Why the register still has a person behind it
Self-checkout has been in stores for more than twenty years, and the job is still here. About 3,089,410 people work as cashiers in the US, at a median wage of $32,880 (BLS, 2025). So when people ask will AI replace cashiers, the honest answer starts with what the machines already took: the scanning and the payment capture. That was the easy part.
The rest of a shift is messier. A card declines. A coupon refuses to apply. A customer wants beer and the ID looks wrong. Someone returns a shirt with no receipt and an argument ready. Those are judgment calls with money, state law and a waiting line attached. A machine can flag the problem. Someone has to settle it.
There is also everything around the lane: counting the drawer at the start and end of a shift, watching for theft, keeping the checkout area stocked and clean, pointing a lost shopper to the right aisle. Most of that is physical work in a crowded space with people moving through it. Our robotics view places the bulk of it in the mobile robot tier, meaning hardware that has to move around a store rather than sit bolted to a counter. That is slower and more expensive to deploy than software. You can see how the physical share is calculated in our scoring method.
What machines do, what they assist, and what lands on staff
The machine group is the mechanical core of the job: reading barcodes, totaling a basket, taking card and mobile payment, printing a receipt, applying a discount the system already knows about. Vision-based checkout can also keep a running total as items go in a bag. That group accounts for 8%.
The assisted group is where software speeds up a person instead of removing one. Prompts flag a price mismatch before it becomes a dispute. A terminal walks a new hire through a void or a split payment. Inventory lookups answer “do you have this in a medium?” in seconds. Scheduling and drawer reconciliation run in the background. Shared work like that comes to 28%.
The human group is the exceptions and the people. Resolving a complaint without losing the customer. Checking ID for age-restricted sales and standing behind that call. Spotting a scan that was never going to happen. Helping an older shopper who will not use a kiosk. Training the person on lane four. Work in that group makes up 64%. That is also why staff-light stores still have attendants pacing the self-checkout bank.
Good to know: self-checkout moves the labor rather than deleting it, because one attendant now covers six lanes instead of one.
What has actually been measured
Very little, head to head. The parity grade on this page is D, and that grade means no direct test of AI against working cashiers has been published and graded. So we give no parity number for this job, and you should treat any site that prints one as a guess.
What would settle it is a study of matched stores that compares staffed lanes and automated checkout on the same measures: throughput at peak, scan and pricing errors, shrink, age-verification failures, and how often an exception needs a staff member anyway. Until something like that exists, the coverage figure is the more useful number, because it is built from task time rather than from opinion. Our coverage method explains how that share is put together.
One outside figure is solid. BLS projects cashier employment to fall 6.5% between 2025 and 2035 (BLS, 2025). That is a real decline across a very large occupation, and it lands hardest on new hires. It is not the same as the job disappearing.
When the balance could shift
Most likely after 2036 (8 in 10 of our scenarios). What that window measures is set out on the replacement-year page.
Two things could pull it earlier. Camera-and-sensor checkout keeps getting cheaper per lane, and the software side of this job is among the lowest-cost automation we track. Large chains also already run formats with few staffed lanes, so the retail model exists and only needs copying.
Two things push it later. Age-restricted sales, returns and payment disputes carry legal and cash risk that stores keep with a named employee. And shrink has pushed several large retailers to add staff back to checkout rather than remove them. Retrofitting older stores costs real money per location, which stretches rollouts over years. The wider picture for the sector is on our retail sector page.
How to stay needed at the front end
Lean into the parts of the shift that the task list above puts in the human group. First, exceptions: returns without receipts, disputed charges, split tenders, price adjustments. Second, verification and loss prevention: age checks, suspicious-transaction judgment, knowing the store policy cold. Third, service recovery: turning a frustrated customer into one who comes back.
Two skills travel well from here. One is point-of-sale and self-checkout troubleshooting, including attendant work across a bank of kiosks, because that role is growing while single-lane work shrinks. The other is de-escalation, which is the skill supervisors hire for when they promote from the registers.
What to do: ask your manager for the self-checkout attendant rotation and the returns desk, since both sit in the part of the job that keeps needing a person.
Nearby work worth comparing: counter and rental clerks, gambling change persons and booth cashiers, and retail salespersons, which leans more on product knowledge and less on the drawer. The supervisor track sits in the same retail sales workers family. You can put any two of these side by side with our job comparison tool, or see where checkout work sits among jobs expected to shrink.