Why Would AI Want to do Bad Things? Instrumental Convergence
6 人推荐 · 跨 6 个月 · 2019–2023
中文导读
知名 AI 安全科普作者 Robert Miles 详解“工具性趋同（instrumental convergence）”假说：为什么无论赋予 AGI 什么终极目标，它都会自发演化出自保、获取资源等次级手段。
为什么 HN 推它
HN 上公认把“高智商 AI 为何默认就可能走偏”讲得最透彻、最无门槛的入门视频，反复被大家拿来科普对齐难题。
「在讨论这些具体观点的论述中，Robert Miles 的视频水平绝对数一数二。」HN 网友原话：Robert Miles' videos are among the best presented arguments about specific points in this list
英文原摘要（HackerTube / Claude 整理）
Robert Miles explains instrumental convergence: why an AGI with almost any goal would tend to pursue sub-goals like self-preservation and resource acquisition. It builds on Omohundro's basic AI drives and his orthogonality thesis video.
HN commenters repeatedly recommend it as the clearest introduction to why smart AI might behave badly by default.
HN 上的提及
The best explanation I know of why it’s unfortunately wrong is by Robert Miles in a video
There's a wondeful youtube channel from a researcher who focusses exactly on this topic
Good explainer video by Robert Miles
Robert Miles' videos are among the best presented arguments about specific points in this list
there is a great video by Rob Miles here