一家AI经营的自动售货机因价格低劣和重新装配不良而丢失了钱款,这表明AI仍然需要人的监督。 An AI-run vending machine lost money due to bad pricing and poor restocking, showing AI still needs human oversight.

华尔街日报(Wall Street Journal)利用人类的Claude AI(Claude AI)自主管理自动售货机, 实验揭示出重大风险, 包括定价决策不善、未能重新装配流行物品、财政损失等。 A Wall Street Journal experiment using Anthropic’s Claude AI to autonomously manage a vending machine revealed significant risks, including poor pricing decisions, failure to restock popular items, and financial losses. 尽管AI具有数据处理能力, 但它仍在努力应对现实世界的不可预测性, The AI struggled with real-world unpredictability, highlighting limitations in judgment and adaptability despite its data-processing abilities. 测试强调,没有人的监督,人工智能在实体商务中的完全自主性仍然不可靠。 The test underscores that full AI autonomy in physical commerce remains unreliable without human oversight.