In May, Starbucks let go of a worker after nine months on the job.
The worker was an AI.
The company had rolled out an AI counting system across more than 11,000 stores (opens in a new tab). Tablets with cameras scanned the shelves and counted stock, so baristas wouldn’t have to. On paper it removed one of the most tedious tasks in retail.
In practice the system kept getting the count wrong (opens in a new tab). It confused one type of milk for another. Miscounts turned into shortages. Staff ended up recounting what the AI had already counted, so the tool doubled the very task it was built to remove. In May an internal newsletter told stores it was being retired. People count the milk again. Reuters reported it from start to finish (opens in a new tab).
Notice what did not happen
No barista lost a job when the AI arrived. None was hired back when it left.
The AI never held a job. It held a task. When it failed at the task, the task went back to a person.
That is the honest answer
AI replaces tasks, not jobs. It takes the repetitive layer: the counting, the data entry, the first drafts (opens in a new tab), the sorting. The parts that need context, judgment, and the sense that something looks off stay with people.
At Starbucks, even the counting came back.
So your value shifts rather than disappears. Less of it sits in doing the work. More of it sits in knowing which work matters, and in catching what the machine gets wrong (opens in a new tab).
Your job isn’t going away. The people who learn to work alongside AI keep their jobs and lose the parts they never liked.
Most people would call that a fair trade.