出 处:《重庆邮电大学学报(社会科学版)》2024年第1期112-120,共9页Journal of Chongqing University of Posts and Telecommunications(Social Science Edition)
基 金:国家社科基金项目:大数据应用中的信息流动与伦理问题研究(19BXW116)。
摘 要:人类创造是人们通过构建关系之网以确证自身在世界之中的位置,塑造意义世界的基础,此种构建需要某种“超越”的外部之物作为支撑。目前,大语言模型已经在智能生成方面表现出巨大的创造潜力,甚至在部分原创性任务中可以取代人类创造,这或将对人类的意义世界造成潜在性威胁。究其症结在于,大语言模型的底层生成逻辑是建立在相关主义基础之上的,而这种基础预设在激进化演绎的同时不可避免地暴露出其深层局限。一方面,大语言模型以“相关主义循环”的方式将自身封闭于字面主义的符号表征与现有的固定范式之内,使得“超越”之维不再可能;另一方面,大语言模型的智能生成功能对于现实架构的渗透性影响又转而超越了相关主义的可理解范畴。从长远角度来看,这种深层矛盾必将冲击一直以来人类从单一主体视域出发建构而成的意义世界。人类何以突破相关主义的内在生成逻辑,维护自身意义世界,这是智能技术时代人类必须直面的问题。立足“物导向哲学”,实现对大语言模型生成逻辑的“超越”,其意义在于这种“超越”的双重性,既能更包容地将大语言模型类智能机器及其功能纳入“物”的可阐释范畴,又能以去中心化的方式突破相关主义束缚,借助“联合实在”的美学活动,通向超越性的外部,为人类创造添注活力,为人类意义世界提供支撑,以此开辟出“超越”大语言模型的意义生成之路。Human creation is the foundation for people to establish their own position in the world and shape the meaning of the world through constructing a network of relationships.This construction requires some kind of“transcendent”external object as a support.Currently,large language models have shown great creative potential in intelligent generation,and even can replace human creation in some original tasks,which may pose a potential threat to the human meaning world.The crux of the matter is that the underlying generation logic of large language models is based on correlationism,which inevitably exposes its deep limitations while simultaneously radicalizing deduction.On the one hand,large language models are enclosed in the symbol representation of literalism and existing fixed paradigms in a“correlative cycle”manner,making the dimension of“transcendence”impossible.On the other hand,the penetrating influence of large language models’intelligent generation function on the real structure has transcended the understandable scope of correlationism.In the long run,this deep contradiction will inevitably impact the meaning world that has been constructed by humans from a single subject perspective.How can humans break through the inherent generation logic of correlationism and maintain their own meaning world?This is a question that humans must face in the era of intelligent technology.Based on the“object-oriented philosophy”,realizing“transcendence”of large language model generation logic is of great significance.This“transcendence”has a dual nature,which can not only more inclusively incorporate large language model-like intelligent machines and their functions into the interpretable scope of“things”,but also break through the constraints of correlationism in a decentralized manner,and lead to the transcendence of external things through the aesthetic activity of“joint reality”,which enlivens human creation and provides support for the human meaning world.In this way,it opens up a path
关 键 词:大语言模型 物导向哲学 意义生成 相关主义 联合实在
分 类 号:TP18[自动化与计算机技术—控制理论与控制工程] B15[自动化与计算机技术—控制科学与工程]
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...
正在载入数据...