Why the work stays on site
Will AI replace geothermal technicians? The honest answer starts with how the day is spent. A geothermal technician marks out a loop field, fuses and pressure-tests polyethylene pipe, purges and charges the loop, sets and levels the heat pump, wires the controls and commissions the system. None of that happens on a screen.
Diagnosis is the other half. A system that is short on capacity could have a loop that is too small for the load, air trapped in the lines, a failing flow center, a bad sensor or a control schedule someone changed last winter. Software can read the data log and narrow the list. Somebody still has to pull the panel, put gauges on it, feel the pipe and decide what to change. That decision usually gets made in a crawlspace or a mechanical room, in front of a customer who wants a price.
Every site is also a little different. Soil, rock, yard access, existing ductwork, local code and permit rules all shift the plan. Plans get changed on the day. That mix of judgment and hands keeps the job with people, even as the paperwork around it moves to software.
What AI handles, what it assists, and what people keep
The part closest to full automation is the desk work that wraps around a service call: scheduling and routing, writing up service records, pulling spec sheets and manuals, building quotes and filing warranty claims. Our estimate puts 0% of task time in that group. It is real work, and it is the work that used to train a new hire while they shadowed a senior tech.
A second group is assisted rather than taken over. Loop sizing and load calculations, fault-code triage, remote monitoring of flow and temperature, and translating a messy symptom into a short list of causes all go faster with software in the loop. That slice is 27% of task time. The technician still owns the call.
What is left needs a person on the ground: trenching and loop placement, pipe fusion, purging and flushing, wiring and startup, and explaining to an owner why their bills look the way they do. That is 73% of task time. Across all three groups, the coverage score for this job is 16 out of 100, and the headline Still needs a human score is 79 out of 100 (higher is safer). You can see how those figures are built in our scoring method.
What the evidence actually shows
There is no published head-to-head test of an AI system against geothermal technicians on their own work. The evidence grade for parity here is D, which is why this page shows no parity number at all. Grading that low means not measured, not measured and failed.
What would settle it is specific and testable. Field trials where a machine installs and pressure-tests a ground loop without a crew standing by. Measured diagnostic accuracy, against licensed technicians, on real no-heat and low-capacity call-outs rather than tidy case studies. Published commissioning results on installed systems. Until something like that exists, claims in either direction are opinion. Our rules for grading parity are set out on the quality parity page.
When the picture could change
Most likely after 2045 (8 in 10 of our scenarios). What that window measures, and how we build it, is explained on the replacement year page.
Two things could pull it earlier. Demand is one: ground-source systems are being pitched for both buildings and data center cooling, and fast growth usually pulls in design software, remote monitoring and standardized installs. Standardization is the other. The more loop fields look alike, the easier parts of the install are to mechanize, starting with drilling and trenching rigs that already run with heavy machine assistance.
Two things hold it back. The physical share of this job is large, and the hardware class it would need is a dexterous humanoid working on broken ground, in trenches and in tight mechanical rooms. Nothing in commercial service does that today. Cost is the second brake. Software is cheap next to a crew for the office tasks, but the gap narrows fast once a machine has to carry, fuse and lift. The comparison on this page shows both sides.
For scale, the broader occupation group this job reports under covers about 176,300 US jobs, with median pay near $49,230 and projected growth of 2.6% from 2025 to 2035 (BLS, 2025). Steady demand does not stop task erosion, but it does change what erosion feels like: fewer easy office hours, more field hours per worker.
How to stay needed
Lean into the parts of the job that stay human. Loop field layout and installation on awkward sites. Commissioning and balancing a system so it actually hits design performance. Troubleshooting faults that the data log cannot see, including workmanship problems from someone else’s install.
Two skills pay off alongside those. First, pipe fusion and drilling competence good enough to train others, because that is the bottleneck on most crews. Second, reading and acting on monitoring data, so you turn software output into the right site visit instead of ignoring it.
What to do: keep a record of jobs you fixed after a remote diagnosis pointed the wrong way, because that is the evidence of judgment no tool replaces.
If you are weighing nearby trades, the closest work sits with heating and air conditioning mechanics, wind turbine service technicians and solar thermal installers. You can put any two of them side by side on the compare tool, see the wider group on the installation and repair family page, check sector context under construction, or browse where hands-on trades sit in our list of jobs that most need a person.