I Told You to Build a Kill Switch. Here’s the Part I Skipped.
中文摘要
作者指出仅靠监控和“杀伤开关”不足以确保AI安全，并探讨了控制智能体时更复杂且被忽视的挑战。
English Summary
Beyond simple monitoring and "kill switches," the author explores the more complex and overlooked challenges of maintaining effective control over autonomous AI agents.
原文节选
Part one was the easy argument: watch your agents, and be able to stop them. Continue reading on Medium »