摘要
提出问题并解决问题是科学发现和知识发展的关键,而目前关于人工智能与人的认知的关系的研究较少去探究两者在“提问”和问题应对方面的差异,因此有必要探讨“问题”与“算法”的联系。从科学问题的形成、应对直到形成理论的过程等来看,AI难以提出创造性的科学问题,而科学问题也难以单纯用算法解决。主要依据是:一些重要科学发现过程中存在假设“先于”观察的现象,以及问题形成和认知跃迁往往是非归纳、非算法的过程等等。因此,在科学发现的能力方面,基于算法的AI可能具有较大的局限。
Proposing and solving questions are pivotal to scientific discovery and knowledge development.Currently,there is limited exploration in the difference between artificial intelligence and human cognition in the realms of"questioning"and coping with problems.Therefore,it is necessary to exam-ine the connection between"questions"and"algorithms".Upon examining the processes such as the formation of scientific questions,coping mechanism,and theory development,it reveals that Al struggles to pose creative scientifc questions.Vice versa,scientific problems are challenging to resolve solely through algorithms.The primary basis for this lies in phenomena where assumptions precede observations in certain significant scientific discovery proces-ses,and the processes of problem formation and cognitive leaps are often non-inductive and non-algorithmic.Therefore,concerning the cabability for scientific discovery,algorithm-based Al may encounter significant limitations.
作者
王耀德
查唐龙
WANG Yao-de;ZHA Tang-long(Academy of Marxism,Jiangxi University of Finance and Economics,Nanchang 330013,China)
出处
《自然辩证法研究》
北大核心
2024年第1期52-58,共7页
Studies in Dialectics of Nature
基金
教育部人文社会科学研究项目(19YJA720015)
江西省社会科学基金项目(21KS05)
浙江省社科联研究项目(2023N143)。
关键词
科学问题
算法
人工智能
AI科学家
scientific problem
algorithm
artificial intelligence
Al scientist