人工智能时代“人机互驯”认知依赖的生成路径研究

Research on the Generation Path of Cognitive Dependence in “Human-Machine Mutual Domestication” in the Era of Artificial Intelligence

摘要: 生成式人工智能的快速普及重塑了传统人机关系，人机之间不再是简单的主客体工具关系，而是演化为人与机器双向作用的“人机互驯”动态格局。在“人机互驯”的过程中，认知依赖逐步生成，带来效率提升的同时也潜藏着主体性消解、思维退化等风险。文章以驯化理论、媒介依赖理论、使用与满足理论为基础，采用扎根理论研究方法，对10位深度使用生成式人工智能的用户开展深度访谈，通过开放式编码、主轴编码、选择性编码三级编码程序提炼范畴，构建“人机互驯”视角下认知依赖的生成路径模型。研究发现，显性驯化(人对机器的驯化)通过需求投射、交互规训、能力延展完成人对AI的改造与占有；隐性驯化(机器对人的反向驯化)经由理解依赖、定向依赖、娱乐依赖实现技术对人的认知渗透；在人主导到机器主导的权力流变过程中，认知依赖逐步生成，最终可能催生算法茧房等认知困境。研究厘清了“人机互驯”的内在逻辑，为规避AI认知依赖风险、构建人机和谐共生的关系提供了理论参考与实践对策。

Abstract: The rapid popularization of generative artificial intelligence has reshaped the traditional human-machine relationship, evolving from a simple subject-object tool relationship to a dynamic pattern of “human-machine mutual domestication” characterized by bidirectional interaction between humans and machines. In the process of “human-machine mutual domestication”, cognitive dependence gradually emerges, bringing efficiency improvements but also potential risks such as subjectivity dissolution and deterioration of thinking. Based on the theories of domestication, media dependence, and uses and gratifications, this paper adopts the grounded theory research method to conduct in-depth interviews with 10 users who deeply engage with generative artificial intelligence. Through a three-level coding process of open coding, axial coding, and selective coding, categories are refined to construct a generation path model of cognitive dependence from the perspective of “human-machine mutual domestication”. The study finds that explicit domestication (human domestication of machines) completes the transformation and possession of AI through demand projection, interactive discipline, and capability expansion; implicit domestication (reverse domestication of humans by machines) achieves technological cognitive penetration of humans through understanding dependence, directional dependence, and entertainment dependence. In the power flow transition from human dominance to machine dominance, cognitive dependence gradually emerges, which may ultimately lead to cognitive dilemmas such as algorithmic cocoons. This study clarifies the inherent logic of “human-machine mutual domestication”, providing theoretical references and practical countermeasures for avoiding the risks of AI cognitive dependence and building a harmonious symbiotic relationship between humans and machines.