Short answer: the hero above gives the verdict for this job, and the reason is the work itself. Loading and moving ore underground means driving a scoop, shuttle car, or load-haul-dump machine through tight headings, over broken floor, under a roof that changes shift by shift. Software can plan a haul route. It still needs a person in the seat or on the remote to handle the heading in front of them. So when people ask whether AI will replace moving machine operators, the honest answer here is task erosion, not a job disappearing.
Why the seat stays with a person
This job is built on reading ground. Operators position the machine by feel, back out of a blind heading, and judge whether a rib looks solid before rolling under it. Prying loose rock from the roof and ribs before loading is not a scripted step. It is a decision made from what you see, hear, and smell in that one place, at that one hour.
The second half of the job is keeping the machine running. Operators check equipment for defects, grease and oil moving parts, free a jammed bucket, move cables clear of the tires, and signal other crew when sight lines vanish. Each of those tasks happens in dust, water, and noise, in a space built for coal or ore rather than for sensors.
Underground conditions also strip away what most autonomous systems lean on. There is no satellite positioning below ground. Radio links drop. Mud coats cameras and lidar within a shift. Those are the practical blockers listed above, and they are the reason the task split leans the way it does rather than any claim that heavy machines cannot drive themselves. On haul loops that repeat without change, they already can.
What AI runs, what it assists, and what the crew keeps
Work that software handles on its own is narrow here. The share AI does alone, from the split above: 0%. It sits with the paperwork side of the shift, such as logging tons moved and machine hours, and flagging a fault code from onboard monitoring before anyone walks the machine down.
Assistance is where the real change shows. The assisted share: 3%. Proximity detection and collision avoidance help an operator place a machine near a rib or another worker. Teleremote control lets a loader be driven from a safe spot outside the heading, with software smoothing the tramming between muck pile and dump point. The operator still decides when to load, when to stop, and when the ground is wrong.
The rest belongs to people. Human-only task time, from the split: 97%. That is the hands-on block: scaling loose rock, spotting a bad roof, repairing and lubricating the machine mid-shift, and working with the face crew when the plan changes. How that share rolls up into the headline figure is set out on the Still needs a human page, and the whole scoring approach sits in our published method.
What the evidence actually shows
There is no direct head-to-head test of an AI system against a trained underground loader operator on this job’s tasks. That is why the evidence grade above reads D, our lowest confidence tier, and why no quality-parity number is given. We do not put a figure on something no one has measured. The way we grade capability against a working professional is explained under Is it better than a person?, and what the capability share means under Can AI do it?
A test that would settle it is easy to describe. Run an autonomous or teleremote loader and a qualified operator on the same production heading across full shifts. Measure tons moved, machine damage, unplanned stops, and how often a person had to walk in to recover the machine. Until something like that is published and repeatable in a working mine rather than a demonstration drift, the honest position is uncertainty.
The market facts around the job are firmer. The Bureau of Labor Statistics counts about 5,930 of these operators in the United States, with median pay of $74,500 a year, and projects employment down 15.8% between 2025 and 2035 (BLS, 2025). That decline is driven by mine closures, ore demand, and consolidation as much as by anything a machine learns to do.
When the picture could change
Most likely after 2048 (8 in 10 of our scenarios). What that window measures, and how it is built, is set out under When could it be replaced?
Two things could pull it earlier. First, the robotics side of this job is dominated by mobile machines rather than fixed arms, and mobile mining equipment is already the most automated hardware class in the industry; progress on underground navigation without satellite positioning transfers straight across. Second, the hourly cost gap shown above is wide, so any operation running a stable, repeating haul loop has a clear reason to fit automation kits to the fleet.
Two things hold it back. Capital cycles in mining are long, and a loader bought today may run for a decade before it is replaced. And the recovery problem is stubborn: when an automated machine gets stuck, buried, or damaged in a confined heading, a person still goes in to fix it, so crews cannot shrink as fast as the equipment catalog suggests.
Good to know: automation in underground mining has mostly moved people away from the face rather than off the payroll, turning seat time into remote and supervisory time.
How to stay needed
Lean into the parts of the shift that stay human. Ground awareness comes first: scaling and assessing roof and rib conditions is a judgment call that carries the crew’s safety with it. Second, machine care, because the operator who diagnoses a hydraulic or drive fault and fixes it underground keeps production moving. Third, crew coordination at the face, where signals and timing between loading, bolting, and haulage decide the shift.
Two skills are worth adding now. One is teleremote and semi-autonomous operation, including the setup, exclusion zones, and handover steps that come with it. The other is reading machine data: fault logs, production dashboards, and condition monitoring, so you can tell a supervisor what the numbers mean before a breakdown.
Close jobs worth comparing are continuous mining machine operators, roof bolters, and surface mining excavating and loading operators, where remote and autonomous equipment is further along. You can also see the wider picture on the extraction workers family page, the mining, oil and gas sector page, and our list of jobs that mostly need a person (our top band, Nah.). To weigh two paths side by side, put this job against another in the compare tool.