■ CRITICAL RISK ■ Retail
Yes, at scale — self-checkout, scan-and-go apps, and computer-vision stores are all aimed squarely at the register, and cashiering is one of the largest occupations in the economy, which makes it automation's biggest single target. The register survives longest where trust, age checks, and human patience are the actual product.
“Self-checkout was just the trailer. The movie is about to start.”
Our AI replacement risk score — how we score jobs
Cashiering is more layered than its reputation: scanning and bagging at speed, handling cash and cards accurately, applying coupons and price overrides, checking IDs for alcohol, defusing the customer whose card declined in front of a line, watching for skip-scanning and sweethearting, and absorbing eight hours of standing, repetition, and the public's moods. It employs millions — for decades it's been one of the most common jobs in America — which is exactly why so much engineering budget points at eliminating it.
The automation offensive runs on several fronts at once. Self-checkout shifted the scanning labor to customers with one attendant overseeing six lanes. Scan-as-you-shop apps and shelf-edge payment remove the checkout entirely. Computer-vision formats like Amazon's Just Walk Out aimed to delete even the app — though its stumbles (including reports that 'automated' stores leaned heavily on human video reviewers) show the last mile is harder than the demos. Meanwhile AI handles the adjacent work: theft detection at self-checkout, dynamic staffing, price management. Retailers don't need the futuristic version to cut headcount; ordinary self-checkout economics already did.
The persistent human core: age-restricted sales and alcohol laws requiring judgment, cash acceptance (still legally and socially necessary), theft deterrence — a staffed lane shrinks losses in ways cameras haven't matched — and the segment of shoppers, especially older ones, who choose stores partly for human contact. Some chains have even re-added staffed lanes after self-checkout shrink and customer irritation. So the job won't hit zero; it will keep thinning per store while the total transaction volume goes machine-first, which for a workforce of millions is an enormous quiet contraction. Hence 97.
Automatability: our editorial assessment of current and near-term AI capability
Already rolling through every supermarket you enter. Self-checkout is mature, scan-and-go is expanding, and per-store cashier counts have been thinning for years even as the occupation remains one of the country's largest — scale is why the disruption is measured in hundreds of thousands of positions this decade. Full cashierless formats keep hitting snags, but they don't need to succeed for the register count to keep falling. The staffed lane becomes the exception lane.
Not fully, but substantially. Age-restricted sales, cash handling, theft deterrence, and customers who want humans keep staffed lanes alive — some chains have even walked back over-aggressive self-checkout after shrink losses. But per-store cashier counts keep falling, and because cashiering is one of the largest occupations in the economy, even partial automation displaces enormous numbers. Fewer registers, not zero.
Official occupational projections have ranked cashiers among the largest projected job declines for years — not because stores close, but because each store needs fewer registers as self-checkout and app-based payment absorb volume. It's attrition-shaped: hours get cut, open positions go unfilled, and one attendant oversees lanes that used to employ six people.
It's still a fine first job — cash handling, reliability, and customer de-escalation are legitimately transferable — but it's a worse last job than ever. Use it as a twelve-to-twenty-four-month platform: collect a shift-lead title, learn the inventory and POS systems, and exit toward service desk, fulfillment, or supervisory work before the hours thin out.
Theft. Self-checkout shrink — accidental and deliberate — plus customer frustration led several chains to restore staffed lanes for larger baskets. That's a real speed bump, not a reversal: retailers are responding with AI camera monitoring and item-recognition at the kiosk rather than rehiring at scale. The correction refines the automation; it doesn't retire it.