Why hand sorting stays with people
Will AI replace reclamation workers? Start with the shift itself. The work is hands on moving, mixed, filthy material: pulling contaminants off a fast conveyor, breaking down appliances, stripping wire and copper, and clearing a jam before a baler chokes. Software can read a stream of material from a camera. It cannot climb into the pit with a pry bar.
Our robotics read puts about 92.5% of this job’s work in the physical column, and the machine class it would take is mobile robots rather than a fixed arm bolted over one belt. That is the hard part. A sorting cell can sit above a single line and pick a single material. A worker walks the floor, opens a door, reaches behind a motor, and decides whether the refrigerant line, the propane cylinder or the swollen lithium battery in front of them is safe to touch. Each load arrives different. The judgment is not a rule you can write down once.
Pay shapes the math too. Median pay in this occupation is $40,240 a year (BLS, 2025), against the capital cost of robotic cells, grippers that survive glass and grit, and the facility rebuild they often need. A materials recovery facility can justify that spending on one high-volume stream. It rarely justifies it for dismantling, salvage and cleanup work that changes hour to hour. You can see how that logic scores on our coverage method page, which measures the share of task time AI can handle today: this job sits at 1 out of 100 on that scale.
What AI does, what it assists, what people keep
No task in this job’s list sits in the AI-does-it group yet. Optical sorters and robotic pickers do real work on single-stream belts, but they take a slice of a line’s throughput rather than a worker’s full task. The automation share our data records for the job is 0%.
The assisted group is empty as well. Nothing in the list is marked as work a person does faster with an AI tool at their elbow, which puts 0% in that column. Camera systems that flag contamination are getting better, and that is where assistance is most likely to show up first.
That leaves the rest: 100% of task time in the needs-a-human group. Sorting and grading mixed material by sight and feel is there. So is dismantling appliances and removing the hazardous parts, clearing jams and blockages on the line, operating forklifts and balers, cleaning equipment, and weighing and recording loads. If you want to see how that pattern compares with the collection side of the industry, put this job next to another on the job comparison tool.
What the evidence actually shows
There is no direct head-to-head test of AI against people in this job. Our quality parity grade is D, which is the grade we use when nothing has measured machine output against a qualified worker’s, so we publish no parity number for it. We would rather say that plainly than guess.
What would settle it is specific and measurable: published pick rates and accuracy for robotic sorters on mixed municipal streams, error and contamination figures from named materials recovery facilities, and uptime data showing how often a cell stops for a jam a person then clears. Dismantling and hazardous-component removal would need its own trial, because the risk sits in the parts nobody photographed in advance. Our grading scale and how each claim is weighted are set out in the scoring methodology.
On the labor-market side there is firmer ground. BLS counts 2,950,280 US workers in this occupation’s reporting group and projects employment change of 1.8% over 2025 to 2035 (BLS, 2025). That is slow growth, not a cliff.
When this could shift
Most likely after 2044 (8 in 10 of our scenarios). What that range measures, and why it is a window rather than a date, is explained on the replacement-year method page.
Two things could pull it earlier. First, cheaper general-purpose mobile robots with hands that tolerate glass, grit and sharp edges, which is the shift tracked in our guide to humanoid robots and physical jobs. Second, facility design: new build materials recovery plants can engineer the line around machines, something a retrofit cannot.
Two things hold it back. Hazard handling is the big one, because refrigerants, pressurized cylinders and damaged batteries carry legal and safety duties that sit with a named person. The other is cost and variability together: unsorted, contaminated, unpredictable feedstock breaks grippers and triggers stoppages, and every stoppage needs someone on the floor. Adoption across transportation and warehousing tends to start where volume is high and material is uniform, which is the opposite of salvage work.
How to stay needed on the floor
Lean into the parts of the job that machines stumble over. Hazardous-component removal is first: appliance refrigerants, batteries and anything pressurized. Jam clearing and basic line maintenance is second, because a plant with robots needs that skill more, not less. Quality control on grading and bale contamination is third, since a buyer rejecting a load costs real money.
What to do: add a forklift certification and hazmat handling training, then learn to tend and troubleshoot sorting equipment rather than only feed it.
Close neighbors worth a look are refuse and recyclable material collectors, machine feeders and offbearers and hazardous materials removal workers, which pays more for the same hazard skills. The wider material moving workers family shows how the scores spread across related roles, and the list of jobs that mostly need a person shows which other hands-on work sits in our top band (Nah.).