Refuse sorters
An AI agent can at least partly carry out 18.0% of the task time of refuse sorters, and finish 0.0% alone. The barrier that holds most of the rest back is access to the systems the work runs on.
310th of 433 · interval 0.0–39.5
201st of 433 · interval 0.0–0.0
329th of 433 · interval 8.2–32.6
task statements rated by two models and weighted by importance and frequency. World employment: 2.9 million.
How the tasks rate
What stops the agent
Share of task weight flagged by each barrier; a task can carry several.
| Task statement (O*NET occupation) | Sonnet · Haiku | Barriers | ||||
|---|---|---|---|---|---|---|
| Drive trucks, following established routes, through residential streets or alleys or through business or industrial areas.Refuse and Recyclable Material Collectors | 8.7% | L0 | L0L0 | 2% | 4.0 h | Physical presence, Accountability, Access to systems |
| Dump refuse or recyclable materials at disposal sites.Refuse and Recyclable Material Collectors | 8.4% | L0 | L0L0 | 4% | 22 min | Physical presence, Access to systems |
| Inspect trucks prior to beginning routes to ensure safe operating condition.Refuse and Recyclable Material Collectors | 8.4% | L1 | L1L1 | 12% | 15 min | Physical presence, Accountability, Access to systems |
| Operate automated or semi-automated hoisting devices that raise refuse bins and dump contents into openings in truck bodies.Refuse and Recyclable Material Collectors | 8.2% | L0 | L0L0 | 4% | 2 min | Physical presence |
| Operate equipment that compresses collected refuse.Refuse and Recyclable Material Collectors | 8.0% | L0 | L0L0 | 4% | 18 min | Physical presence, Access to systems |
| Dismount garbage trucks to collect garbage and remount trucks to ride to the next collection point.Refuse and Recyclable Material Collectors | 7.8% | L0 | L0L0 | 1% | 3 min | Physical presence |
| Refuel trucks or add other fluids, such as oil or brake fluid.Refuse and Recyclable Material Collectors | 7.8% | L0 | L0L0 | 4% | 15 min | Physical presence, Accountability, Access to systems |
| Communicate with dispatchers concerning delays, unsafe sites, accidents, equipment breakdowns, or other maintenance problems.Refuse and Recyclable Material Collectors | 7.3% | L1 | L1L2 | 32% | 5 min | Physical presence, Access to systems |
| Fill out defective equipment reports.Refuse and Recyclable Material Collectors | 7.3% | L2 | L2L2 | 58% | 12 min | Physical presence, Accountability, Access to systems |
| Check road or weather conditions to determine how routes will be affected.Refuse and Recyclable Material Collectors | 6.6% | L2 | L2L3 | 75% | 20 min | Access to systems |
| Tag garbage or recycling containers to inform customers of problems, such as excess garbage or inclusion of items that are not permitted.Refuse and Recyclable Material Collectors · supplemental task | 6.5% | L0 | L0L1 | 9% | 2 min | Physical presence, Access to systems |
| Clean trucks or compactor bodies after routes have been completed.Refuse and Recyclable Material Collectors | 6.0% | L0 | L0L0 | 4% | 30 min | Physical presence, Access to systems |
| Make special pickups of recyclable materials, such as food scraps, used oil, discarded computers, or other electronic items.Refuse and Recyclable Material Collectors · supplemental task | 4.8% | L1 | L1L2 | 20% | 38 min | Physical presence, Accountability |
| Organize schedules for refuse collection.Refuse and Recyclable Material Collectors · supplemental task | 4.2% | L2 | L2L2 | 68% | 1.8 h | Access to systems |
Level: consensus of the two models (the lower level when they disagree). Share: part of the task’s working time an agent could take over. Human time: minutes a competent person needs for one instance, as estimated by the models.
Exposure is not delegation
| Measure | Value | Percentile | Pct |
|---|---|---|---|
| ADI (this index) | 18.0% | 29 | |
| Eloundou et al. (2024), GPT exposure β | 0.167 | 29 | |
| Felten et al. (2021), AI occupational exposure | -1.283 | 10 | |
| Webb (2020), AI patent exposure | 53 | 44 | |
| Eisfeldt et al. (2026), generative AI exposure | 0.031 | 20 | |
| Anthropic Economic Index, observed use | 0 | 16 |
Earlier indices measure whether AI can touch a task; ADI measures how much of it an agent can carry out from a goal. – : the source does not cover this occupation.
Neighbouring occupations
| Occupation | ADI | % |
|---|---|---|
| Meter readers and vending-machine collectors 9623 | 39.2 | |
| Odd job persons 9622 | 23.8 | |
| Garbage and recycling collectors 9611 | 23.2 | |
| Sweepers and related labourers 9613 | 20.6 | |
| Messengers, package deliverers and luggage porters 9621 | 20.5 | |
| Elementary workers not elsewhere classified 9629 | 10.1 | |
| Water and firewood collectors 9624 | 9.2 |
Dashed line: refuse sorters (18.0%).
Espinal Maya, Cristian, and Santiago Jiménez Londoño. 2026. ADI v2026.1: Agent Delegability Index data, code and MCP server. Zenodo. https://doi.org/10.5281/zenodo.23552918. Record: Refuse sorters (ISCO-08 9612), https://balamlabs.org/adi/occupations/9612
@misc{espinal2026adidata,
title = {{ADI} v2026.1: Agent Delegability Index data, code and {MCP} server},
author = {Espinal Maya, Cristian and Jim{\'e}nez Londo{\~n}o, Santiago},
year = {2026},
publisher = {Zenodo},
version = {v2026.1},
doi = {10.5281/zenodo.23552918},
url = {https://doi.org/10.5281/zenodo.23552918},
note = {Record: Refuse sorters (ISCO-08 9612)}
}The same record as static JSON, with its citation: /adi/v2026.1/isco/9612.json.