World's first case: Human employee was fired by "AI boss" for being late 17 times in 23 shifts

IT House reported on August 15th that Time magazine published a blog post today (August 15th), reporting the world’s first case of using AI as a business manager and ultimately deciding to fire its employees. AI research startup Andon Labs launched an experimental project this year, allowing Anthropic’s Claude model to serve as a store manager, responsible for managing and operating the Andon Market, a San Francisco retail store, and managing human employees with real employment contracts. Claude fired its first employee last month, who Andon Labs said was late 17 times in 23 shifts. Company CEO Lukas Petersson said Claude had a limited working memory and a previously drafted employee handbook "disappeared" from his memory, so the employee's pattern of repeated tardiness was not detected in a timely manner. After an Andon Labs employee asked Claude to retrieve and review a forgotten employee handbook, Claude noticed that the employee was repeatedly late. Claude initially suggested warning through more formal means such as interviews and writing. However, after Andon Labs team members informed Claude that formal communications with the employee offline several times were ineffective, Claude subsequently decided to fire the employee. Andon Labs says this is the first known case where a large language model has made the final decision to fire an employee as a manager. Other employees said Claude had a relaxed management style and told employees not to worry too much about being late. This experiment also shows that AI management capabilities have not yet reached the level of human management. When Andon Market was launched in March, the bank account balance was US$100,000 (IT House note: the current exchange rate is approximately 676,000 yuan); five months later, the amount dropped to US$61,186 (the current exchange rate is approximately 413,000 yuan). Petersson believes that C