独异性社会带来了结构性孤独常态，智能陪伴成为人机数字化交往重要趋势。可供性作为本研究的理论视角，其概念核心关照主客体间的关系属性，映现人机交往中机器主体性增强的未来图景，与智能陪伴议题有较强理论贴合性。基于文本数据挖掘、内容分析、参与观察、深度访谈等方法，本研究围绕智能陪伴的需求产生、产品供给、用户使用、社会影响四个递进式问题展开。智能陪伴的需求产生，蕴含着结构性和个体性的双重驱动力量。宏观层面，工业化、媒介化、智能化等变迁重塑了陪伴关系。微观层面，基于15535条微博话题数据，研究发现想象可供性中的物质性与社交成本、中介化与专属体验、情感化与情绪价值等因素构成了智能陪伴的个体化需求动机。智能陪伴供给方面，研究选取了34款代表性算法驱动型智能陪伴产品，收集整理其官方宣传文案作为研究样本，编码分析发现智能陪伴产品宣传聚焦于产品设定、功能特点、获取方式、技术优势四项要素。智能陪伴产品的功能形态基于本能层、行为层、反思层的情感化设计，在身体理想、数字人格、关系程度三重层次持续迭代演化。智能陪伴的消费体验方面，研究提炼出响应、工具、内容、情感、交往五重可供层次。以“智能陪伴”为搜索关键字，研究爬取京东商城的279件商品、78174条商品评价信息，形成1239个可供性因素关键词词表，包括词性、词频、TF_IDF值、情感值、主客体属性、可供性类别六个维度。研究发现智能陪伴设备整体具有较高的移动可供性（71.3%）与生产可供性（66.1%）；而社交可供性（49.2%）稀缺且重要，可传情、可致意等成为影响用户消费体验的关键可供性因素。用户对于智能陪伴设备不仅抱有工具理性的期待，更具有情感陪伴与深度交往的渴望。智能陪伴的现实影响兼具价值与风险，基于4个智能陪伴用户趣缘社群，研究从2022年11月到2023年3月开展了为期5个月的观察记录，发现智能陪伴用户在社交过程中存在感知层面乐观偏误、话题层面由虚向实、观点层面偏向强化、关系层面流动蔓延的特点。同时，设计者意图和使用者期待的夹角、虚拟陪伴与现实社交的区隔，也会导致用户主体性损害、效能感失控等风险。研究类比内容层面信息茧房的概念，创新性提出关系层面情感温室的设想，探讨人机交往中的双向驯构与情感抵抗，并探索智能陪伴对于现实社会可供、可用、可信、可替、可塑的影响路径，为理解智能时代人机交往中的陪伴关系提供了新的研究视角。
The society of singularities has resulted in a structural norm of loneliness, and intelligent companionship has become an important trend of human-machine digital association. As the theoretical perspective of this study, the core of the concept of affordance focuses on the relationship between subject and object, which reflects the future enhancement of machine subjectivity in human-machine interaction, so it is theoretically compatible with the topic of intelligent companionship. Based on methods such as text data mining, content analysis, participant observation, and in-depth interviews, this study investigates four progressively interrelated issues surrounding intelligent companionship: demand generation, product supply, user experience, and social impact.The demand for intelligent companionship is driven by both structural and individual factors. At the macro level, changes such as industrialization, mediatization, and digitization have reshaped companion relationships. At the micro level, based on a study of 15,535 Weibo posts, it was found that materiality and social costs, intermediation and exclusive experiences, affect and emotional value constitute individualized motivational factors for intelligent companionship. In terms of supply, the study selected 34 algorithm-driven intelligent companion products that are representative on the market and collected their official promotional materials as research samples. Content analysis revealed that the promotional focus of intelligent companion products is on product settings, functional features, acquisition methods, and technological advantages. The functional design of intelligent companion products is based on emotionalization at the instinctual, behavioral, and reflective levels, and continues to evolve iteratively in terms of bodily ideals, digital personalities, and relationship degrees.Regarding the consumer experience of intelligent companionship, the study identified five levels of satisfaction: responsiveness, utility, content, emotion, and interaction. Using ＂intelligent companion＂ as the search keyword, the study screened and crawled 279 products, along with their product descriptions and sales data, and 78,174 corresponding user comments from JD.com, forming a keyword table of 1,239 elements of affordance, including six analytical dimensions: part of speech, word frequency, TF-IDF value, sentiment value, subject-object category, and affordance category. The study found that intelligent companion devices overall have relatively high levels of mobile affordance （71.3%） and production affordance （66.1%）, while social affordance （49.2%） still exhibits scarcity in the user experience. Additionally, factors such as the ability to convey emotion and intent have become critical elements that influence user purchasing and evaluation. Users not only have expectations for intelligent companion devices in terms of tool rationality, but also have a desire for emotional companionship and deep interaction.The real-world impact of intelligent companionship is both valuable and risky. Based on four user interest communities of intelligent companionship, this study conducted a five-month observation from November 2022 to March 2023, and found that intelligent companion users have the characteristics of optimistic bias at the perception level, topic-level movement from virtual to reality, reinforced viewpoints, and fluid spread at the relationship level during social processes. At the same time, due to the angle between designer intent and user expectation, as well as the distinction between virtual companionship and real-world social interaction, risks such as damage to user subjectivity and loss of efficacy may arise. Based on the comprehensive consideration of the demand, supply, use, and impact of intelligent companionship, this study innovatively proposes the concept of a relationship-level emotional greenhouse by analogy to the content-level information cocoons, exploring the bidirectional domestication and emotional resistance in human-machine interaction, and exploring the influence paths of intelligent companionship on the real-world society in terms of availability, usability, trust, substitutability, and plasticity, providing a new research perspective for understanding companion relationships in the era of intelligent technology.
