内涵型人工智能及其伴生安全问题  

Connotative AI and Its Associated Safety Problems

作　　者：杨义先[1] 钮心忻[1] YANG Yixian;NIU Xinxin(Beijing University of Posts and Telecommunications,Beijing 100876,China)

机构地区：[1]北京邮电大学,北京100876

出　　处：《信息安全与通信保密》2023年第3期1-9,共9页Information Security and Communications Privacy

基　　金：国家重点研发计划资助项目(No.2020YFB1805403);国家自然科学基金项目(No.62032002);高等学校学科创新引智计划资助项目(No.B21049)。

摘　　要：人工智能的未来发展有两大趋势,一是继续向外扩展人类的外延,二是努力向内丰富人类的内涵,前者称为外延型AI,后者称为内涵型AI。与外延型AI的发展方向相反,内涵型AI意在借助现有的机器智能来提高人类自身的整体智能。在全面综述内涵型人工智能及其发展趋势的同时,从宏观、中观和微观角度分析了相关新技术和新理念的伴生安全问题,其中有些问题已有解决思路,有些问题还需在发展中不断探索解决方案。There are two major trends in the future development of artificial intelligence:one is to continue to expand the extension of human beings outward,and the other is to strive to enrich the connotation of human beings inward.The former is called extensional AI,and the latter is called connotative AI.Contrary to the development direction of extensional AI,connotative AI intends to improve the overall intelligence of humans themselves by leveraging existing machine intelligence.While comprehensively reviewing the connotative AI and its development trends,this paper analyzes the associated safety problems of relevant new technologies and concepts from macro,meso,and micro perspectives.Some of these problems have already been solved,and some of them still need to be continuously explored for solutions during development.

关 键 词：人工智能 脑机接口 意念控制 安全智能 信息安全 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TP393[自动化与计算机技术—计算机科学与技术]