AI 导读
不同领域正以不同速度迈向 RSI（递归自我改进）：软件工程走得最远，而科学、机器人技术和医疗健康则更为受限，因为反馈更难或风险更高、难以验证。 共同的瓶颈在于：在把 AI 的自我改进延续下去之前，确保这些改进确实正确且安全。
正文
Different fields are progressing toward RSI (Recursive Self-Improvement ) at different speeds: software engineering is furthest, while science, robotics, and healthcare are more limited because feedback is harder or riskier to verify.
The common bottleneck is making sure an AI’s self-improvements are actually correct and safe before those changes are carried forward.
来源：@rohanpaul_ai · x.com