以人工智慧应对人工智能威胁.PDFVIP

40 4 236 自 然 辩 证 法 通 讯 Vol 40 No 4 (Serial No 236) 第 卷 第 期（总 期）　 　　　 　 　　　 . , . . 2018 年 4 月　　　　　　　　　Journal of Dialectics of Nature　　　　　　 　　April 2018 以人工智慧应对人工智能的威胁 Developing Artificial Wisdom to Deal With the Threat of Artificial Intelligence 汪凤炎 /WANG Fengyan　魏新东 /WEI Xindong 210097 （南京师范大学道德教育研究所暨心理学院，江苏南京， ） (Institute of Moral Education Research, School of Psychology, Nanjing Normal University, Nanjing, Jiangsu, 210097) 摘　要：近年来人工智能的飞速发展引发了人们对其安全问题的关注与讨论。现有研究更多地从预防 强人工智能、引导弱人工智能的角度提出应对方案，却很少有直面强人工智能的有效方案。本文从“智 慧的德才一体理论”中获得灵感，主张将人工智能升级为人工智慧来应对这一威胁，以图灵智慧测验来 检测是否达到人工智慧，并将人工智慧分为弱人工智慧与强人工智慧两种类型。最后探讨了人工智慧实 现的具体路径。 关键词： 智慧 人工智能 人工智慧 图灵智慧测验 Abstract: In recent years, with the rapid development of artificial intelligence (AI), its security problems have attracted many people’s attentions and provoked much discussions. However, more attention, in existing studies, was paid to preventing strong AI or guiding weak AI rather than directly facing the real threat from strong AI. The latest theory of wisdom of integration of morality and cleverness has shed light on this problem by advocating that AI should be upgraded as artificial wisdom. Only in the way of equipping such software with the corresponding hardware, and making it pass the Turing wisdom test successfully, can the AI be upgraded as artificial wisdom with the capability of integrating morality and cleverness. There are two types of artificial wisdom. Weak artificial wisdom is controlled by human beings, while strong artificial wisdom is not. According to this approach, specific ways of turning AI into artificial wisdom are discussed.