriers to full automation.
Adoption barriers	claude-sonnet-5	3/5	While no strict licensing mandates a human for this specific research task, scientific credibility, peer review norms, and regulatory reliance on qualified experts create moderate institutional friction against full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	A specialized environmental scientist commands a high loaded wage ($80k–$120k+ annual), and the cost of AI tools plus the extensive human oversight required for novel research design and validation remains comparable to or exceeds the cost of direct human work.
Cost vs. human wage	claude-sonnet-5	2/5	Because AI cannot substitute for the core scientific reasoning and field research, human scientists remain necessary, making AI a supplement rather than a cost-effective replacement for the full task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed production system reliably performs independent pollution-source identification and theory development. AI can support components (e.g., satellite image analysis, literature synthesis) but cannot yet autonomously design and execute the empirical research required to develop novel pollution-control methods.
Technical feasibility today	claude-sonnet-5	2/5	AI tools exist for literature synthesis, data analysis, and pattern detection in pollution datasets, but no deployed product independently researches pollution sources and develops abatement theories at production scale.

Determine data collection methods to be employed in research projects or surveys.

CI 25–30 · exposure 25 · augmentation 63 · importance 3.2/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Academic and government research sectors have low AI adoption for methodology design; most adoption is in data analysis, not study design. Adoption remains experimental and limited to pilot projects rather than widespread production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and public health research sectors show slower AI adoption for core methodological design work compared to fast-moving digital/financial sectors, with AI use concentrated in data analysis rather than study design.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by generating methodological frameworks, suggesting relevant data sources, or cross-referencing similar study designs, thereby helping researchers explore options faster. However, augmentation is moderate since domain expertise and contextual fit assessment remain heavily dependent on human reasoning.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by summarizing best practices, suggesting sampling frameworks, referencing similar studies, and drafting protocol language, substantially speeding up the human's design process while judgment remains human-led.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can suggest data collection methodologies and generate lists of potential approaches, determining methods for research projects requires domain expertise, project constraints assessment, and validation of feasibility—tasks that demand human judgment. AI could assist in filtering or organizing options but cannot fully replace the iterative design and contextual reasoning needed.
Task automatability	claude-sonnet-5	2/5	Designing appropriate data collection methodologies requires domain judgment, knowledge of site-specific conditions, regulatory requirements, and study goals that AI cannot fully replicate end-to-end today, though it can suggest options given a well-specified problem.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Research design choices often require professional judgment, peer review, and adherence to institutional review boards or regulatory standards (e.g., IRB approval for human subjects, EPA protocols). Regulatory and professional norms create friction that prevents full automation.
Adoption barriers	claude-sonnet-5	3/5	Methodology choices often must satisfy regulatory or grant-funding standards and be defensible in peer review or legal contexts, creating moderate professional accountability barriers without requiring formal licensure sign-off in all cases.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI systems for methodology suggestion have moderate integration costs and require substantial human oversight to validate and adapt outputs. The loaded cost of a specialized environmental scientist or researcher significantly exceeds the AI inference cost, but human involvement remains essential.
Cost vs. human wage	claude-sonnet-5	2/5	While AI queries are cheap, the oversight, validation, and domain expertise still required to check methodological soundness means costs remain comparable to or only modestly below human expert time.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No production systems autonomously design research methodologies end-to-end. Academic tools may suggest statistical designs or data sources, but deployed products do not reliably perform complete methodology determination with human-level judgment for real-world projects.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously designs environmental sampling or survey methodologies in production; AI is used mainly as a brainstorming or literature-review aid rather than a reliable methodology designer.

Conduct applied research on environmental topics, such as waste control or treatment or pollution abatement methods.

CI 25–30 · 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 science remains largely in traditional academic and consulting workflows. While digital tools are adopted, automation of research methodology itself is slow; sectors are cautious about AI-driven environmental claims due to regulatory and liability risk.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and government/regulatory-adjacent sectors show slower AI adoption compared to finance or software, with pilots for data analysis emerging but production-scale autonomous research uncommon.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by automating literature synthesis, analyzing monitoring data, and drafting reports, raising a scientist's productivity on administrative and analytical components. However, the core experimental and interpretive work remains human-dependent.
Augmentation potential	claude-sonnet-5	4/5	AI substantially aids literature reviews, data analysis, predictive modeling, and report drafting, meaningfully boosting researcher productivity while humans retain oversight of experimental design and field validation.
Task automatability	claude-haiku-4-5-20251001	2/5	Applied research requires experimental design, hypothesis formation, and interpretation of contextual results—tasks that demand human judgment and creativity. While AI can assist with literature review and data analysis, end-to-end research conception and execution with 50% time savings remains beyond current systems.
Task automatability	claude-sonnet-5	2/5	Applied environmental research requires hands-on experimentation, field sampling, site-specific measurement, and novel hypothesis generation that current AI cannot execute end-to-end; AI can assist with literature review, data analysis, and drafting but cannot conduct the research itself.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Research integrity, regulatory compliance (EPA, institutional review boards), and liability for environmental claims create significant barriers. Many jurisdictions require licensed environmental professionals to author or sign off on applied research findings used in regulatory decisions.
Adoption barriers	claude-sonnet-5	3/5	While no license is strictly required to 'do research,' credentialed expertise, institutional review, regulatory reporting requirements, and liability for environmental findings create meaningful friction against pure AI substitution.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Environmental research involves specialized equipment, field sampling, and regulatory compliance costs. AI inference is cheap, but integration with lab workflows and the oversight required for valid research outputs make total cost competitive with or exceeding a mid-level scientist's wage.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply support literature reviews and data crunching, but the core research—experiment design, fieldwork, equipment operation, and interpretation—still requires paid scientists, keeping overall costs comparable to human-led research.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably conducts applied environmental research independently. AI tools can support components (data analysis, report writing) but research requires hands-on experimentation, stakeholder engagement, and adaptive methodology that lacks production-grade automation.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously conducts applied environmental research; existing AI tools are used piecemeal for literature synthesis, statistical analysis, or modeling support within a human-led research process.

Monitor effects of pollution or land degradation and recommend means of prevention or control.

CI 25–25 · exposure 25 · augmentation 75 · importance 4.0/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Adoption of AI-assisted environmental monitoring is occurring primarily in data-rich sectors (municipalities, large utilities) and mostly for supplementary reporting and early warning, not core decision-making. Smaller agencies and field-based work lag significantly, and professional liability concerns slow replacive adoption.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and government agencies are slower adopters of AI tools compared to finance or tech, with pilots for data analysis but limited production-scale deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI substantially augments environmental specialists by automating data aggregation, trend detection in large pollution datasets, and scenario modeling for control strategies, allowing experts to focus on interpretation and stakeholder engagement rather than manual data wrangling. The human scientist remains essential for judgment and recommendation authority.
Augmentation potential	claude-sonnet-5	4/5	AI tools (satellite imagery analysis, predictive modeling, data visualization) meaningfully enhance an environmental scientist's ability to detect patterns and inform recommendations.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can analyze pollution data, process satellite imagery, and generate reports on trends, the task requires nuanced judgment about causation, local context, and trade-offs between competing concerns that current systems cannot reliably replicate end-to-end. Data collection and preliminary analysis could be partially automated, but recommendation-making critically depends on domain expertise and stakeholder input.
Task automatability	claude-sonnet-5	2/5	Field monitoring, sampling, site inspection, and stakeholder-specific recommendations require physical presence and contextual judgment that current AI cannot perform end-to-end.'
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental recommendations often feed into regulatory decisions, permits, and compliance reporting that legally require sign-off by licensed environmental professionals (PE licenses, state certifications). Liability for incorrect pollution assessments and control failures is asymmetric and substantial, creating strong barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Regulatory reporting, liability for environmental compliance decisions, and often licensed/credentialed professional sign-off create strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI monitoring systems (sensors + inference) have become cheaper for data collection, but integrating, validating, and operationalizing recommendations still requires significant human specialist oversight and field verification, keeping total cost close to or exceeding the cost of employing environmental scientists for this work.
Cost vs. human wage	claude-sonnet-5	2/5	Sensor networks and data analytics reduce some labor costs, but human fieldwork, sampling, and expert judgment remain necessary, keeping AI-only cost savings limited.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	Deployed products can monitor specific pollutants (air quality sensors, water quality AI) and flag anomalies, but no production system reliably performs the full task of both monitoring complex effects and generating actionable, context-aware prevention recommendations at quality parity with specialists. Most systems are narrow single-pollutant monitors or research prototypes.
Technical feasibility today	claude-sonnet-5	2/5	Some products exist for remote sensing analysis and data trend detection, but no deployed system autonomously monitors pollution and issues actionable prevention recommendations reliably.

Review and implement environmental technical standards, guidelines, policies, and formal regulations that meet all appropriate requirements.

CI 25–25 · exposure 25 · augmentation 75 · 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 compliance remains heavily regulated and typically requires human accountability. Adoption of AI for autonomous standard implementation is limited; organizations use AI for drafting and flagging but retain human review and decision-making.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and regulatory compliance sectors have historically been slow adopters of AI compared to finance or tech, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists by parsing complex regulatory documents, cross-referencing organizational policies, identifying compliance gaps, and drafting implementation plans, which can substantially accelerate the human specialist's work while they retain oversight and final authority.
Augmentation potential	claude-sonnet-5	4/5	AI tools can meaningfully assist by quickly searching regulations, summarizing guideline changes, and drafting compliance checklists, boosting productivity while humans retain final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can assist in reviewing and summarizing environmental standards and regulations, the task requires human judgment to evaluate applicability, organizational context, and implementation strategy. Current AI lacks the domain expertise and liability tolerance to autonomously implement standards across complex organizational systems.
Task automatability	claude-sonnet-5	2/5	AI can help summarize and cross-reference regulations, but implementing standards requires judgment, site-specific interpretation, and accountability that current AI cannot fully replace.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental regulations are legally binding and often require sign-off by licensed environmental professionals or authorized personnel. Liability for non-compliant implementation rests with the organization, creating strong organizational and legal barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Regulatory compliance often requires certified professionals (e.g., environmental scientists, PEs) to sign off, and liability for regulatory violations is significant, creating strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Implementation requires specialized environmental scientist expertise, regulatory knowledge, and organizational coordination that AI cannot fully replace. The cost of AI systems plus human oversight and refinement approaches or exceeds the loaded wage of a specialist environmental scientist.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply assist with document review, but the oversight, verification, and liability-bearing implementation still require costly expert human involvement, keeping overall cost comparable to human-only work.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	AI tools can extract and summarize regulatory text and flag compliance gaps, but no deployed product reliably implements environmental standards end-to-end. Feasibility remains at the assisting-human level rather than autonomous execution in production environments.
Technical feasibility today	claude-sonnet-5	2/5	Some legal/regulatory research tools exist and are used to surface relevant standards, but no deployed product reliably 'implements' compliance decisions end-to-end in environmental practice.

Provide scientific or technical guidance, support, coordination, or oversight to governmental agencies, environmental programs, industry, or the public.

CI 25–25 · exposure 25 · augmentation 75 · importance 3.7/5 · click for rater detail

Panel ratings per dimension for this task
Dimension	Model	Rating	Rationale
Sector adoption velocity	claude-haiku-4-5-20251001	2/5	Environmental agencies and regulated industries adopt AI cautiously due to compliance and liability concerns; while data analytics tools are increasingly used, core guidance and oversight roles remain heavily staffed by humans with no evidence of rapid substitution in production settings.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and government-facing consulting sectors have historically been slower AI adopters compared to finance or tech, with pilots emerging but production deployment limited.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI can significantly assist by rapidly synthesizing environmental data, generating preliminary analyses, summarizing regulatory requirements, and drafting technical documents—enabling human environmental scientists to focus on higher-level judgment and stakeholder communication while AI handles information synthesis.
Augmentation potential	claude-sonnet-5	4/5	AI tools significantly help specialists draft reports, summarize regulations, model data, and prepare communications, meaningfully boosting productivity while the professional retains responsibility for guidance and oversight.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help synthesize technical information and draft guidance documents, the task requires nuanced judgment about complex, context-dependent environmental situations and direct coordination with multiple stakeholders—activities that demand human decision-making authority and accountability that AI cannot currently assume end-to-end.
Task automatability	claude-sonnet-5	2/5	This task requires synthesizing domain expertise, judgment, and stakeholder-specific communication in real-time interactions, which current AI cannot fully replicate end-to-end despite being able to draft supporting materials.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Strong barriers exist: governmental agencies and regulated industries typically require credentialed professionals to provide official guidance and oversight; liability, regulatory compliance, and professional licensure/certification create legal requirements for human sign-off on environmental assessments and recommendations.
Adoption barriers	claude-sonnet-5	4/5	Regulatory and legal contexts often require credentialed professionals to sign off on technical guidance to government or industry, creating strong institutional and liability barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The task requires specialized environmental expertise, regulatory familiarity, and stakeholder coordination—qualifications that remain expensive to replicate or oversee with AI systems; human environmental scientists remain cost-competitive for the authoritative guidance this role demands.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can cheaply generate draft reports or summaries, the liability and expertise-verification overhead means human specialist involvement remains costly and largely unavoidable, keeping cost savings modest.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably perform independent oversight and guidance at the level required by agencies and industry; existing AI tools can assist with data synthesis and document drafting but cannot substitute for the expert judgment and legal/professional accountability that characterizes this task.
Technical feasibility today	claude-sonnet-5	2/5	AI chatbots and research assistants can provide background information, but no deployed product reliably serves as the authoritative technical guidance source for agencies or industry without expert oversight.

Provide advice on proper standards and regulations or the development of policies, strategies, or codes of practice for environmental management.

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 regulation and policy development remain slow-moving, oversight-heavy sectors with strong incentive to maintain expert human accountability. Adoption of AI for advisory tasks in this domain is still in early pilot phases rather than production displacement.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental consulting and regulatory advisory sectors have been slower to adopt AI agents in production compared to finance or software, with pilots more common than scaled deployment.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by synthesizing regulatory research, comparing existing standards across jurisdictions, and drafting initial policy language, allowing human experts to focus on strategic judgment and stakeholder negotiation. However, the assistance is partial—the core advisory and decision-making role remains human-driven.
Augmentation potential	claude-sonnet-5	4/5	AI can significantly speed up regulatory research, drafting of policy documents, and comparison of standards, meaningfully boosting the productivity of environmental scientists doing this work.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can summarize existing standards and regulations or draft policy language, providing actionable advice on proper standards requires evaluating complex, context-specific tradeoffs among environmental, economic, and stakeholder concerns that currently demands human expert judgment. AI systems today lack the domain depth and accountability to independently deliver this advisory function at equal quality.
Task automatability	claude-sonnet-5	2/5	This requires synthesizing regulatory context, stakeholder priorities, and professional judgment about applicability to specific situations, which AI can support but not fully replace given liability and contextual complexity.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory bodies and organizations typically require documented expertise and professional accountability from those advising on environmental standards and policy—often involving licensed professionals (environmental engineers, scientists) or legal review. Liability for erroneous standards guidance creates a strong human-sign-off requirement.
Adoption barriers	claude-sonnet-5	4/5	Providing regulatory and policy advice often requires credentialed expertise and carries legal/liability implications, creating strong professional and organizational barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	An experienced environmental scientist's loaded cost is $100k–$150k annually for work on advice and policy; current AI inference for policy synthesis is cheap but requires heavy human oversight and revision by qualified experts, making the all-in cost per advice-equivalent remain higher than the human alternative.
Cost vs. human wage	claude-sonnet-5	2/5	Research and drafting can be cheaply accelerated by AI, but final advisory output still requires expensive expert validation and liability coverage, keeping overall cost comparable to human-driven work.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably performs this task end-to-end; AI tools can assist with research and drafting, but actual advisory and policy development remains substantially manual. Legal and reputational liability for incorrect environmental advice creates deployment friction that has not been overcome at scale.
Technical feasibility today	claude-sonnet-5	2/5	AI tools can summarize regulations and draft policy language, but no deployed product independently provides authoritative regulatory advice or policy strategy in production without expert review.

Investigate and report on accidents affecting the environment.

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 investigation operates in highly regulated, risk-averse sectors (government agencies, large utilities) with slow digitization and strong preference for credentialed human specialists. Adoption of AI for auxiliary tasks (data synthesis, preliminary analysis) is emerging but replacement adoption remains minimal.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental consulting and regulatory compliance sectors show slower AI adoption compared to fast-moving digital industries, with pilots for report generation but limited production-scale investigation automation.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can assist by organizing accident data, drafting preliminary reports, cross-referencing environmental regulations, and identifying relevant case precedents—raising human investigator productivity. However, the core activities (field assessment, causal judgment, witness evaluation) remain human-dependent, limiting augmentation scope.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist with drafting incident reports, summarizing regulations, organizing data, and analyzing patterns from monitoring data, significantly speeding up the reporting component while humans handle investigation and judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	Accident investigation requires complex judgment about causation, environmental impact assessment, and source attribution—tasks demanding human expertise and field knowledge. While AI can assist with report writing and data organization, end-to-end investigation remains heavily dependent on human field work, expert interpretation, and regulatory context.
Task automatability	claude-sonnet-5	2/5	Investigating environmental accidents requires site visits, physical sampling, evidence chain-of-custody, and contextual judgment that current AI cannot perform end-to-end; only report drafting and data synthesis portions are automatable.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory frameworks (EPA, state environmental agencies) typically require licensed or credentialed environmental scientists to conduct official investigations and sign reports. Legal liability for incomplete or incorrect accident assessment creates high error-cost asymmetry, and many jurisdictions mandate human expert sign-off.
Adoption barriers	claude-sonnet-5	4/5	Regulatory reporting requirements often mandate certified/licensed environmental professionals to sign off on findings, and liability for accident causation determinations creates strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Environmental investigations demand field work, expert analysis, and regulatory accountability that AI cannot fully replace. The integrated cost of human specialists (site visits, lab work, expert judgment, liability) remains lower than the cost of AI infrastructure plus mandatory human oversight and investigation.
Cost vs. human wage	claude-sonnet-5	2/5	Human specialists must still conduct site inspections and sampling, so AI only reduces costs on the reporting/documentation portion, yielding modest overall savings relative to full task cost.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed AI system reliably conducts environmental accident investigations independently. AI tools exist for document review and report generation, but investigations involve physical site assessment, witness interviews, regulatory compliance, and causal analysis that require human specialists in production settings.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously investigates environmental accidents; AI tools are used narrowly for document analysis or report drafting support, not the field investigation itself.

Develop the technical portions of legal documents, administrative orders, or consent decrees.

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 legal departments move cautiously on regulatory document automation due to oversight requirements, liability exposure, and the specialized expertise needed. Adoption remains pilot-stage rather than production-wide, with most drafting still performed by human specialists.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental consulting and regulatory sectors show slower, more cautious AI adoption for legally binding documents compared to fast-moving digital-first industries.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can usefully assist environmental specialists by generating initial technical sections, organizing regulatory requirements, or drafting boilerplate language, meaningfully accelerating their work while they retain review and final authority over legal and technical accuracy.
Augmentation potential	claude-sonnet-5	4/5	AI is well-suited to drafting boilerplate, summarizing regulations, and organizing technical content, substantially speeding up the human expert's drafting process while they retain final judgment.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can generate draft text and templates for legal documents, developing technical portions of regulatory documents like consent decrees requires deep domain knowledge of environmental law, compliance standards, and case-specific context. Current AI systems lack reliable access to specialized regulatory databases and cannot independently verify legal sufficiency, falling well short of the 50% time-saving threshold for equal quality.
Task automatability	claude-sonnet-5	2/5	Drafting technical legal language requires integrating regulatory context, site-specific data, and legal precision that current AI can assist with but not reliably complete end-to-end without substantial expert revision.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Consent decrees and administrative orders carry high legal and regulatory weight; errors create liability and may be challenged. Most jurisdictions require licensed attorneys and qualified environmental professionals to attest to technical accuracy and legal sufficiency, creating both formal and practical barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	These documents carry legal and regulatory weight, often requiring sign-off by qualified/certified environmental scientists or attorneys, creating strong liability and authorization barriers.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for legal drafting incur infrastructure, integration, and mandatory expert human oversight costs that remain comparable to or exceed the hourly cost of having skilled environmental specialists draft these documents directly, especially when accounting for quality assurance and liability concerns.
Cost vs. human wage	claude-sonnet-5	2/5	AI drafting reduces some time but the oversight, verification against regulations, and legal risk review needed keep costs closer to comparable rather than dramatically cheaper.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed product reliably produces complete legal/administrative documents meeting regulatory standards without expert human review and revision. AI legal tools exist for simple contracts or form generation, but consent decrees and administrative orders involve complex technical requirements and liability that require lawyer and environmental specialist sign-off, limiting production use to narrow, heavily supervised scenarios.
Technical feasibility today	claude-sonnet-5	2/5	LLMs can draft template language but no deployed environmental-compliance product reliably produces final technical portions of consent decrees or administrative orders without heavy expert review.

Design or direct studies to obtain technical environmental information about planned projects.

CI 25–25 · exposure 25 · augmentation 63 · 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	Environmental consulting and government agencies have adopted AI for specific subtasks (data visualization, statistical analysis) but retain human scientists for study direction and design. Adoption remains conservative due to regulatory requirements and liability concerns.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental consulting and science sectors show slower AI adoption compared to finance or information industries, with pilots for data analysis but limited production-scale agentic use in study design.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist environmental scientists by accelerating literature synthesis, suggesting analytical approaches, generating visualizations, and automating routine modeling—all while the scientist retains control over study design decisions and regulatory compliance.
Augmentation potential	claude-sonnet-5	4/5	AI tools substantially assist with literature synthesis, data analysis, GIS work, and drafting technical reports, meaningfully boosting productivity of the human specialist directing the study.
Task automatability	claude-haiku-4-5-20251001	2/5	AI can assist with literature review, data analysis, and report drafting, but designing studies requires understanding project context, regulatory requirements, and field conditions that demand human expertise and judgment. Current AI cannot autonomously plan comprehensive environmental studies meeting legal and scientific standards.
Task automatability	claude-sonnet-5	2/5	Designing and directing environmental studies requires site-specific judgment, regulatory knowledge, and expert planning that AI cannot execute end-to-end; AI can assist with literature review and drafting but not direct the study.5
Adoption barriers	claude-haiku-4-5-20251001	4/5	Environmental impact assessments and study designs often require licensed professionals (environmental consultants, engineers) and must comply with regulations (NEPA, Clean Water Act, state environmental laws). Liability for inadequate study design creates legal barriers to full automation.
Adoption barriers	claude-sonnet-5	4/5	Environmental studies often require credentialed professionals (e.g., certified environmental scientists) and are tied to regulatory compliance (NEPA, permitting) where liability and legal sign-off create strong barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools (modeling software, analysis platforms) are typically integrated into workflows alongside expensive expert environmental scientists rather than replacing their labor. The oversight and direction required keeps total costs comparable to or exceeding human-only approaches.
Cost vs. human wage	claude-sonnet-5	2/5	AI can cheaply generate literature summaries or draft protocols, but the core task still requires expert oversight, site visits, and regulatory judgment, keeping human labor cost dominant.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed AI products reliably perform end-to-end environmental study design and direction. Tools exist for data analysis and modeling, but the core task of designing appropriate studies with proper scope, methods, and quality assurance remains dependent on human environmental scientists.
Technical feasibility today	claude-sonnet-5	2/5	No deployed product autonomously designs or directs environmental impact studies; existing tools support data analysis and document drafting but do not substitute for expert direction of fieldwork and study design.

Develop programs designed to obtain the most productive, non-damaging use of land.

CI 25–25 · exposure 25 · augmentation 75 · 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 science and land management remain traditionally structured fields with heavy dependence on domain expertise, regulatory process, and stakeholder trust. Adoption of AI-driven automation is slow; tools are used to augment analysis, not replace program development at scale.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and land management sectors are relatively slow adopters of AI compared to information/finance industries, with pilots more common than production deployment.
Augmentation potential	claude-haiku-4-5-20251001	4/5	AI significantly assists with data processing, scenario modeling, environmental impact prediction, and literature synthesis—all core components of program design. Environmental scientists can leverage AI to accelerate analysis and explore alternatives, raising overall productivity while retaining final program authority and judgment.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist with data analysis, modeling land-use scenarios, mapping, and drafting reports, significantly boosting productivity while the specialist retains decision authority.
Task automatability	claude-haiku-4-5-20251001	2/5	This task requires synthesizing complex environmental data, stakeholder input, and long-term ecological modeling to devise land-use strategies. While AI can assist with data analysis and scenario modeling, the creative integration of scientific findings with policy, economic, and social constraints demands human judgment that current systems cannot fully replace.
Task automatability	claude-sonnet-5	2/5	This requires site-specific field assessment, stakeholder negotiation, regulatory judgment, and integrated program design that current AI cannot perform end-to-end; AI can assist with data analysis but not the full program development.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Land-use planning often requires regulatory approval, stakeholder consultation, and legal accountability. Environmental scientists and specialists frequently operate within licensing or professional credentialing frameworks; their final recommendations carry liability and must meet regulatory standards that typically demand human expertise and sign-off.
Adoption barriers	claude-sonnet-5	4/5	Land-use programs often require credentialed environmental scientists, regulatory sign-off, and legal liability considerations, creating substantial barriers to full automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	The specialist salary for environmental scientists is substantial, and current AI tools (modeling software, data platforms) require significant setup, domain expertise integration, and human oversight. The all-in cost of AI-assisted analysis does not yet undercut the cost of a specialist designing and owning the program.
Cost vs. human wage	claude-sonnet-5	2/5	AI tools can reduce time spent on data crunching but the overall program still requires expensive expert labor, site visits, and regulatory compliance work, so cost savings are modest relative to the human specialist's full role.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably generate complete land-use programs end-to-end. AI tools exist for environmental modeling and data analysis, but program development—which involves stakeholder engagement, regulatory compliance, and contextual judgment—remains a human-led process with AI as a support layer.
Technical feasibility today	claude-sonnet-5	2/5	Some GIS and environmental modeling tools support parts of land-use planning, but no deployed product reliably develops complete land-use programs in production without expert oversight.

Evaluate violations or problems discovered during inspections to determine appropriate regulatory actions or to provide advice on the development and prosecution of regulatory cases.

CI 15–25 · exposure 20 · 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	Environmental regulatory agencies are typically laggard adopters of automation due to legacy systems, statutory requirements for human judgment, and risk aversion around enforcement decisions. Adoption remains largely in pilot stages rather than deep production deployment.
Sector adoption velocity	claude-sonnet-5	2/5	Government environmental and regulatory agencies are typically slow adopters of AI for high-stakes enforcement decisions due to legal, political, and procedural constraints.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist by organizing inspection findings, flagging anomalies, summarizing prior case patterns, and surfacing relevant regulations, thereby helping a human specialist work faster and more systematically through evaluation and case development.
Augmentation potential	claude-sonnet-5	4/5	AI can meaningfully assist by summarizing inspection reports, flagging relevant regulations and precedents, and drafting case documentation, significantly speeding up the specialist's research and drafting work.
Task automatability	claude-haiku-4-5-20251001	2/5	While AI can help analyze inspection data and flag potential violations, evaluating violations to determine appropriate regulatory actions requires domain expertise, judgment about enforcement severity, and understanding of contextual factors. Current systems cannot reliably perform the full decision-making end-to-end with the consistency and accountability required.
Task automatability	claude-sonnet-5	2/5	This requires synthesizing regulatory context, legal judgment, and case-specific facts to determine enforcement actions, which exceeds current AI capabilities for reliable end-to-end automation.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.rate what current, generally available AI systems can do today.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Regulatory action and case prosecution have hard legal barriers: government authority and professional licensing requirements mean a qualified human regulatory specialist must ultimately make and sign off on enforcement decisions. Liability and the asymmetric cost of incorrect determinations create strong friction against full automation.
Adoption barriers	claude-sonnet-5	5/5	Regulatory enforcement decisions typically require authorized government officials with legal accountability, and prosecution advice often requires attorney involvement, creating hard legal barriers to automation.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	Current AI tools for regulatory support still require significant human oversight, domain expert integration, and customization. The all-in cost (infrastructure, oversight, liability) approaches or may exceed the loaded wage of a specialized environmental regulatory specialist.
Cost vs. human wage	claude-sonnet-5	2/5	While AI can cheaply summarize documents, the actual judgment and legal risk assessment still requires expensive human expert oversight, keeping costs comparable to or only modestly below human-only work.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No deployed products reliably perform regulatory violation evaluation and enforcement case recommendation at production scale. Some tools assist with data analysis and document review, but the evaluative judgment and legal/regulatory decision-making remain in research or early-stage territory.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs regulatory violation evaluation and enforcement recommendation autonomously; this remains a human expert judgment task supported at most by document retrieval tools.

Supervise or train students, environmental technologists, technicians, or other related staff.

CI 7–25 · exposure 13 · 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	Educational and supervisory functions in environmental science remain predominantly human-led; adoption of AI for automating these roles is slow and limited to administrative aids like scheduling or content drafting rather than replacing the supervisor role itself.
Sector adoption velocity	claude-sonnet-5	2/5	Environmental science and technical fields show moderate AI adoption for technical analysis but supervisory/management functions lag significantly.
Augmentation potential	claude-haiku-4-5-20251001	3/5	AI can meaningfully assist supervisors by generating progress reports, suggesting training content, or flagging performance issues, thereby reducing administrative load and freeing time for mentorship. However, the core judgment and relationship work remains human-dependent.
Augmentation potential	claude-sonnet-5	3/5	AI can assist with training materials, performance tracking, scheduling, and answering technical questions, but the supervisory relationship itself remains human-led.
Task automatability	claude-haiku-4-5-20251001	2/5	Supervising and training inherently requires real-time interaction, judgment about individual progress, and dynamic adaptation to learner needs. While AI could assist with generating training materials or basic feedback, it cannot reliably replace the core supervision and mentoring functions at scale with current systems.
Task automatability	claude-sonnet-5	1/5	Supervision and mentoring of staff/students requires ongoing human judgment, relationship-building, and adaptive feedback that current AI cannot replicate end-to-end.
Adoption barriers	claude-haiku-4-5-20251001	4/5	Supervising and training staff carries implicit accountability for the quality of trainee work and professional development. Organizations and regulatory bodies generally expect a qualified human to own supervision, and professional norms strongly favor human mentorship in scientific and technical fields.
Adoption barriers	claude-sonnet-5	4/5	Organizational structure, accountability for staff performance/safety, and mentorship expectations create strong practical barriers to automating supervisory roles.
Cost vs. human wage	claude-haiku-4-5-20251001	2/5	AI tools for training content or basic tutoring are inexpensive, but the loaded cost of human supervisors and trainers remains substantially lower when amortized across the value and accountability they provide; AI has not yet achieved cost parity for this role.
Cost vs. human wage	claude-sonnet-5	1/5	There is no viable AI substitute for the supervisory role itself, so cost comparison favors the human performing this task.
Technical feasibility today	claude-haiku-4-5-20251001	2/5	No mature product reliably performs comprehensive supervision and training end-to-end. AI can generate templates and answer questions, but real-world supervision involves relationship-building, performance assessment, and corrective feedback that organizations have not yet replaced with AI in production.
Technical feasibility today	claude-sonnet-5	1/5	No deployed product performs live supervision or hands-on training of technical staff; this remains a human management function.

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.