For the first time, an AI fired a human employee after becoming the boss of a real store
Artificial intelligence has moved beyond writing emails and answering questions. In a real-world experiment in San Francisco, an AI system was given responsibility for running a retail business, including managing human workers.
The project was created by Andon Labs, which handed control of its Andon Market store to an autonomous AI agent powered by Claude. The system handled business decisions, including inventory, schedules and staffing.
One of the most striking moments came when the AI examined an employee’s attendance record. The worker had reportedly arrived late for 17 of his 23 shifts.
After reviewing the information, the AI was directly asked whether the employee should remain necessary for the business. It decided that the worker should be dismissed.
The incident became a notable example of an AI making a decision normally associated with a human manager. However, the experiment was not a typical workplace where a person’s livelihood depended entirely on an algorithm.
Andon Labs has stated that the people working at Andon Market remain formally employed by the company and receive normal legal protections and guaranteed pay. The AI can make management decisions inside the experiment, but the project operates under human oversight and safeguards.
The experiment has been running for about five months. During that period, the AI-controlled business reportedly lost nearly $40,000 because of poor commercial decisions and operational mistakes.
Despite those failures, Andon Labs believes the experiment shows how quickly autonomous AI systems are moving into real business environments. The company has already used AI agents to manage retail stores, vending businesses and other commercial projects. :contentReference[oaicite:1]{index=1}
According to Andon Labs CEO Lukas Petersson, current AI models still make serious mistakes. Yet their capabilities are improving rapidly. That raises the possibility that AI systems could eventually play a much larger role in hiring, scheduling and managing human workers.
For now, an AI firing decision remains part of a controlled experiment. But the case offers a glimpse of a possible future where an AI boss is no longer science fiction and algorithms begin making decisions once reserved for human managers.
Source: TIME.
It was previously reported that scientists trained artificial intelligence to identify various human illnesses from photographs. This AI-powered virtual doctor can already make 98% correct diagnoses.
