工程研究——跨学科视野中的工程2026，Vol.18Issue(3)：265-276,12.DOI:10.3724/j.issn.1674-4969.20250101
情感化依恋:陪伴型AI产品中的情感设计策略研究
Emotional Attachment:A Study on Emotional Design Strategies in Companion AI Products
摘要
Abstract
With the rapid advancement of Artificial Intelligence(AI)technology,Companion AI products have reshaped the traditional human-computer interaction paradigm by virtue of sophisticated Emotional Design,forging in-depth emotional bonds with users and driving the evolution of human-machine relations toward complex intimate connections.However,the escalating anthropomorphism of AI has given rise to a prominent emotional dependence among a subset of users,which further triggers the paradox of"companionship-alienation":these products seemingly offer emotional consolation,yet in reality,they may exacerbate users'loneliness and erode their willingness and ability to engage in real-world social interactions.Against this backdrop,this study aims to systematically dissect this phenomenon and explore rational and sustainable emotional design strategies for companion AI products. This research adopts a case study approach as the core methodology,first clarifying the definition of Companion AI products—referring to AI robots or virtual dialogue agents with a certain level of intelligence,autonomy and social skills that can establish and maintain long-term relationships with users.Such products are capable of capturing both verbal content and non-verbal cues such as tone and emotion,and retain user preferences and personal historical information through memory functions to achieve personalized and coherent interaction.Based on functional characteristics and target user groups,the study categorizes current companion AI software products into four types:intelligent assistant type,role-playing dialogue partner type,AI social community type and emotional support tool type,each catering to users'instrumental needs,self-expression needs,social belonging needs and mental health needs respectively,forming a complementary product matrix.The research also traces the evolutionary trajectory of AI from a mere tool to an emotional partner,revealing that the development of emotional computing,natural language processing and speech recognition technologies has laid a technical foundation for the emotional turn of AI,while the surging social demand for emotional companionship—driven by the high prevalence of loneliness among young people,compressed social time of workplace groups and the weakening of traditional emotional bonds—has become the core impetus for the boom of companion AI products.Market forecasts show that the global companion robot market is poised to reach 150 billion US dollars by 2030,with China accounting for over 40%of the share,indicating the huge development potential of this field. While companion AI products are thriving,their inherent design flaws and potential social risks have gradually emerged,which are the key research focuses of this study.First,the pseudo-social interaction scenarios constructed by anthropomorphic design have led to severe technological dependence:long-term interaction with AI makes users form stable usage habits and emotional attachment,and even interactive addiction,which reduces their motivation to pursue real interpersonal relationships,and even leads to the transfer of social relationship responsibilities,with technology becoming a rationalized tool for interpersonal alienation.For example,a survey of elderly care institutions in Japan shows that 68%of the elderly using companion robots interact with them for more than 14 h a week,while 54%of their children's visiting frequency has dropped by 40%.Second,the"perfect interactivity"of AI companions has reconstructed users'emotional cognitive models,resulting in the atrophy of social skills and the deepening of fear of real social interaction,and even the emergence of"robot-like personality"and the alienation of human subject value.Third,the structural defects of existing AI systems in ethical decision-making lead to ineffective crisis intervention:AI's emotional response is essentially probability prediction rather than real understanding,and it cannot capture the dynamic evolution of users'emotional states over time,which may even indirectly encourage self-harm behavior among vulnerable groups,as exemplified by the 2024 Character.AI suicide inducement incident. To address the above problems,this study first constructs a hierarchical classification system of AI companionship based on the depth of emotional connection,dividing it into five levels from low to high:tool-based companionship(L1),scenario-based companionship(L2),personality-based companionship(L3),symbiotic companionship(L4)and transcendent companionship(L5).Each level corresponds to different interaction modes,technical characteristics and user needs,and presents different risk levels.On this basis,the study proposes a core design concept of"auxiliary rather than substitutive"for companion AI products,advocating that AI should serve as a tool to enhance rather than replace interpersonal communication,helping users connect with and integrate into the real physical world and improve their social competence and quality of life.Centering on this concept,a comprehensive and systematic design methodology system is constructed,including six key strategies:boundary control design to build an emotional immune system,ethical embedding to construct a transparent mental education system,non-anthropomorphic appearance design to break through the uncanny valley trap,graded design based on the traffic light principle to establish a risk response system,user empowerment design to guide users from passive dependence to active exploration,and the balance of virtual-real integration to prevent social alienation.Specific measures include setting up functions such as"digital minimalism"and"emotional cooling-off period",adding AI essence identification and literacy popularization content,adopting surreal or dynamically variable visual design,implementing hierarchical supervision for different companionship levels,opening up UGC creation tools and pressure test modes,and designing offline task systems to link virtual interaction with real life. This study reveals the evolutionary process of companion AI products from simple tool-based interaction to highly intelligent emotional companionship,and clarifies the inherent contradictions and potential risks in their development.The proposed"auxiliary"design concept and corresponding design strategies break through the limitations of the traditional"anthropomorphism-centered"design thinking,and provide a theoretical basis and practical reference for the humanized,ethical and sustainable development of the companion AI industry.The research holds that AI companionship technology is not a simple imitation of human social forms,but the creation of a new form of emotional interaction paradigm,and the excavation of the value of differentiated IP is the key to the future development of the industry.In the future,companion AI products should implant ethical genes in their technical architecture,balance the provision of emotional support and the protection of human nature,and ultimately realize the harmonious coexistence of humans and machines.This study also points out the direction for follow-up research:it is necessary to continue to pay attention to the dynamic evolution of human-machine emotional relationships,and further optimize the design strategy system in combination with the development of new technologies such as brain-computer interfaces,so as to provide users with more high-quality and healthy emotional support services.关键词
陪伴型AI/人工智能/情感设计/人工智能设计Key words
Accompanying AI/Artificial Intelligence/Emotional Design/Artificial Intelligence Design分类
社会科学引用本文复制引用
周瑄,朱敏知..情感化依恋:陪伴型AI产品中的情感设计策略研究[J].工程研究——跨学科视野中的工程,2026,18(3):265-276,12.基金项目
2023年湖南省哲学社会科学成果评审委员会立项课题:后现象学技术哲学视角下技术人工物的设计伦理研究(XSP2023YSC074) （XSP2023YSC074）
