EXPOSED
The work is physical — loading hoppers, adjusting feed rates and screen settings on crushers and mills, clearing jams, changing grinding wheels and liners, sampling product for particle size — and language AI does none of that. The real threat is conventional plant automation: sensor-driven feed control, automated screening, and vision-based quality checks already run mills with fewer tenders, and AI makes those control loops better. Monitoring gauges, logging output, and adjusting to spec are the parts most easily absorbed; no license protects the role and buyers of aggregate or ground pigment do not pay for a human relationship.
Mixed — a routine tier and a judgment tier. At 13 the split is real: PLC loops and belt scales already handle feed rate, amperage trending, and screen deck monitoring on modern crushing circuits, but nobody has automated unbolting a worn jaw plate, chipping out a bridged rock in a cone crusher, or re-dressing a grinding wheel on an older mill, which keeps this above the low end rather than at 8.
Hands-on in uncontrolled environments. 15 reflects a shift spent in dust, noise, and vibration walking crusher decks and mill floors — lockout/tagout before entering a chute, sledging jams free, hauling liner segments and screen cloth, hosing down fines — in environments where the material itself is unpredictable and the plant layout is not a controlled cell.
No licence, no signature requirement. 2 is nearly the floor because no state licenses a crusher operator; MSHA Part 46/48 or OSHA hazard training is required of the employer, not credentialed to you personally, so a plant can put a new hire on the same panel after a week of training.
Anonymous artifact production. 3 acknowledges that the reader may be the person the foreman calls when the tertiary cone starts surging, but the customer buying #57 stone or ground calcium carbonate sees only a gradation certificate and never learns which operator ran the shift.
Executes defined procedures on defined inputs. 6 sits at the top of the procedural band because the calls you make — tightening the closed-side setting when the gradation drifts coarse, slowing the feeder before the mill chokes, pulling a sample for sieve analysis — are real decisions but measured against a written spec and a QC lab that overrides you.
Roof Bolters, Mining EXPOSED
Has AI actually changed your work?