Authors: Ye Wenjie, Gao Qiqi
Abstract: AI large-scale models, exemplified by GPT and DeepSeek, have had a profound impact on the global order. Extending the “system-lifeworld” theory to large language models not only reveals how two technological trajectories—closed-source hegemony and open-source cooperation—differently construct the global order, but also enables a paradigmatic shift of the theory from a critique of modernity to an interpretation of the global digital order.Closed-source large models epitomized by GPT are progressively reinforcing U.S.-centric global hegemony, seeking to establish a worldwide “digital system.” Conversely, open-source counterparts like DeepSeek are emerging as transformative technological forces that facilitate globally negotiated construction, thereby endeavoring to shape a global “digital lifeworld.”GPT-driven technology aims to construct a global “digital system” that consolidates technological and cultural hegemony while eroding communicative rationality among pluralistic actors in intersubjective interactions. Its training paradigm imposes exceptionally high demands for data accumulation and computational reserves, creating a monopolistic developmental trajectory for large-scale models. Concurrently, AI oligopolies leverage GPT to advance the “uni-acculturation” of knowledge across global domains, thereby perpetuating the reproduction of knowledge imperialism.DeepSeek, however, functions as a transformative force. It facilitates the global development of large-scale models toward “co-ownership, co-governance, and sharing,” promoting “trans-acculturation” of knowledge and negotiated construction within the global digital space. This fosters communicative rationality among pluralistic actors across technological and cultural “dual fields,” thereby contributing to the construction of a global “digital lifeworld.”
Keywords: Era of Large Models; GPT; DeepSeek; Global Order; Digital Governance; System; Lifeworld
Journal：Journal of Hohai University (Philosophy and Social Sciences), 1–11.
Publish date：2025/9/10
