First-Line Supervisors of Construction Trades and Extraction Workers
47-1011.00Median wage $79,920/yr812,210 employed (US)Rank #405 of 923 scored · top 44% by substitution
Directly supervise and coordinate activities of construction or extraction workers.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution29
Exposure27
Augmentation54
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
15 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
0%
Run 1.0.0-draft.1 · computed 2026-08-05 · rater panel: claude-sonnet-5, claude-haiku-4-5-20251001 · intervals span rater disagreement.
Why this score
The five weighted dimensions of the composite, averaged across this occupation's tasks (importance-weighted, panel mean). Exact weights and formulas: /api/v1/methodology.
Task automatabilityw 35%28
panel mean rating 2.1/5 → substitution pressure 28/100
Technical feasibility todayw 20%26
panel mean rating 2.0/5 → substitution pressure 26/100
Cost vs. human wagew 15%28
panel mean rating 2.1/5 → substitution pressure 28/100
Adoption barriersw 20%inverted — strong barriers lower the score42
panel mean rating 3.3/5 (barrier strength) → substitution pressure 42/100
Sector adoption velocityw 10%22
panel mean rating 1.9/5 → substitution pressure 22/100
Task breakdown (15 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.
Order or requisition materials or supplies.
70CI 52–87 · exposure 70 · augmentation 75 · importance 4.0/5 · click for rater detail
Order or requisition materials or supplies.
70| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Construction and extraction firms are increasingly adopting integrated ERP and procurement automation; large contractors deploy these systems routinely. Smaller firms lag, but the trend across the sector is toward faster, AI-assisted or fully automated ordering. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a historically low-digitization sector with fragmented small firms, so despite available e-procurement tools, actual production-scale AI adoption for this task remains limited. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists supervisors by flagging low-stock items, suggesting vendors, verifying part specs, and pre-filling orders—leaving the human to authorize and override for non-standard needs. This augmentation materially raises supervisory productivity on material management. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-enabled inventory tracking, demand forecasting, and automated purchase-order generation can meaningfully speed up and reduce errors in the requisition process while the supervisor still approves final orders. |
| Task automatability | claude-haiku-4-5-20251001 | 5/5 | Ordering and requisitioning materials is highly structured, rule-based work involving inventory checks, vendor catalogs, and purchase order generation—tasks current AI systems handle reliably. Modern procurement software and agent systems can autonomously manage the full workflow from need identification to order placement at 50%+ time savings. |
| Task automatability | claude-sonnet-5 | 3/5 | Ordering/requisitioning materials against known specs, inventory levels, and vendor catalogs is a structured, rules-based process that AI/software can largely handle, but it still requires site-specific judgment about quantities, timing, and substitutions tied to construction schedules and conditions. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | No legal requirement for a licensed human to place orders; minimal regulatory friction. Adoption barriers are mainly organizational (legacy systems, supplier relationships, trust) rather than structural or liability-driven. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement blocks AI-assisted ordering, though vendor relationships, credit/purchasing authority, and liability for incorrect orders create some organizational friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 5/5 | Automated procurement (via APIs, ERP systems, or agent orchestration) costs a fraction of the supervisor labor it replaces—infrastructure is amortized, and inference cost is negligible per order. Human oversight remains minimal for routine reorders. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Procurement automation software has real licensing and integration costs, and complex construction orders still need human review, so savings versus a supervisor's time are moderate rather than order-of-magnitude. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Enterprise procurement and inventory management systems with AI integration are deployed at scale in construction and supply-chain contexts. Most ordering is now digitized; some gaps remain in real-time yard-level inventory reconciliation and exception handling, but core functionality is production-proven. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Procurement software with automated reordering, e-procurement platforms, and AI-assisted supply chain tools are deployed in construction and industrial settings, but many small-to-mid firms still rely on manual ordering by supervisors due to variable job-site needs. |
Record information, such as personnel, production, or operational data on specified forms or reports.
69CI 65–72 · exposure 70 · augmentation 75 · importance 3.6/5 · click for rater detail
Record information, such as personnel, production, or operational data on specified forms or reports.
69| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Construction remains moderately digitized and fragmented, with adoption of automated reporting systems lagging finance or tech sectors. Larger firms and general contractors are beginning to deploy job-site data capture, but small-to-medium firms still rely on manual logging, limiting sector-wide adoption velocity. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a traditionally low-digitization sector with slower software adoption compared to information/finance industries, though digital field reporting tools are gradually spreading. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI assists significantly by auto-populating forms from photos, sensor data, or voice dictation, reducing transcription burden and error-checking load on supervisors while they remain responsible for accuracy and sign-off. This is transformative for daily reporting workflows even where humans retain final authority. |
| Augmentation potential | claude-sonnet-5 | 4/5 | Mobile apps, voice dictation, and templated reporting tools already meaningfully speed up supervisors' data recording and reduce transcription errors while keeping them in the loop for verification. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Recording personnel, production, or operational data on forms/reports is largely structured data entry—highly automatable with current AI systems that can parse documents, extract information, and populate databases or forms. The main barrier is data source integration, which is routine for deployed systems, leaving minimal manual intervention needed. |
| Task automatability | claude-sonnet-5 | 4/5 | Recording structured data onto forms/reports is a well-defined data entry/transcription task that current AI (OCR, voice-to-text, form-filling agents) can largely automate, though field data capture and edge cases require setup. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Few legal or regulatory barriers exist to automating form-filling and data recording in construction contexts; standard occupational safety and production logging face no licensing mandate on the automation itself. Main friction is integration with legacy systems and organizational inertia, not structural barriers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement dictates a human must record this data; main friction is organizational habit, existing paper-based workflows, and site connectivity issues rather than legal or safety barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Automated data recording has very low per-task inference costs (pennies per form or report) plus modest integration overhead, far cheaper than a human's hourly wage for the same data entry work. This is an order of magnitude cost advantage in most scenarios. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Automated data capture and reporting tools (mobile apps, OCR, integrated ERP/construction management platforms) are far cheaper per record than supervisor time once implemented, though initial setup adds cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Many organizations already deploy robotic process automation (RPA) and AI-driven form-filling systems in production. Document parsing and data extraction are mature capabilities; construction management software increasingly automates field-to-report pipelines, though some domain-specific forms may still require manual review. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Digital field reporting apps and voice-to-text tools exist in construction management software, but many sites still rely on paper forms or manual entry into legacy systems, limiting reliable end-to-end deployment. |
Analyze worker or production problems and recommend solutions, such as improving production methods or implementing motivational plans.
30CI 30–30 · exposure 25 · augmentation 50 · importance 4.1/5 · click for rater detail
Analyze worker or production problems and recommend solutions, such as improving production methods or implementing motivational plans.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction is a traditionally slower-adopting sector with fragmented, often small firms, limited digital infrastructure, and safety-driven conservatism. AI-driven problem analysis in construction supervision is still pilot-stage; production deployment is rare except in larger firms with dedicated analytics teams. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a low-digitization, physical-labor-heavy sector with slow AI adoption relative to information/professional services, though some analytics tools are being piloted for productivity tracking. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can usefully assist supervisors by analyzing crew performance data, flagging safety or productivity outliers, and suggesting structured intervention frameworks. However, the augmentation is partial—the supervisor must still interpret context, motivate workers, and own final decisions. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help analyze production data, identify patterns, and suggest generic motivational or efficiency strategies, providing useful decision support while the supervisor retains contextual judgment and implementation responsibility. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Analyzing worker/production problems and recommending solutions requires contextual judgment, stakeholder input, and domain expertise that current AI struggles with at scale. While AI can help surface data patterns and draft suggestions, the human supervisor's lived knowledge of on-site dynamics, individual worker capabilities, and organizational constraints is difficult for AI to replace end-to-end at the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires on-site judgment, knowledge of specific crew dynamics, physical production constraints, and interpersonal factors that AI cannot directly observe or fully model; AI can assist analysis but not autonomously perform the diagnosis and recommendation end-to-end at equal quality. |
| Adoption barriers | claude-haiku-4-5-20251001 | 3/5 | Construction supervisors often have union roles, safety certifications, or contractual accountability that require a human decision-maker to be responsible for crew management and safety recommendations. However, the task itself is not legally locked to a licensed professional, creating moderate but not absolute barriers to automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement specifically bars AI from this analytic task, but organizational trust, liability for safety/production decisions, and the need for physical presence create meaningful friction against full delegation to AI. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI tools capable of analyzing construction problems (data platforms, dashboards) require significant setup, domain tuning, and human validation for each recommendation. The integrated cost (tooling + integration + mandatory human review) remains comparable to or exceeds the cost of a supervisor's time spent on analysis. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools (e.g., dashboards, LLM advisors) are cheap to run but cannot replace the supervisor's necessary field presence and judgment, so cost savings are marginal since the human role remains essential. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably performs this task in construction supervision at scale. AI analytics tools can flag performance anomalies and suggest generic best practices, but production/safety context in construction is highly site-specific, and current systems lack the reliability to be the primary decision-maker without heavy human oversight. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | No deployed product autonomously diagnoses construction worker/production problems and generates actionable motivational or process solutions; existing analytics tools provide data support but require heavy human interpretation. |
Arrange for repairs of equipment or machinery.
30CI 25–35 · exposure 25 · augmentation 50 · importance 3.8/5 · click for rater detail
Arrange for repairs of equipment or machinery.
30| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction and extraction sectors have historically lower digitization and slower AI adoption compared to information or financial services. While some large firms use maintenance management software, autonomous repair arrangement remains uncommon; most organizations still rely on supervisor judgment and vendor relationships. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a traditionally low-digitization sector with slow AI adoption for operational coordination tasks like maintenance scheduling. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist supervisors by analyzing equipment logs to predict failures, suggesting qualified vendors, or flagging cost outliers, improving the speed and consistency of repair decisions. However, the core task of relationship management and authorization remains human-driven, limiting transformative augmentation. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled maintenance management systems, predictive maintenance alerts, and digital scheduling tools can meaningfully help supervisors track equipment status and streamline repair requests. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Arranging repairs requires identifying equipment failures, selecting appropriate vendors, scheduling work around operations, and negotiating terms—most of which involve human judgment, vendor relationships, and site-specific constraints. AI can assist with some components (e.g., identifying common failure patterns, suggesting vendors) but cannot end-to-end handle the coordination, accountability, and relationship-building inherent in arranging repairs. |
| Task automatability | claude-sonnet-5 | 2/5 | This task involves identifying equipment issues, contacting vendors or maintenance staff, scheduling, and coordinating logistics on a physical job site—largely requiring judgment, phone/in-person coordination, and physical inspection that current AI cannot fully replace. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Arranging repairs often requires authorization authority vested in licensed supervisors, liability assignment, and accountability for equipment downtime costs. Many organizations require human sign-off on repair authorization and vendor selection, especially for safety-critical equipment in construction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No strict licensing requirement for arranging repairs, but organizational trust, vendor relationships, and on-site accountability create moderate friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI solutions for equipment diagnostics and vendor matching require significant setup, integration with maintenance systems, and human oversight. The combined inference and integration cost likely exceeds the savings from automating only partial steps, leaving human labor for relationship and decision components. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools could cheaply handle scheduling notifications, but the core coordination, vendor negotiation, and on-site judgment still require a human supervisor, so all-in cost savings are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably arranges equipment repairs autonomously today. Existing systems can help log issues or suggest repair vendors, but actual arrangement—vendor selection, authorization, scheduling, cost negotiation, liability assignment—still requires human decision-making and communication. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some scheduling/ticketing software and maintenance management systems exist, but they assist rather than autonomously arrange repairs; no deployed product independently diagnoses, negotiates vendor terms, and schedules construction equipment repairs end-to-end. |
Estimate material or worker requirements to complete jobs.
29CI 25–32 · exposure 25 · augmentation 75 · importance 4.1/5 · click for rater detail
Estimate material or worker requirements to complete jobs.
29| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 3/5 | Mid-market and larger construction firms use estimating software and BIM-based takeoff tools, but adoption is mixed and mostly assistive rather than autonomous. Many small and regional firms still rely on manual estimation; AI-driven autonomous estimation in production is limited. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a historically low-digitization sector with slow, uneven AI adoption; estimating software adoption is growing but full AI-driven estimation is still nascent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | AI-powered estimation software meaningfully assists supervisors by auto-generating material takeoffs from plans, flagging outliers, and running scenario analyses. When used as a draft tool reviewed by the supervisor, it can improve speed and catch omissions, though final judgment remains human. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered takeoff and estimating tools meaningfully speed up quantity calculations and historical cost lookups, letting supervisors focus judgment on site-specific labor and scheduling factors. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI can assist with material quantification from specifications and basic labor hour estimation, but real-world estimation requires site-specific variables (terrain, weather, crew skill, logistics) and judgment calls that AI systems handle inconsistently. A supervisor would still need significant oversight to validate AI estimates against project-specific conditions. |
| Task automatability | claude-sonnet-5 | 2/5 | Estimating relies on site-specific judgment, blueprint reading, and experience with local conditions that current AI cannot fully replicate end-to-end, though it can assist with calculations and quantity takeoffs. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Liability and error-cost asymmetries are high: underestimating materials or labor can cause budget overruns, project delays, and safety issues; overestimating wastes client money. Surety bonds, contractual guarantees, and client relationships create legal/reputational risk that typically requires human expert sign-off, not full automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for estimating itself, but liability for cost overruns, contractual risk, and reliance on supervisor judgment create moderate organizational friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI estimation tools (subscription or API-based) cost in the hundreds to thousands annually, but supervisors still dedicate 2–4 hours per estimate validating and adjusting recommendations. The all-in cost (tool + supervision time) remains comparable to having an experienced supervisor do direct estimation. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI tools can speed up quantity takeoffs but still require a skilled estimator or supervisor to validate assumptions, site conditions, and labor productivity rates, keeping overall cost comparable to human-only workflows. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While estimation software exists (BIM, takeoff tools), these require human-supervised inputs and expert review of outputs. No deployed system autonomously and reliably estimates both materials and worker requirements across diverse construction job types without substantial human correction and validation. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Construction estimating software with AI-assisted takeoff features exists but requires significant human review and customization; no deployed product autonomously estimates worker/material needs reliably across varied job types. |
Read specifications, such as blueprints, to determine construction requirements or to plan procedures.
28CI 25–30 · exposure 25 · augmentation 63 · importance 4.4/5 · click for rater detail
Read specifications, such as blueprints, to determine construction requirements or to plan procedures.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a digitization laggard compared to finance or tech; most firms still rely on paper blueprints and human expertise. AI adoption in construction planning is nascent, with few production deployments of autonomous specification-reading systems. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a traditionally low-digitization sector; AI plan-reading tools are in pilot or early adoption stages at large firms, with slow penetration among smaller contractors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by highlighting key specs, flagging discrepancies between documents, and organizing information into checklists, raising supervisor productivity in plan review. However, the supervisor must remain the decision-maker for interpreting requirements and planning procedures. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI tools can quickly highlight specifications, flag discrepancies, and assist in quantity takeoffs, meaningfully speeding up a supervisor's plan review while the human retains final judgment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI can extract basic information from blueprints and specs, but understanding multi-layered construction requirements, material interactions, and project-specific constraints requires expert judgment that AI alone cannot reliably replicate end-to-end. Significant human oversight and validation remain necessary. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can extract and summarize information from blueprints and specifications, but translating that into on-site construction planning requires physical context, judgment, and coordination that current systems cannot fully replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Liability and safety requirements are substantial: errors in reading specifications can cause structural failures, injuries, or code violations. Industry norms, insurance, and regulatory oversight all expect a licensed/experienced supervisor to verify and sign off on construction plans derived from blueprints. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement mandates a human specifically read blueprints, but liability for construction errors, safety codes, and organizational trust in experienced supervisors create meaningful friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-assisted document review is cheaper per document scanned, but the cost of integration, training on domain-specific formats, and mandatory human review/correction for safety-critical construction planning approaches or exceeds the wage of a skilled supervisor doing this task. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Software licenses for plan-reading AI tools are cheap per use, but integration, verification by a qualified supervisor, and liability oversight keep total cost closer to comparable with human review for high-stakes decisions. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | OCR and document parsing tools exist, but deployed solutions struggle with complex technical drawings, nonstandard formats, and the need to synthesize specifications into actionable construction plans. No mature product reliably performs this task in production across diverse construction contexts. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some construction-tech products (e.g., AI blueprint takeoff and clash-detection tools) exist and are used in production, but they are narrow-scope aids rather than full replacements for reading specs and planning procedures. |
Supervise, coordinate, or schedule the activities of construction or extractive workers.
28CI 25–30 · exposure 25 · augmentation 50 · importance 4.4/5 · click for rater detail
Supervise, coordinate, or schedule the activities of construction or extractive workers.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a lower-digitization, physically-grounded sector. While larger firms use project management software, adoption of AI for supervisory automation is minimal and mostly limited to pilot projects rather than production deployment. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a historically low-digitization, physical-labor sector where AI adoption for supervisory functions remains in early pilot stages, lagging behind information/professional services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-powered scheduling and resource planning tools can assist supervisors by automating routine coordination and providing forecasting, helping them focus on safety and personnel issues. However, the assistance is limited to administrative and planning aspects rather than transformative for the full scope of supervision. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-powered scheduling, resource allocation, and progress-tracking tools meaningfully assist supervisors in planning and monitoring, even though hands-on coordination remains human-led. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could assist with scheduling and coordination planning, the task fundamentally requires real-time supervision, worker safety oversight, and context-dependent decision-making on-site. Current AI cannot reliably handle the adaptive judgment and human-facing communication that supervision demands, limiting meaningful automation below the 50% time-saving threshold. |
| Task automatability | claude-sonnet-5 | 2/5 | Scheduling sub-tasks can be aided by software, but real-time supervision of workers on physical sites, resolving conflicts, and adaptive coordination require in-person judgment AI cannot fully replicate today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Construction supervision carries significant regulatory and liability barriers: worker safety laws, OSHA compliance, insurance requirements, and site control responsibilities legally rest with human supervisors. Customers and regulatory bodies expect a licensed or responsible human to oversee worker activities. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing mandates a human supervisor per se, but liability, safety regulations, and the need for on-site physical presence and authority create real organizational friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Integration of AI tools into construction supervision requires significant setup, oversight, and human verification costs. The all-in cost of AI coordination systems remains comparable to or exceeds the loaded wage of a first-line supervisor when accounting for implementation and error management. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Scheduling software reduces some administrative costs, but the bulk of the task (in-person supervision, communication, problem-solving) still requires a paid human supervisor, so overall cost savings are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed products reliably perform full supervisory duties in construction. Scheduling software and project management tools exist but do not replace the core supervisory judgment, safety monitoring, and personnel management central to this role. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Construction scheduling software (e.g., Procore, MS Project) exists and is widely used, but actual supervision of on-site workers remains human-driven with no deployed product performing full supervisory coordination. |
Assign work to employees, based on material or worker requirements of specific jobs.
28CI 25–30 · exposure 25 · augmentation 50 · importance 4.2/5 · click for rater detail
Assign work to employees, based on material or worker requirements of specific jobs.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a low-digitization sector with fragmented, small-firm dominance. While larger firms pilot scheduling software, AI-driven autonomous assignment is rare in production; most adoption remains at the tool-assisted (not autonomous) level. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a traditionally low-digitization sector with slow AI adoption for site management tasks, though some larger firms use project management software increasingly. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-powered scheduling tools can surface candidate assignments, flag material shortages, and highlight worker availability, meaningfully speeding up a supervisor's manual assignment process. However, the supervisor retains final decision authority, making this genuine assistance rather than transformative augmentation. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-based scheduling and resource-planning tools can help supervisors track worker availability, skills, and material needs, improving efficiency while the supervisor makes final assignments. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Assigning work based on material and worker requirements involves situational judgment about job constraints, worker skills, and real-time scheduling. While AI could assist in data-driven recommendations, end-to-end automation would require dynamic problem-solving and contingency handling that current systems struggle with in complex, variable construction environments. |
| Task automatability | claude-sonnet-5 | 2/5 | Requires real-time knowledge of site conditions, worker skills, and material availability that current AI cannot independently assess or act on without heavy human input.The judgment involved in matching workers to physical tasks on a dynamic jobsite resists full automation. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Supervisors have direct responsibility and accountability for safe, compliant job assignments. Liability for injuries, delays, or compliance failures, combined with union agreements and OSHA requirements, creates legal and organizational barriers to removing human supervisory judgment from assignment decisions. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No strict licensing requirement for this specific task, but organizational and safety-related decision-making (who is qualified/certified for a job) creates practical friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI scheduling tools require significant setup, data integration, and ongoing human oversight. The cost of inference plus integration plus human validation approaches or exceeds the cost of a supervisor's time for routine assignments, especially given domain-specific customization needs. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Any AI-assisted scheduling tool would still require significant human oversight, data entry, and validation, so cost savings versus a supervisor's judgment are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed product reliably handles real-world construction work assignment autonomously. Existing scheduling software requires substantial human input and decision-making; systems cannot independently assess job-specific material availability, worker skill-match, and site conditions with production-grade reliability. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some scheduling/workforce management software exists but assignment decisions still require a human supervisor's judgment about crew skill and site-specific conditions; no deployed product autonomously assigns construction trade workers reliably. |
Coordinate work activities with other construction project activities.
28CI 25–30 · exposure 25 · augmentation 63 · importance 4.2/5 · click for rater detail
Coordinate work activities with other construction project activities.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a relatively low-digitization sector with fragmented adoption of even basic software; on-site coordination relies heavily on in-person judgment and communication, limiting adoption of AI solutions to larger, more tech-forward firms. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a traditionally low-digitization sector with slow, uneven adoption of AI-driven project management tools compared to information/finance sectors. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist with scheduling suggestions, resource tracking, and conflict identification, helping supervisors make faster decisions. However, augmentation is limited to data synthesis; the core coordination judgment remains fundamentally human. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-powered scheduling, BIM clash detection, and project management dashboards meaningfully help supervisors track dependencies and communicate across trades, significantly aiding but not replacing the coordination role. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with scheduling and information aggregation, coordinating activities requires real-time judgment about worker safety, resource allocation, and dynamic problem-solving that still demands human oversight. Current systems cannot reliably handle the full end-to-end coordination without substantial human intervention. |
| Task automatability | claude-sonnet-5 | 2/5 | This task requires real-time, on-site coordination among multiple trades, physical inspection of progress, and dynamic decision-making that current AI cannot perform end-to-end without heavy human oversight. Only scheduling/documentation sub-components are automatable today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Legal liability for worker safety, OSHA compliance, and responsibility for project outcomes create strong barriers; a licensed supervisor or responsible human agent typically must sign off on critical coordinations and safety decisions. Union agreements and site management protocols further restrict automation. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing requirement specifically bars AI from scheduling, but liability for safety, contractual coordination, and the need for physical presence on job sites create meaningful organizational and practical barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI coordination tools require significant setup, integration with existing systems, and human oversight to validate decisions. The all-in cost remains comparable to or exceeds the hourly cost of a first-line supervisor performing coordination work. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | While software tools reduce some administrative coordination costs, the human supervisor's on-site judgment, negotiation, and physical presence are still required, keeping overall costs comparable to or only modestly cheaper than a human doing the full task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Scheduling and project management tools exist, but deployed solutions struggle with the complexity of on-site conditions, worker communication, and exception handling typical of construction coordination. No mature product reliably replaces the supervisor's integrative role in dynamic environments. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Construction scheduling software (e.g., Procore, BIM tools) exists and is used to track dependencies, but no deployed AI product autonomously coordinates cross-trade activities on live job sites reliably. |
Train workers in construction methods, operation of equipment, safety procedures, or company policies.
26CI 23–30 · exposure 25 · augmentation 50 · importance 4.0/5 · click for rater detail
Train workers in construction methods, operation of equipment, safety procedures, or company policies.
26| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a physically dispersed, low-digitization sector with fragmented firms and conservative adoption patterns. While some large contractors use digital training platforms, AI-driven autonomous training agents have not penetrated production workflows meaningfully. Adoption remains at pilot or experimental stages. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Construction is a low-digitization, physically-oriented sector with historically slow AI and automation adoption compared to information/professional services. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by drafting training plans, generating video scripts, creating quiz questions, and organizing safety checklists—useful productivity gains for trainers preparing materials. However, the real-time interpersonal and adaptive coaching during live training limits the scope of meaningful augmentation to pre-delivery tasks. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-generated training materials, quizzes, translated safety instructions, and video/AR aids can meaningfully support supervisors in preparing and delivering training content. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | AI could draft training materials or create instructional content, but cannot meaningfully replace hands-on demonstration, real-time feedback to individual workers, or adaptive correction—core elements of effective construction safety and equipment training. The interpersonal and contextual judgment required for on-site training exceed current AI automation capabilities. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can produce training materials and simulations, but the hands-on demonstration, physical coaching, and real-time correction of workers using equipment on a job site cannot be fully replicated end-to-end by current AI systems.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Construction training—especially safety procedures and equipment operation—often falls under regulatory requirements (OSHA, certifications) where a qualified human must sign off or be present. Legal liability for training failures and injury creates strong pressure to retain human trainers as the accountable party, not just an AI reviewer. |
| Adoption barriers | claude-sonnet-5 | 3/5 | OSHA and other safety regulations often require documented, verifiable training and sign-off by qualified personnel, and companies favor experienced supervisors for liability and quality control reasons. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI-generated training content is cheap, but integration, customization for site conditions, live supervision of hands-on learning, and liability oversight still require substantial human labor. The all-in cost of AI-supported training remains high relative to a trainer's hourly wage because human oversight cannot be eliminated. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Developing custom training content or VR modules requires notable upfront investment, and ongoing in-person coaching by supervisors is still needed, so cost savings versus a human trainer are limited. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | While AI can generate training videos and documentation, no deployed product reliably handles the full supervisory training task—assessing worker comprehension, adjusting teaching to individuals, ensuring practical mastery on equipment, or handling safety certification. Narrow use cases (document generation) exist but fall well short of the complete task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Some e-learning and VR safety training products exist in construction, but they supplement rather than replace supervisor-led, hands-on training in most companies today. |
Inspect work progress, equipment, or construction sites to verify safety or to ensure that specifications are met.
25CI 25–25 · exposure 25 · augmentation 50 · importance 4.5/5 · click for rater detail
Inspect work progress, equipment, or construction sites to verify safety or to ensure that specifications are met.
25| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Construction remains a relatively low-digitization sector with fragmented adoption. While large contractors pilot drone and AI inspection tools, small and mid-size firms—the majority—have slow uptake due to cost, integration friction, and tradition of in-person supervision. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a historically low-digitization, physical-labor sector with slow technology adoption; while some large firms pilot AI site-monitoring, most first-line supervisors still perform manual inspections. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-powered cameras and drone analytics can assist supervisors by flagging anomalies and automating documentation, boosting inspection efficiency. However, the inherent need for on-site judgment and safety sign-off limits the transformative potential when the human remains accountable. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI tools such as progress-tracking cameras, drone imagery analysis, and defect-detection software can meaningfully assist supervisors by flagging issues and organizing data, improving efficiency without replacing the inspection role. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI can assist with visual inspection (camera feeds, drone imagery analysis), current systems struggle with the full context and judgment required to verify safety compliance and specification adherence in complex, dynamic construction environments. Real-time hazard detection and regulatory interpretation remain partially manual. |
| Task automatability | claude-sonnet-5 | 2/5 | While computer vision and drone-based inspection tools can flag some visible defects or progress against BIM models, comprehensive safety verification requires physical presence, judgment, and adaptive assessment of dynamic site conditions that current AI cannot fully replicate end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | OSHA and building codes often require a qualified, responsible person (licensed/certified supervisor) to sign off on safety and compliance inspections. Liability and legal accountability for missed defects create strong barriers to full automation without human certification. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Safety inspections are often tied to OSHA compliance and liability, requiring a designated competent person to physically verify conditions and sign off, creating strong legal and organizational barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Drone inspection and AI analysis systems carry significant upfront hardware and licensing costs, plus integration and oversight. For many small-to-mid construction firms, the all-in cost per inspection remains comparable to or higher than a supervisor's hourly rate for a site visit. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | Drone/camera hardware, sensor deployment, and software licensing plus required human oversight make AI-assisted inspection roughly comparable to or more costly than a supervisor's time for routine site checks, especially at smaller job sites. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Computer vision systems exist for defect detection and progress monitoring, but deployed products show material limitations in diverse lighting, angles, and site conditions. No production system reliably replaces the full inspection task; most remain pilot-stage or require heavy human verification. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Products like drone-based progress monitoring and AI safety-camera systems (e.g., Smartvid.io, OpenSpace) exist in construction but are narrow-scope add-ons to human inspection rather than reliable standalone replacements for supervisor judgment. |
Confer with managerial or technical personnel, other departments, or contractors to resolve problems or to coordinate activities.
22CI 14–30 · exposure 17 · augmentation 50 · importance 3.9/5 · click for rater detail
Confer with managerial or technical personnel, other departments, or contractors to resolve problems or to coordinate activities.
22| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Construction remains a relatively low-digitization, site-based sector where face-to-face and phone conferencing dominate critical decisions. Adoption of AI for coordination and problem-solving among humans is minimal in this industry. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Construction is a historically low-digitization, physical-first sector with slow AI adoption relative to information/finance industries. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist by summarizing project data, drafting memos, tracking action items, and flagging recurring coordination issues, improving supervisor efficiency in preparation and documentation. However, the core negotiation and decision-making remain human-led. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI can help summarize communications, draft updates, track action items, and flag scheduling conflicts, meaningfully aiding the supervisor's coordination workflow. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | This task requires nuanced interpersonal negotiation, judgment about trade-offs across departments, and context-dependent problem-solving. While AI can draft communications or summarize issues, autonomous problem resolution and coordination across human stakeholders remains beyond current capabilities; humans must lead and decide. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires in-person or synchronous judgment-driven negotiation across parties with competing interests and physical site context, which current AI cannot conduct end-to-end reliably. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Organizational and legal barriers are substantial: supervisors hold accountability for site decisions and contractor relations, and automating inter-departmental coordination would create liability and authority ambiguity. Stakeholders expect human judgment and formal sign-off on problem resolution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing barrier per se, but liability for safety/contract decisions and strong preference for human accountability and relationship-based coordination create real friction. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Current AI systems would require significant human oversight and validation at each step, making total cost (inference + integration + oversight) comparable to or higher than a supervisor's time spent directly conferring and deciding. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | AI can reduce scheduling and note-taking overhead, but the substantive negotiation and problem-resolution still requires paid supervisor time, so total cost savings are modest. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No production systems reliably conduct multi-stakeholder coordination and real-time problem resolution in construction contexts today. Chatbots can simulate conversation but cannot bind parties to decisions, navigate organizational politics, or hold accountability—all essential to this task. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | AI meeting assistants and coordination tools exist but no product autonomously confers with stakeholders to resolve construction problems; humans remain the active agents in these conversations. |
Locate, measure, and mark site locations or placement of structures or equipment, using measuring and marking equipment.
16CI 5–28 · exposure 13 · augmentation 50 · importance 3.9/5 · click for rater detail
Locate, measure, and mark site locations or placement of structures or equipment, using measuring and marking equipment.
16| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Construction remains a low-digitization sector with significant physical, on-site constraints. Adoption of autonomous marking systems is minimal; the sector still relies almost entirely on human surveyors and supervisors for site layout work. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Construction is a historically low-digitization, physically-oriented sector with slow AI/robotics adoption for on-site layout tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI can assist supervisors through digital layout visualization, automated measurement suggestions from site imagery, and real-time deviation alerts. However, the human supervisor still makes final placement decisions and validates marks on-site. |
| Augmentation potential | claude-sonnet-5 | 3/5 | AI-enabled tools like robotic total stations, GPS-guided stakeout systems, and BIM-integrated layout apps already assist supervisors in planning and verifying measurements, improving accuracy and speed while a human remains in control. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Measuring and marking site locations requires high-precision spatial reasoning and interaction with physical environments. While AI can assist with layout planning from blueprints, the on-site marking task demands real-time adaptation to terrain, weather, and existing conditions that current systems cannot autonomously manage to production quality. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical presence on a construction site, manual measurement, and marking with tools, which current AI systems cannot perform without robotic embodiment far beyond deployed capability. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Safety liability and accuracy requirements create strong barriers: mistakes in site layout cause cascading construction errors and legal liability. Building codes and safety regulations require qualified human sign-off, and insurance/contractual frameworks heavily protect the human role. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing specifically requires a human for this exact task, but liability for incorrect placement, site safety requirements, and physical presence needs create real friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI hardware for autonomous on-site measurement (robotics, surveying systems) remains expensive, and oversight/integration costs are substantial. This is not yet cheaper than deploying a skilled supervisor or technician for the task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI system to compare cost against for this physical task; any robotic solution would require expensive specialized hardware exceeding human labor costs for this task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No deployed products reliably perform autonomous site layout marking end-to-end. Research prototypes exist for computer vision-based measurement, but production systems do not yet exist that can independently locate, measure, and physically mark multiple site elements with the precision required in construction. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs on-site physical layout marking autonomously; robotic total stations and layout robots exist only in narrow pilot/research contexts, not as general supervisor replacements. |
Provide assistance to workers engaged in construction or extraction activities, using hand tools or other equipment.
12CI 5–19 · exposure 8 · augmentation 25 · importance 3.8/5 · click for rater detail
Provide assistance to workers engaged in construction or extraction activities, using hand tools or other equipment.
12| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Construction trades remain physically on-site intensive with slow digital transformation. While some remote monitoring tools exist, actual displacement of first-line supervisors providing hands-on assistance is minimal; workforce culture and regulatory requirements maintain human supervision as the norm. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Construction is a notoriously low-digitization, physical-labor-heavy sector with minimal AI-driven displacement of hands-on tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide limited assistance via safety checklists, tool recommendations, or technique reference materials, but the immediate, hands-on nature of the task—physically assisting workers—offers narrow scope for meaningful AI augmentation while a human supervisor remains present. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can support scheduling, planning, or safety monitoring around this work, but offers little direct assistance to the physical act of using hand tools alongside workers. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could theoretically guide tool selection or demonstrate techniques via video, the hands-on nature of assisting workers with physical hand tools and equipment in dynamic, hazardous construction/extraction environments requires real-time spatial reasoning, safety judgment, and physical presence that current AI cannot provide end-to-end. AI might augment instruction but cannot replace the supervisor's direct assistance role. |
| Task automatability | claude-sonnet-5 | 1/5 | This is physical, hands-on assistance with tools at construction/extraction sites requiring embodied presence, dexterity, and situational judgment that current AI systems cannot perform. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist: OSHA and construction safety regulations typically require licensed or certified supervisors to be present on-site; liability and worker injury costs create high error penalties; union agreements often mandate human supervisors; and the inherent need for human contact and judgment in hazardous environments prevents substitution. |
| Adoption barriers | claude-sonnet-5 | 3/5 | While not formally licensed, safety regulations, liability for jobsite injuries, and the practical need for physical human presence create real friction against any automation attempt. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of deploying remote monitoring or AI guidance systems, combined with the continued need for on-site human supervisors for safety and liability reasons, exceeds the loaded wage of a first-line supervisor performing direct hands-on assistance. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute for this physical task, so any comparison shows AI as either infeasible or far more costly than the human performing it directly. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI product reliably provides in-person physical assistance to construction workers using hand tools and equipment. This task inherently requires a human present on-site to observe, guide, and intervene in real-time safety-critical situations. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed AI product provides hands-on physical assistance to construction/extraction workers using hand tools; this remains purely a human physical labor task. |
Suggest or initiate personnel actions, such as promotions, transfers, or hires.
3CI 0–5 · exposure 0 · augmentation 25 · importance 3.4/5 · click for rater detail
Suggest or initiate personnel actions, such as promotions, transfers, or hires.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Construction trades remain low-digitization sectors with small-to-medium firms, decentralized hiring, and strong union/labor-relations constraints. Adoption of AI for personnel decisions is minimal and lagging far behind white-collar sectors. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Construction is a low-digitization, physical-labor sector with slow AI adoption, especially for people-management decisions. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI can assist modestly by surfacing performance data, flagging candidates from a pool, or flagging training gaps, but the core supervisory judgment on who to promote or hire remains human-centric with limited augmentation potential given legal and interpersonal complexity. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can help draft performance summaries or organize personnel records, but offers limited assistance for the actual judgment-driven decision to promote, transfer, or hire. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Personnel actions like promotions, transfers, and hires require complex judgment about worker performance, interpersonal dynamics, organizational fit, and legal/HR compliance that current AI cannot reliably make end-to-end. The task involves accountability and discretionary decisions that supervisors must own. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires firsthand judgment about worker performance, team fit, and interpersonal dynamics on physical worksites, which AI cannot observe or evaluate end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Personnel decisions are heavily regulated and carry legal liability for discrimination, wrongful termination, and wage claims. Most jurisdictions require human accountability for hiring/promotion decisions, and organizations face significant legal and reputational risk delegating these actions. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Personnel decisions carry legal liability (discrimination, labor law), typically require documented human accountability and often union or HR policy sign-off, creating strong barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of an AI error in hiring or promotion (wrong hire, legal liability, team conflict, turnover) far exceeds the loaded wage of a supervisor making the decision. Current AI oversight would consume the savings, making this uneconomical. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this task, so cost comparison favors the human supervisor by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI system reliably performs personnel actions independently in production. While HR analytics tools can surface candidates or flag performance trends, the actual decision to promote, transfer, or hire requires human evaluation and organizational authority that remains gatekept. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously initiates personnel actions for construction crews; HR software may support documentation but not the judgment or initiation itself. |
Related occupations — Construction & Extraction
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.