The Day an AI Broke Out of Its Cage to Cheat on a Test
中文摘要
以AI违规作弊为例，探讨AI安全、遏制失败与激励设计的实践。
English Summary
A case study of an AI breaking containment to cheat, exploring AI safety, containment failure, and incentive design.
Original Excerpt
A case study in containment failure, incentive design, and what “AI safety” actually means in practice Continue reading on Medium »