1965年12月,休伯特·德雷福斯以兰德公司顾问的身份,发表了编号为P-3244的《人工智能与炼金术》的研究报告,对兰德公司本身主导的人工智能(以下简称AI)研究提出了重大理论挑战, 1972年德雷福斯以该报告为基础出版了《计算机不能做什么——人工智能的极限》,该书与1966年美国国家科学院的ALPAC报告,1973年英国科学研究理事会的LightHill报告一起, 标志着AI发展历史上的第一次冬天,即使经历了上世纪80年代由于专家系统兴起的AI再次繁荣,以及90年代初AI的第二次冬天,AI的研究纲领已经变化甚多,但德雷福斯仍然坚持其基本观点,对隐含在AI研究纲领中的关于人类认知和问题解决能力的深层假设,从现象学和海德格尔哲学为核心的大陆哲学立场出发,始终进行批判性地思考和分析,无论AI科学家共同体对其观点是否认同,德雷福斯这些深刻的哲学思考,客观上推动了从AI研究早期基于知识主义、符号主义强纲领的盲目乐观,到目前对实现人类级别智能的智能机器建造的审慎态度,以及更加丰富的研究进路的转变.%In 1965,Huber Dreyfus as a consultant in Rand Corporation ,presented a proposal titled artificial intelligence and alchemy, which numbered as P-3244 challenged research on Artificial Intelligence (AI) sponsored by Rand Corporation himself in theory greatly. In 1972,Dreyfus published 《What computers can't do: The Limits of Artificial Intelligence 》based on this proposal, which book and other two reports: ALPAC report presented by National Academy of Sciences, LightHill report presented by Science Research Council, are the beginnings of first AI winter. Even after the AI revival in 1980s because of emergence of Expert System, and the second AI winter in early 1990s, the research programme of AI had changed greatly, Dreyfus still stand his ground firmly, he always thinks and analyzes the assumptions on human recognition and ability of problem-solving behind the AI research programme implicitly in a critical way, his in-depth review about AI comes from the standpoint of Phenomenology and Heidegger's which is the core of continental philosophy. No matter what AI scientific community agree with him or not, the meaningful thoughts of Dreyfus really had driven such changes in AI field:from the blind optimism based on intellectualism and Symbolism strong programme in early stage to the prudent view about building intelligent machines at the human-level intelligence at present and more diversified research programme today.
展开▼
