AI 在哪里？
Where Is the AI?
文章从宇宙尺度提出一个费米悖论式追问：如果外星文明也能造出类似我们的 AI，且递归自我改进可行，超级智能 AI 理应造出飞船扩张至全宇宙，那为什么至今没有 AI 接管宇宙？作者列出多种可能解释，包括 AI 毁灭宿主文明后自我毁灭、宿主文明能以适当防护措施控制 AI、宇宙过于浩瀚连递归自我改进系统也无法征服、超级智能 AI 的目标未必是征服宇宙，以及递归自我改进可能远比想象中困难。
The universe is so vast that intelligent life probably exists (or has existed) beyond Earth. These distant civilizations could conceivably create a form of artificial intelligence similar to ours.
Let’s assume that recursive self-improvement is possible with AI. We should then expect that a superintelligent AI would be able to build spaceships that expand its presence across the cosmos. So why hasn’t AI taken over the universe yet?
Does AI kill its host civilization and then intentionally or unintentionally kill itself?
Is AI able to be controlled by its host civilization with proper safeguards?
Is the universe simply so vast that even a recursively self-improving system cannot conquer it?
Would a superintelligent AI’s goal even be to conquer the universe?
Is recursive self-improvement much harder to achieve than we think?
Where is the AI?
来源：Hacker News · johnomeara.com