Refuse and Recyclable Material Collectors
53-7081.00Median wage $49,690/yr147,240 employed (US)Rank #631 of 923 scored · top 68% by substitution
Collect and dump refuse or recyclable materials from containers into truck. May drive truck.
Sub-scores
0–100 · band = confidence interval from rater disagreement
Substitution22
Exposure14
Augmentation27
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
14 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
7%
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%14
panel mean rating 1.6/5 → substitution pressure 14/100
Technical feasibility todayw 20%13
panel mean rating 1.5/5 → substitution pressure 13/100
Cost vs. human wagew 15%16
panel mean rating 1.6/5 → substitution pressure 16/100
Adoption barriersw 20%inverted — strong barriers lower the score54
panel mean rating 2.8/5 (barrier strength) → substitution pressure 54/100
Sector adoption velocityw 10%7
panel mean rating 1.3/5 → substitution pressure 7/100
Task breakdown (14 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.
Organize schedules for refuse collection.
74CI 72–75 · exposure 75 · augmentation 75 · importance 3.3/5 · click for rater detail
Organize schedules for refuse collection.
74| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 4/5 | Waste management is a digitized industry with strong financial incentives for route efficiency; scheduling automation is already common in medium-to-large municipal and private waste operations. Adoption is faster in larger, better-resourced organizations. |
| Sector adoption velocity | claude-sonnet-5 | 3/5 | Waste management is a traditionally slow-digitizing sector, but route optimization software has seen steady adoption over the past decade, placing it in the middle range rather than cutting-edge fast adoption. |
| Augmentation potential | claude-haiku-4-5-20251001 | 4/5 | Scheduling AI assists human coordinators by proposing optimal routes and adjusting for constraints, allowing staff to focus on exception handling and service quality improvements. AI significantly amplifies coordinator productivity while keeping judgment in human hands. |
| Augmentation potential | claude-sonnet-5 | 4/5 | AI-based routing tools significantly boost the productivity of schedulers/dispatchers by handling optimization while humans manage exceptions, customer complaints, and special circumstances. |
| Task automatability | claude-haiku-4-5-20251001 | 4/5 | Scheduling refuse collection involves route optimization, calendar management, and constraint satisfaction—tasks that current AI and optimization software handle efficiently. Automated systems can generate schedules accounting for collection frequency, crew availability, and service areas, though human override for last-minute disruptions remains common, limiting it from a full 5. |
| Task automatability | claude-sonnet-5 | 4/5 | Scheduling and route organization is a logistics/optimization problem well suited to software; route-planning and scheduling algorithms already handle this in many municipalities with minimal human oversight beyond exceptions. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Scheduling is not a licensed activity and carries no legal requirement for human sign-off; it is already widely automated. Customer service preferences and organizational inertia create modest friction, but nothing hard prevents full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for scheduling; main friction is organizational (union agreements, legacy systems, public-sector procurement cycles) rather than legal or safety-critical barriers. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Scheduling software costs are typically a small fraction of labor for a scheduling coordinator's time; a mature platform easily delivers 50%+ cost savings per schedule cycle. However, integration costs and ongoing tuning prevent the highest tier. |
| Cost vs. human wage | claude-sonnet-5 | 4/5 | Once implemented, software-driven scheduling is far cheaper per task-equivalent than a human dispatcher manually planning routes and calendars, though initial integration has some cost. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 4/5 | Scheduling software and route optimization tools are in production use across waste management companies (e.g., Routific, OptimoRoute, specialized waste-management systems). These systems reliably generate and adjust collection schedules at scale, though final approval and exception handling often still require human input. |
| Technical feasibility today | claude-sonnet-5 | 4/5 | Commercial waste management routing and scheduling software (e.g., RouteSmart, AMCS, Rubicon) is deployed at scale by municipalities and haulers today to generate and adjust collection schedules. |
Check road or weather conditions to determine how routes will be affected.
46CI 39–52 · exposure 30 · augmentation 63 · importance 3.8/5 · click for rater detail
Check road or weather conditions to determine how routes will be affected.
46| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Waste collection remains a physical, lower-digitization sector where many small operators and municipalities have not yet deployed advanced route optimization or real-time condition monitoring systems, though larger firms are beginning to adopt telematics and mobile solutions. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | Waste collection is a low-digitization, physically-oriented sector with slower AI adoption compared to office-based information work, though route optimization software is spreading gradually. |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | Weather apps and real-time traffic tools can assist drivers by providing timely alerts and suggested route changes, meaningfully improving decision-making on which roads to avoid without replacing the need for human judgment on ground conditions. |
| Augmentation potential | claude-sonnet-5 | 4/5 | GPS, weather apps, and traffic alert systems already meaningfully assist drivers in real time, improving decision speed and route safety while the human remains in control. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Current AI can gather weather and traffic data from APIs and predict some route impacts, but real-time assessment of local road conditions (debris, flooding, ice on specific streets) and dynamic decision-making about actual collection route adjustments requires human judgment and live observation that AI cannot fully replace today. |
| Task automatability | claude-sonnet-5 | 2/5 | AI can retrieve and interpret weather/road data (via apps, GPS, traffic APIs), but integrating this into route decisions for a physical collection job still requires human judgment and on-the-ground verification, so full automation with equal quality is not yet achieved.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 1/5 | There are no licensing, regulatory, or legal barriers to automating weather and route-condition checks; this is a straightforward information-gathering and analysis task with no human-contact or authorization requirement. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing barrier prevents using AI-generated data, but liability for missed pickups or accidents due to route miscalculation creates some caution in fully automating this decision. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 4/5 | Weather and traffic data services are inexpensive and widely available; integrating them with a route system costs far less than paying a human to manually check conditions and update routes daily. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Weather/traffic data APIs are cheap, but the human driver still needs to check and interpret this information as part of their job, so cost savings are moderate rather than transformative. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Weather APIs and traffic data are readily available, but no deployed commercial system reliably combines this data with live local road assessment and automatic route recalculation specifically for refuse collection operations at the granularity needed by individual collectors. |
| Technical feasibility today | claude-sonnet-5 | 3/5 | Products like fleet management software, GPS navigation apps, and weather services already provide real-time road/weather alerts to drivers, but they don't autonomously replan collection routes without human oversight. |
Fill out defective equipment reports.
34CI 23–45 · exposure 33 · augmentation 50 · importance 4.3/5 · click for rater detail
Fill out defective equipment reports.
34| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection remains a low-digitization, small-firm-dominated sector with minimal automation of administrative tasks. Production adoption of AI for equipment reporting is virtually absent in this industry, with most facilities still using manual paper or basic digital forms. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Refuse collection is a low-digitization, physically oriented sector with minimal AI adoption in daily operations or reporting workflows. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide modest assistance by suggesting defect categories or auto-populating routine fields based on equipment type, but the core task—identifying and assessing physical defects—requires human presence and judgment. Limited augmentation potential because the task is straightforward and already relatively quick for trained workers. |
| Augmentation potential | claude-sonnet-5 | 4/5 | Voice dictation and templated text generation can meaningfully speed up and standardize this reporting task even though a human must still identify and describe the defect. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While AI could help structure and categorize equipment defect information, the task requires on-site physical inspection, judgment about defect severity, and context-specific knowledge of equipment conditions that current AI cannot reliably perform end-to-end. Significant human involvement remains necessary for accurate assessment and form completion. |
| Task automatability | claude-sonnet-5 | 3/5 | Filling out structured defect reports from dictated or typed notes is well within current LLM capability, though it requires capturing accurate equipment details from the worker in the field.imony.rounding. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory and safety oversight requirements present meaningful barriers: equipment maintenance and defect reporting are often tied to safety certifications, fleet compliance standards, and liability frameworks that mandate human inspection and sign-off. Organizational responsibility for equipment condition also creates resistance to full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | There's little regulatory or licensing barrier to using software for this administrative task, though some employer verification of accuracy may be required. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | AI systems capable of structured defect reporting (computer vision + form generation) remain expensive relative to a refuse worker completing a report. Integration, training data collection, and human oversight costs exceed the labor savings for this relatively brief task. |
| Cost vs. human wage | claude-sonnet-5 | 3/5 | Voice-to-text and text-generation tools are cheap relative to a worker's time spent writing a report, but integration and verification overhead narrows the savings for such a small task. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | No mature product reliably performs equipment defect reporting in production environments. Vision-based inspection systems exist in research, but automated defect detection on refuse equipment involves variable conditions, physical safety concerns, and regulatory compliance that deployed systems do not handle dependably at scale. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | While generic form-filling and dictation tools exist, no widely deployed product specifically integrates with waste collection fleet maintenance systems to autonomously generate defect reports. |
Operate automated or semi-automated hoisting devices that raise refuse bins and dump contents into openings in truck bodies.
28CI 25–30 · exposure 25 · augmentation 38 · importance 4.0/5 · click for rater detail
Operate automated or semi-automated hoisting devices that raise refuse bins and dump contents into openings in truck bodies.
28| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 2/5 | Refuse collection remains a labor-intensive, geographically dispersed sector with slow digital transformation. Most fleets still rely on traditional semi-automated hoisting rather than advanced autonomous systems, reflecting low tech adoption velocity. |
| Sector adoption velocity | claude-sonnet-5 | 2/5 | placeholder |
| Augmentation potential | claude-haiku-4-5-20251001 | 3/5 | AI-assisted monitoring (computer vision for bin detection, load sensors for weight, predictive alerts for jams) could help operators work more efficiently and safely, though the operator remains essential for positioning and troubleshooting. |
| Augmentation potential | claude-sonnet-5 | 2/5 | placeholder |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While the hoisting mechanism itself is already automated, the task requires human judgment to position the bin correctly, monitor for jams, and respond to real-time conditions (varying bin weights, stuck debris, misalignments). Current AI cannot reliably handle these physical contingencies end-to-end without human intervention. |
| Task automatability | claude-sonnet-5 | 2/5 | The physical driving, navigation, and positioning of the truck at each stop still requires a human, though the hoisting mechanism itself is already automated hardware rather than an AI task; AI doesn't meaningfully change this beyond existing automation.NB the core operator task remains human-driven vehicle control.a} |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Operational safety liability is substantial—equipment failure could injure workers or damage property, creating legal and insurance barriers. Many jurisdictions also require a licensed operator present during refuse truck operations, and DOT/OSHA oversight favors human supervision of mechanical hazards. |
| Adoption barriers | claude-sonnet-5 | 3/5 | placeholder |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | The hoisting device itself is already capital-intensive; adding autonomous control would require significant sensor and software investment with ongoing maintenance, likely exceeding the wage cost of one operator over several years in most markets. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | placeholder |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Automated hoisting systems exist but are already semi-automated and require human operators to initiate, supervise, and troubleshoot. No fully autonomous AI system reliably replaces the operator's role in real-world refuse collection today. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | placeholder |
Tag garbage or recycling containers to inform customers of problems, such as excess garbage or inclusion of items that are not permitted.
19CI 15–24 · exposure 8 · augmentation 13 · importance 3.8/5 · click for rater detail
Tag garbage or recycling containers to inform customers of problems, such as excess garbage or inclusion of items that are not permitted.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Waste collection is a physical, labor-intensive sector with low automation investment to date; most fleets lack digital infrastructure, and adoption of field robotics remains negligible. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a physical, low-digitization sector with minimal AI agent deployment in the field; adoption of automation for this specific inspection task is essentially nonexistent. |
| Augmentation potential | claude-haiku-4-5-20251001 | 1/5 | Current AI tools offer no meaningful assistance in this task; workers rely on visual inspection and local knowledge rather than decision support systems. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Computer vision could potentially help identify contamination in bins from route cameras, offering some assistive value, but this is not a common current practice and adds only marginal support to the tagging task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical presence at containers, visual inspection of contents, understanding of local waste rules, and customer interaction to communicate problems—all of which are beyond current AI capabilities in a mobile field setting without significant infrastructure overhaul. |
| Task automatability | claude-sonnet-5 | 2/5 | While an AI vision system could theoretically detect prohibited items and flag containers, the physical act of tagging containers during a collection route requires physical presence and manipulation that current AI cannot perform end-to-end.dll The task fundamentally couples perception with physical action outdoors in variable conditions.dll |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Although there are no explicit licensing requirements for tagging containers, customer preference for human inspectors and the need for safe, contextual decision-making in residential settings create moderate organizational friction. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement exists for this task, but it requires physical presence on a truck route and interacts with municipal/customer relations, creating moderate organizational friction against pure automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The capital cost of deploying mobile robots with computer vision, plus integration and remote oversight, would far exceed the modest hourly wage of a waste collector for this task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-only solution for this physical inspection-and-tagging task, so any AI cost would be additive to, not a replacement for, the human collector performing the route. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed commercial product can autonomously inspect containers, identify prohibited items, and attach or leave tags in real-world waste collection environments at production scale today. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs this full task; while some smart bin sensors exist in pilot programs, no production system autonomously inspects and physically tags containers during curbside collection. |
Communicate with dispatchers concerning delays, unsafe sites, accidents, equipment breakdowns, or other maintenance problems.
19CI 5–33 · exposure 13 · augmentation 38 · importance 4.0/5 · click for rater detail
Communicate with dispatchers concerning delays, unsafe sites, accidents, equipment breakdowns, or other maintenance problems.
19| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection is a physical, low-digitization sector with small firms; adoption of AI communication automation is minimal and not on the horizon given the workforce composition and regulatory environment. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a low-digitization, physical-labor sector with minimal AI agent deployment for field worker communications. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | Basic communication tools (voice-to-text transcription, automated log templates) could offer modest assistance, but the task itself is already lightweight and leaves little room for meaningful productivity gains. |
| Augmentation potential | claude-sonnet-5 | 3/5 | Dispatch software, GPS tracking, and mobile apps with AI-assisted logging can help streamline reporting and route communication, offering moderate productivity gains. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Communicating with dispatchers about operational issues requires real-time judgment, situational awareness, and human-to-human coordination that current AI cannot perform autonomously. The task involves selective reporting of problems and prioritization decisions that depend on context only a human on-site can assess. |
| Task automatability | claude-sonnet-5 | 2/5 | This requires real-time situational judgment, physical presence at unpredictable field conditions, and verbal communication tied to a physically demanding job; AI could log or transcribe but not perform the core sensing and reporting end-to-end. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Strong barriers exist because safety accountability and real-time incident reporting typically require a human to take responsibility for accuracy, and regulatory/liability frameworks expect a person-to-person dispatch chain for hazardous work environments. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement for this communication task itself, but safety reporting responsibilities and liability for missed hazards create some organizational caution about full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Communication overhead for a refuse collector is minimal and already embedded in human work; AI systems would add integration and monitoring costs without meaningful cost savings, making automation economically irrational. |
| Cost vs. human wage | claude-sonnet-5 | 2/5 | A worker's radio call is nearly free at the margin; adding AI monitoring or communication systems would require sensors, integration, and oversight costing more than the trivial human action. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product autonomously handles communication with dispatchers for this occupational context; such tasks remain within human workflow. Current chatbots lack the real-time field awareness, safety accountability, and organizational integration required. |
| Technical feasibility today | claude-sonnet-5 | 2/5 | Voice/radio communication and fleet dispatch software exist, but no deployed product autonomously detects unsafe sites or breakdowns and communicates them without human initiation. |
Dump refuse or recyclable materials at disposal sites.
15CI 15–15 · exposure 0 · augmentation 0 · importance 4.3/5 · click for rater detail
Dump refuse or recyclable materials at disposal sites.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Waste collection occurs primarily in small enterprises and municipal operations with low digitization and heavy reliance on physical labor; automation adoption in this sector remains minimal and concentrated in vehicle dispatch rather than the physical dumping task. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste management is a low-digitization, physically intensive sector with minimal AI/robotics adoption for this specific end-of-route task. |
| Augmentation potential | claude-haiku-4-5-20251001 | 1/5 | AI offers minimal assistance for the core physical act of dumping materials; existing systems assist with route optimization and scheduling rather than the task of unloading refuse itself. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no assistance to the physical act of dumping materials at a disposal site; route optimization AI exists but doesn't touch this specific task. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Dumping refuse at disposal sites requires physical manipulation of heavy materials in unstructured environments with variable terrain and equipment—beyond the capability of current AI systems. Current robots cannot reliably navigate waste facilities, position containers, and manage diverse waste streams autonomously. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical task requiring driving to a site and operating hydraulic dumping mechanisms; no current AI system performs the physical dumping action, though vehicle automation is a separate emerging technology. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Minor barriers exist; the task requires vehicle operation and site access but no licensing requirement for the dumping act itself. Workplace safety regulations and liability for spills provide some friction but are not hard legal blockers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing uniquely protects this specific dumping action, but safety regulations, vehicle certification, and site protocols create some procedural friction against full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Current autonomous systems capable of handling refuse collection infrastructure are prohibitively expensive compared to the labor costs of human collectors who operate standard collection vehicles. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-driven substitute for this physical task, so any AI cost comparison is moot; human labor with equipment remains the only functional option. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed AI products perform this task. While robotics research exists for some waste handling, no production systems reliably dump refuse at disposal sites without human operators managing vehicle operation and positioning. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously drives collection vehicles to disposal sites and executes dumping operations at scale in production today. |
Dismount garbage trucks to collect garbage and remount trucks to ride to the next collection point.
15CI 15–15 · exposure 0 · augmentation 0 · importance 4.0/5 · click for rater detail
Dismount garbage trucks to collect garbage and remount trucks to ride to the next collection point.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Waste collection remains a traditional, labor-intensive sector with minimal AI/robotics adoption in the core physical collection task, despite some investment in route optimization software. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a low-digitization, physically intensive sector with minimal AI/robotics adoption in production; this is a laggard sector for automation. |
| Augmentation potential | claude-haiku-4-5-20251001 | 1/5 | AI offers minimal assistance for the physical dismounting and garbage collection actions themselves, though route optimization tools may support planning before work begins. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no assistance to a worker physically dismounting and remounting a truck to collect garbage at each stop. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical dismounting and remounting of trucks across varied collection routes, involving balance, coordination, and real-time adaptation to street conditions. Current AI robotics cannot reliably perform these dynamic physical movements at scale in unstructured urban environments. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical task requiring mobility, dexterity, and dynamic navigation of varied street environments that no current AI system can perform; it requires robotic embodiment, not software AI.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | While there are no strict licensing requirements, the task involves public safety (vehicle mounting/dismounting on active streets) and worker injury liability concerns that create moderate friction to automation adoption. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human specifically, but physical safety, liability for injury/property damage, and municipal labor practices create some friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Humanoid robots or specialized collection robots capable of this task would cost hundreds of thousands of dollars per unit with high maintenance, making them far more expensive than the typical loaded wage of a waste collector ($40–60k annually). |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI/robotic substitute available at any price for this physical collection task, so cost comparison favors the human by default. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed commercial systems reliably perform the full cycle of dismounting, collecting garbage, and remounting trucks. Robotics for waste collection remain largely in pilot stages with severe limitations on handling diverse waste types and uneven terrain. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs curbside garbage collection via dismounting/remounting trucks; robotic refuse collection remains experimental at best with no production deployment. |
Make special pickups of recyclable materials, such as food scraps, used oil, discarded computers, or other electronic items.
15CI 15–15 · exposure 0 · augmentation 25 · importance 3.4/5 · click for rater detail
Make special pickups of recyclable materials, such as food scraps, used oil, discarded computers, or other electronic items.
15| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection remains a physical, lower-digitization sector with limited automation adoption. Small firms and municipal operators dominate, with minimal AI agent deployment in production. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste management is a low-digitization, physically intensive sector with minimal AI-driven automation of collection tasks in production today. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist with route optimization and scheduling, but the core act of pickup and material identification offers limited augmentation benefit given the task's primarily physical nature. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can help with route optimization, scheduling special pickups, and dispatch logistics, but offers little assistance to the physical act of collecting and transporting the materials. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical collection of materials from dispersed locations, manual sorting/identification of item types, and navigating variable real-world environments. Current AI cannot operate mobile robots reliably enough for unstructured material pickup at human-equivalent speed and quality at scale. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical driving and lifting task requiring vehicle operation, navigation to varied sites, and manual handling of bulky or hazardous items; no AI system can perform this end-to-end today. |
| Adoption barriers | claude-haiku-4-5-20251001 | 2/5 | Physical presence and direct customer interaction are built into the task, and liability for material handling creates some friction, though no strict licensing requirement exists for this occupational class. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing requirement specifically bars automation, but safety regulations around hazardous e-waste/oil handling and municipal contracts create some friction, though these are not fundamental legal barriers to automation attempts. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The hardware, maintenance, energy, and oversight costs for autonomous collection robots vastly exceed the loaded wage of a refuse collector, especially given current reliability and operational constraints. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI substitute for the physical collection itself, so any AI cost would be additive to, not a replacement for, the human labor and vehicle costs. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed products perform end-to-end special pickup operations reliably in production. Robotics research exists but production systems handling heterogeneous recyclables in the field remain experimental and narrow in scope. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs physical special-pickup collection; this remains firmly in the domain of human labor and vehicles, with robotics for unstructured curbside pickup still research-stage. |
Drive trucks, following established routes, through residential streets or alleys or through business or industrial areas.
14CI 5–23 · exposure 13 · augmentation 25 · importance 4.3/5 · click for rater detail
Drive trucks, following established routes, through residential streets or alleys or through business or industrial areas.
14| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection is a traditionally low-digitization sector in small to mid-sized firms; autonomous adoption is negligible in production, with only scattered pilots. The industry lags professional services and information sectors. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a low-digitization, physical-labor sector with minimal AI/autonomous vehicle adoption in production fleets currently. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | GPS routing and fleet optimization software provide modest assistance, but AI has limited ability to augment the core manual driving task itself without the driver being relieved of responsibility for real-time navigation decisions. |
| Augmentation potential | claude-sonnet-5 | 2/5 | GPS/route optimization software can assist route planning, but this offers only modest productivity gains to the core driving task itself. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | While autonomous vehicles exist, full end-to-end automation of refuse collection routes—including navigation through diverse residential/industrial areas, handling edge cases, and meeting the 50% time-saving threshold reliably—remains in pilot stages. Current systems lack robust real-world deployment at scale. |
| Task automatability | claude-sonnet-5 | 1/5 | Driving a collection truck through varied residential and industrial routes while stopping repeatedly requires real-world physical navigation and vehicle control that current AI/autonomous systems cannot reliably perform end-to-end today.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Regulatory oversight of autonomous commercial vehicles is maturing but not yet blanket-permissive; liability for collisions and injuries in public streets remains legally asymmetric; many jurisdictions require licensed drivers for waste collection routes. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Commercial driving of heavy vehicles on public roads requires licensing (CDL), insurance, and regulatory compliance, and liability for accidents involving pedestrians/property creates strong barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Autonomous vehicle systems remain capital-intensive with high integration and maintenance costs; the economics do not yet favor AI over human drivers in refuse collection when total deployment, liability, and oversight costs are included. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Autonomous heavy-vehicle systems capable of this task don't exist commercially, so any hypothetical cost would include expensive sensor suites, safety systems, and oversight far exceeding a driver's wage today. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 2/5 | Autonomous truck products in controlled environments (mining, warehouses) show promise, but consumer-grade route-driving through residential streets and alleys with real-time safety constraints is not yet a reliably deployed commercial product in refuse collection. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product autonomously drives refuse trucks through mixed residential/alley/industrial routes in production; autonomous driving remains limited to geofenced passenger applications and research pilots. |
Operate equipment that compresses collected refuse.
14CI 5–23 · exposure 8 · augmentation 25 · importance 4.0/5 · click for rater detail
Operate equipment that compresses collected refuse.
14| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection is a laggard sector with limited digitization, small and medium-sized operators, and predominantly physical work. Adoption of AI-driven automation in refuse operations is minimal; the industry is not pushing toward autonomous compressor operation in production. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a low-digitization, physical-labor sector with minimal AI adoption for the operational task itself, though some fleet-level automation trials exist. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could provide minor assistance (e.g., monitoring load sensors, alerting to unsafe conditions, optimizing compaction cycles), but such assistance is limited and not transformative because the operator must remain physically present and in direct control of the equipment for safety and legal reasons. |
| Augmentation potential | claude-sonnet-5 | 2/5 | AI can support route optimization or maintenance alerts for the vehicle, but offers little direct assistance to the physical act of operating compaction equipment. |
| Task automatability | claude-haiku-4-5-20251001 | 2/5 | Compression equipment operation requires physical presence, dexterity, and real-time environmental awareness (checking load, safety hazards, proper positioning). While the mechanical action of activating a compressor is simple, the full task includes monitoring, adjusting, and troubleshooting that resists full automation with current systems. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical equipment-operation task requiring on-site presence, driving, and manual controls; no current AI system can perform this end-to-end.' |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | This task has substantial barriers to automation: equipment operation often requires operator licensing or certification, liability falls heavily on the equipment operator if something fails, and OSHA regulations tightly govern refuse-equipment use and operator presence. Legal responsibility cannot be easily transferred to an autonomous system. |
| Adoption barriers | claude-sonnet-5 | 3/5 | No licensing barrier specifically, but physical safety regulations, equipment liability, and the need for human presence on collection routes create real friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 2/5 | Custom robotic systems capable of compressor operation would require significant capital investment (hundreds of thousands to millions) that far exceeds the loaded cost of a single operator over several years. Current AI-powered automation does not achieve cost parity. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI substitute performing this physical task, so AI cost is effectively infinite relative to human labor for this specific action. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No current deployed AI system can reliably operate refuse-compression machinery end-to-end in the field. Robotics that perform this task exist only in research or highly controlled lab settings; production-grade automation is not commercially available at scale in the refuse industry. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product operates refuse compaction equipment autonomously in production; this remains firmly manual/human-operated work. |
Refuel trucks or add other fluids, such as oil or brake fluid.
10CI 5–15 · exposure 0 · augmentation 13 · importance 4.3/5 · click for rater detail
Refuel trucks or add other fluids, such as oil or brake fluid.
10| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | Refuse collection remains a physical, equipment-dependent sector with low digitization and capital constraints; adoption of autonomous refueling systems is virtually nonexistent in production. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a physically intensive, low-digitization sector with minimal AI/robotics adoption for manual vehicle maintenance tasks like refueling. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could potentially assist through fluid-level monitoring systems or alerts, but the core task—physically adding fuel or fluids—remains human-performed, with only modest augmentation value from sensors or diagnostics. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no meaningful assistance for the physical act of refueling or adding fluids to a truck; this is a manual, hands-on task with no digital augmentation pathway. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical manipulation in unstructured environments (truck fuel ports, fluid reservoirs) with safety-critical decisions. Current AI systems lack the dexterous robots and environmental perception necessary to perform refueling reliably without human oversight. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical manual task requiring a person to handle nozzles, fluids, and vehicle-specific fittings; no off-the-shelf AI system performs this end-to-end today.ateway It requires physical dexterity and mobility that current AI/robotics cannot deliver reliably in this context. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | Safety regulations, worker compensation liability for equipment failure, and operational requirements for human verification of fluid levels and truck readiness create strong organizational and legal barriers to full automation. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing is required to refuel a truck, but there are practical organizational and safety barriers (fuel handling protocols, liability for spills or errors) that create some friction against automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Deploying robotic systems capable of safe autonomous refueling would cost significantly more than the wages of maintenance workers or drivers who perform this task as part of their routine, with substantial integration and safety infrastructure overhead. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | Any robotic solution capable of this task would require expensive specialized hardware, sensors, and manipulation systems far exceeding the low hourly cost of a human worker performing simple fluid checks and refueling. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed products reliably perform autonomous refueling or fluid-addition tasks in real-world refuse collection operations. While robotic arms exist in controlled factory settings, they cannot yet handle the variability of truck maintenance environments. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | There are no deployed commercial products that autonomously refuel trucks or add oil/brake fluid to refuse vehicles in production settings; this remains firmly in the research/prototype stage even for robotic fueling generally. |
Clean trucks or compactor bodies after routes have been completed.
10CI 5–15 · exposure 0 · augmentation 0 · importance 3.6/5 · click for rater detail
Clean trucks or compactor bodies after routes have been completed.
10| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | The waste collection sector is traditionally labor-intensive, lower-digitized, and concentrated in small to mid-sized operations with limited capital for advanced automation. Adoption of AI or robotics in this domain remains minimal. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste management and sanitation is a low-digitization, physically intensive sector with minimal AI/robotics adoption for such maintenance tasks. |
| Augmentation potential | claude-haiku-4-5-20251001 | 1/5 | AI offers no meaningful assistance for the manual physical task of cleaning truck bodies; no autonomous system or tool-based AI augmentation meaningfully aids a human worker in this specific cleaning task. |
| Augmentation potential | claude-sonnet-5 | 1/5 | AI offers essentially no assistance for the physical act of cleaning trucks or compactor bodies. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | Cleaning truck or compactor bodies requires physical manipulation in unstructured environments with variable debris, confined spaces, and safety hazards. Current AI systems cannot perform this embodied task end-to-end without human intervention. |
| Task automatability | claude-sonnet-5 | 1/5 | This is a physical cleaning task requiring manipulation of hoses, brushes, and access to irregular truck/compactor surfaces, which is far outside current AI or robotic system capabilities in general deployment. |
| Adoption barriers | claude-haiku-4-5-20251001 | 4/5 | This task is inherently tied to physical presence on-site and involves worker safety, equipment liability, and operational workflows that create substantial friction against automation. Union agreements and worker-safety regulations in waste collection also raise adoption barriers. |
| Adoption barriers | claude-sonnet-5 | 2/5 | No licensing or legal requirement mandates a human specifically clean trucks, but the physical nature and equipment variability create practical barriers to automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | Autonomous cleaning systems capable of this task would require significant capital investment and site-specific integration, making them considerably more expensive than the direct labor cost of human workers performing the task. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no viable AI-based substitute for this physical task, so any AI-based approach would require costly robotic hardware far exceeding the cost of human labor. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed products reliably clean refuse trucks or compactor bodies autonomously in real-world operations. Specialized robotics research exists but is not in production use for this task. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No commercial product performs autonomous truck or compactor cleaning today; this remains manual labor performed by workers or occasionally automated wash bays, not AI-driven. |
Inspect trucks prior to beginning routes to ensure safe operating condition.
3CI 0–5 · exposure 0 · augmentation 25 · importance 4.5/5 · click for rater detail
Inspect trucks prior to beginning routes to ensure safe operating condition.
3| Dimension | Model | Rating | Rationale |
|---|---|---|---|
| Sector adoption velocity | claude-haiku-4-5-20251001 | 1/5 | This task occurs in fragmented, physically-intensive industries (waste collection, recycling) with low AI adoption rates and strong reliance on driver accountability and physical presence at the vehicle. |
| Sector adoption velocity | claude-sonnet-5 | 1/5 | Waste collection is a low-digitization, physical-labor-intensive sector with minimal AI adoption for hands-on vehicle inspection tasks; telematics adoption is slow and inspection remains manual. |
| Augmentation potential | claude-haiku-4-5-20251001 | 2/5 | AI could assist by flagging visual anomalies in photos or reminding drivers of checklist items, but the core task—physical inspection and safety certification—requires human judgment and legal responsibility that AI cannot augment meaningfully. |
| Augmentation potential | claude-sonnet-5 | 2/5 | Telematics and IoT sensors can alert workers to certain mechanical issues (e.g., tire pressure, engine diagnostics) ahead of or during inspection, offering modest assistance, but do not transform the core physical inspection process. |
| Task automatability | claude-haiku-4-5-20251001 | 1/5 | This task requires physical inspection of truck components (brakes, lights, tires, mirrors, hydraulics) and real-time judgment about safety-critical conditions that vary by vehicle state. Current AI lacks embodied inspection capability and cannot reliably assess mechanical safety across the diversity of truck types and wear patterns without human verification. |
| Task automatability | claude-sonnet-5 | 1/5 | This requires physical inspection of a vehicle (tires, brakes, fluids, hydraulic lifts, lights) which demands physical presence and manipulation; no current AI system can perform this end-to-end.4th-generation sensor systems exist but aren't a full substitute for the task as described.5th, this remains a manual physical task. |
| Adoption barriers | claude-haiku-4-5-20251001 | 5/5 | Department of Transportation (DOT) regulations legally require that drivers or certified mechanics perform and sign off on vehicle safety inspections before operation; autonomous AI cannot satisfy legal accountability or liability requirements for safety-critical transportation tasks. |
| Adoption barriers | claude-sonnet-5 | 4/5 | Vehicle safety inspections are often governed by DOT/company safety regulations requiring a qualified human to physically verify roadworthiness before operation, creating strong regulatory and liability barriers to full automation. |
| Cost vs. human wage | claude-haiku-4-5-20251001 | 1/5 | The cost of building a robot capable of safe, reliable truck inspection, combined with integration and liability oversight, far exceeds the loaded wage of a driver or mechanic performing a 10-15 minute pre-route check. |
| Cost vs. human wage | claude-sonnet-5 | 1/5 | There is no AI system that can substitute for the physical inspection labor, so any 'AI cost' would be for supplementary sensors/telematics layered on top of, not replacing, human labor costs, making AI more expensive as a substitute. |
| Technical feasibility today | claude-haiku-4-5-20251001 | 1/5 | No deployed product performs autonomous pre-route truck safety inspections in production. Computer vision systems exist for limited specific checks, but comprehensive safety certification requires certified mechanics or drivers conducting physical hands-on inspection, not AI-only systems. |
| Technical feasibility today | claude-sonnet-5 | 1/5 | No deployed product performs full pre-trip physical vehicle inspections autonomously; while some fleets use telematics/sensor dashboards to flag issues, the hands-on inspection itself is not replicated by any production AI system. |
Related occupations — Transportation & Material Moving
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.