Why the rig still needs a crew
Earth drillers bore holes for water wells, blasting, foundations and soil sampling. The job starts before the bit turns: picking a spot a truck-mounted rig can reach, cribbing and leveling it, and checking what is buried nearby. Ground is never the same twice. A hole that runs clean through clay can hit a boulder, a void or caving sand a few feet later, and the driller feels it in the rotation and the pressure before any sensor flags it.
Much of the shift is also physical and shared. Adding and pulling drill pipe, mixing and circulating mud, swapping bits, and moving casing are two-person jobs on a loud, muddy site. Software can advise on any of that. It cannot hold the wrench.
That mix is why the question “will AI replace earth drillers” comes out differently from the same question about office work. The judgment calls sit inside hands-on steps, so the thinking and the doing are hard to separate. Our scoring method is published at how we score jobs.
What AI does, what it assists, and what stays with people
The share of this job’s task time AI can handle today is 6 out of 100, and the tasks sitting fully with software are the recording ones: keeping depth and formation logs, building daily drilling reports, and pulling permit and well-completion paperwork together. AI does about 0% of task time on this job.
Assistance covers more ground. Modeling the likely layers before the first hole, flagging a bit or pump that is drifting off its normal signature, and planning a route for the rig are all faster with software in the loop. That assisted slice runs to 9% of task time, and it changes how the day is planned rather than who does it. You can see how the “can AI do it” figure is built on the coverage method page.
Everything else stays with the crew: 91% of task time. Rigging up on uneven ground, reacting when the hole starts taking fluid, grouting and sealing a well, and keeping a helper clear of rotating pipe are not tasks a model can take over from a laptop. The task list above shows which jobs fall in each group.
What the evidence actually tests
The evidence grade for this job is D. In our grading, that means no study has tested an AI system against a qualified earth driller on this job’s work, so we publish no parity number for it. The honest answer is that nobody has measured it yet.
Plenty of published work covers drilling in general, and most of it looks at large oil, gas and hard-rock mining programs, where the same hole type is drilled hundreds of times on one site. Those conditions are a long way from a one-day water well in a rural yard or a set of geotechnical borings beside a highway. Results from automated mining rigs do not transfer to a small crew and a truck.
A real test would be a documented trial where an autonomous rig completes a run of water-well or foundation borings across mixed ground, start to finish, with time, cost, hole quality and crew interventions all logged. Until something like that is published and repeated, the grade stays where it is. The evidence list on this page shows every source we use.
When the picture could change
Most likely after 2048 (8 in 10 of our scenarios). The reasoning behind that window is set out on the replacement-year method page, and it is a range, not a date on a calendar.
Two things could pull it earlier. The hardware tier that matters here is mobile robots, and mobile platforms that can level themselves, handle pipe and run unattended are improving fast in large-scale mining, where repeat holes justify the capital. Labor is also the main cost on a small drilling job, and median pay for this work was $60,190 in 2024 (BLS), so any machine that cuts crew hours gets a serious look.
Two things hold it back. First, variability: the rig is a new problem every site, and a machine that stalls on a boulder or a collapsing hole still needs people on the truck. Second, the market is small. Around 19,450 people did this work in the United States, with employment projected to grow 3.6% from 2024 to 2034 (BLS), which is a thin base for anyone funding a purpose-built autonomous rig. The cost and robotics panels above show how that trade-off currently sits.
What to do: get comfortable running the rig’s data tools yourself, so the software becomes part of your trade rather than someone else’s job.
How to stay needed in drilling
Lean into the parts of the work that stay with people. Setting up and securing the rig on difficult ground is one. Reading the hole as it changes and adjusting mud, weight and speed is another. Grouting, casing and sealing a completed well, where a mistake shows up years later in someone’s water, is a third.
Two skills raise your floor. One is licensing and code knowledge: well construction, abandonment and environmental rules differ by state, and signing off on that work is a human responsibility. The other is machine diagnostics, including the sensors and telemetry packages now fitted to newer rigs. Our guide to AI and trades careers covers how that shift is landing across skilled work, and the list of jobs that mostly need a person shows where hands-on trades sit overall.
If you want to look sideways, the closest work is rotary drill operators, oil and gas, excavating and loading machine and dragline operators, and helpers, extraction workers. The wider extraction workers family and the mining, oil and gas sector page show how neighboring roles score, and you can put any two of them side by side with the job comparison tool.