ATMs automated the core teller task, yet teller jobs survived — what does that history say about AI and jobs today?

This explores why automating a core task, like ATMs handing out cash, didn't wipe out the bank teller job, and what that history suggests about AI and work today. First, a caveat: the collection has no paper on ATMs or tellers. The well-known version of the story comes from outside it. ATMs made each branch cheaper to run, so banks opened more branches, and tellers moved from counting cash toward customer service and selling. What the collection does have is a set of findings that explain how that kind of survival works, and when it doesn't.

The clearest one: it matters how much of a job the machine touches, not just whether it touches it. Firm-level data from 2010 to 2023 shows that when AI exposure is concentrated in a few tasks, workers shift to the tasks that are left, and the net job loss stays modest Does concentrated AI exposure enable workers to adapt and reallocate?. Dispensing cash was one slice of a teller's day, which is exactly that pattern. The same thing turns up in more recent cases. Companies using generative AI kept their headcount and moved people from doing the work to supervising it Does generative AI adoption actually eliminate jobs or reshape them?. A shipbroking firm turned its manual classification work into checking and correcting the algorithm Does automation eliminate manual work or repurpose it?. Most skills carry over into the reshaped job instead of disappearing Will most worker skills survive AI automation?.

The tellers' new role points to a deeper idea. One argument holds that the safest jobs aren't the most complex ones. They're the ones where people pay for a particular person, not just a result Which jobs will actually survive automation by AI?. The ATM took the transaction, and what was left for tellers was the relationship. Banking has a modern sequel, too. Commonwealth Bank, Klarna and IBM cut customer-service staff expecting AI to do the work, then hired people back when it couldn't Do banks accurately assess whether AI can replace customer service jobs?. Employers tend to overestimate how much of a job a machine can actually take.

Here's what you might not expect: survival in aggregate can hide damage at the entry point. Payroll data shows no economy-wide job losses from AI. But young workers in exposed occupations are being hired at 19% lower rates, while experienced workers see no such gap Is generative AI displacing workers at economy-wide scale?. If AI takes over the routine work juniors used to learn on, it can also erode the training that produces the experts who will later supervise the AI How does automating junior work destroy verification oversight?. So 'the job survived' and 'the career path survived' are separate questions. The ATM story is mostly about the first one.