Why the route matters more than the reading
The part of this job that software can take is the easiest part: look at a dial or display, write down the number, move on. Smart meters and advanced metering infrastructure send that number by themselves, and image-recognition tools can read a legacy dial from a photo. That is real task erosion, and it has been running for years through hardware, not chatbots.
What software does not do is walk the route. Meter readers find meters behind locked gates, in basements, under snow, and in yards with dogs. They inspect meters for damage, leaks, corrosion and signs of tampering. They connect and disconnect service, leave notices, and explain a strange bill to the person standing at the door. Those tasks are why the work has not gone quietly with the meters.
The job is shrinking all the same. The Bureau of Labor Statistics counts about 19,430 meter readers in US utilities, with median pay of $48,150, and projects employment falling 10.5% between 2025 and 2035 (BLS, 2025). Fewer routes, fewer entry-level openings, and more of the remaining work pushed toward inspection and field service: that is the honest shape of it, and it is why this job sits on our list of jobs expected to shrink.
What software captures, what it assists, and what stays with a person
Reading and recording consumption is the clearest automation target. Remote meters report usage on a schedule, and optical tools can pull a number off a mechanical dial without replacing the meter. Our task split puts a share of task time in the group AI can already handle (0%), and the headline coverage figure above the task list is scored the same way we score every job, explained on how we measure coverage.
Assistance is where most of the change shows up on a working day. Software flags readings that look wrong before a bill goes out, routes a reader to the stops that actually need a visit, and holds the history that tells you whether a jump in usage means a leak, a faulty register or a new hot tub. The share of task time in that assisted group is printed above (44%).
The rest stays with a person, and the chart shows how large that part is (56%). It is the physical and social work: getting access to a hard-to-reach meter, judging whether a seal has been cut, turning service on or off safely, and dealing with a customer who is angry, confused or simply not home. Our robotics estimate above marks the physical slice and the hardware class that would be needed to do it — mobile robots, not fixed arms — which is a far harder and costlier build than reading a digit.
What the evidence actually shows
There is no published head-to-head test of AI against meter readers in the field, so our parity grade for this job is D, and we give no parity number. Vendor accuracy claims for optical reading are not the same thing as a measured comparison, and we do not treat them as evidence.
What would settle it is specific: a dated trial across meter types and conditions — glass fogged, dial worn, display dead, enclosure flooded — reporting read accuracy, exception rates, revisit rates and billing disputes for automated capture against a trained reader doing the same routes. Access and inspection outcomes would need to be measured too, because that is where the human share of the work sits. Until something like that is published, the honest answer is that the reading task is well covered and the field task is untested. The grading scale, including what a D means, is set out in how we grade quality parity.
When the balance could tip
Most likely after 2036 (8 in 10 of our scenarios). The method behind that window, and what it does and does not claim, is described in how we estimate the replacement year.
Two things could pull it earlier. Funded smart meter rollouts finishing in a service territory remove the daily read in one step. And cheap retrofit modules that read legacy dials in place lower the cost of covering the meters a utility is not ready to replace, which is why our cost comparison above puts automated reading well under a staffed route.
Two things hold it back. Customers refuse or opt out of smart meters in some states, and an opt-out meter still needs a visit. And access is stubbornly physical: a meter in a locked basement, a crawlspace or a flooded pit needs a person, as do shutoffs, tamper investigations and leak checks. Replacing a working meter fleet is capital spending that moves on utility-commission timelines, not software release cycles.
What to do: if your territory has an active AMI program, ask now which crew keeps the inspection, connect and disconnect work once the reads go remote.
How to stay needed on the route
Lean into the tasks that stay with people. Inspection is first: spotting damage, corrosion, tampering and the usage pattern that means a leak. Service work is second: safe connects, disconnects and reconnects, and the paperwork and safety rules around them. Customer contact is third: handling the disputed bill, the refusal, the no-access property and the exception that software kicks out.
Two skills travel well from here. One is meter data work — reading AMI exception reports and knowing which flag needs a truck roll and which does not. The other is utility field safety and compliance: gas, water or electric basics, lockout rules and documentation that holds up in an audit.
If you want a next step rather than a side step, the nearest work by code and by day-to-day task is weighers, measurers, checkers and samplers. Field inspection skills carry into work as an energy auditor, and the licensed, hands-on end of utility work sits with electrical power-line installers and repairers. You can put any two of them side by side on our job comparison tool.
For the wider picture, this job sits in the material recording and dispatching family and in the utilities sector, where scores for related roles are listed together. How every figure on this page is built is set out in our scoring method.