Breaking Emotional Barriers: How AI Mitigates Interpersonal Distance and Fosters Social Capability Growth

Breaking Emotional Barriers: How AI Mitigates Interpersonal Distance and Fosters Social Capability Growth | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Breaking Emotional Barriers: How AI Mitigates Interpersonal Distance and Fosters Social Capability Growth Qiyin Hu, Bao Jiao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7495693/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract In today’s “liquid modernity”, rapid technological change and hyper-mediated relationships often leave youth’s emotional needs unmet, fostering fatigue, social pressure, and loneliness. Yet, as interaction is intrinsic to human sociality, intelligent agents are emerging as viable alternatives to traditional social partners. Using grounded theory, this study identifies a prevalent “sense of interpersonal distance” among youth. It examines its moderating role in a 2 (social type: interpersonal vs. human–AI) × 2 (scenario type: competence vs. emotional) experiment. Findings reveal that interpersonal distance significantly shapes evaluations of social types and that AI can function as a social actor in both competence- and emotion-oriented contexts. Such human–AI connections may offer new forms of technologically mediated intimacy, shifting sociality from co-presence to co-existence. Business and commerce/Information systems and information technology Biological sciences/Psychology Social science/Psychology human–AI interaction interpersonal communication sense of interpersonal distance technologically mediated intimacy Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 1. Introduction With industrialization accelerating social differentiation in China, traditional relationship patterns have lost their binding force, giving way to an individualized society (Bauman, 2000). As Han (2015) notes, the “achievement society” compels self-discipline and intensifies individualization, fragmenting interpersonal ties. The rise of internet technologies has reshaped interaction, merging virtual and physical spaces. Online exchanges now dominate daily communication, while offline meetings have become rare, often planned rituals. Smart devices fragment attention, creating a state of being “alone together”. Algorithmic communities transcend geography and foster tailored connections, yet hyperconnectivity often deepens feelings of alienation. Under conditions of time scarcity, social interaction follows an efficiency logic, with partner choice and time allocation treated as cost–benefit decisions. Constant connectivity can lead to cognitive overload, emotional exhaustion, and social media fatigue. Many young people respond with “digital fasting” or “social downgrading” to regain balance. This reflects a more profound contradiction: when promised “free connectivity” becomes “forced connectivity,” social interaction turns into labor that is both consumptive and exploitative. True freedom lies in regaining control over life rhythms and creating space for meaningful connection. Intelligent technologies, particularly AI, may offer more controllable interaction partners. Human–machine relations have evolved from subordination, opposition, and equivalence to a fourth stage—human–AI symbiosis (Xiang & Xu, 2023). In today’s liquid modernity, many young people under academic or work stress find AI an appealing outlet—meeting conversational needs, enabling emotional resonance, and easing the burdens of human-to-human interaction. 2. Literature review 2.1 The social dynamics of contemporary youth groups As one of the most dynamic groups in modern society, youth constitute a core driving force in social development, and their value orientations reflect the social mentality of different historical periods. In empirical research, the age boundaries for defining youth are typically set between 13 and 45, yet the range of 18 to 35 years old is the most commonly adopted and broadly compatible. Contemporary youth have grown up in a stage of deep modernization, assuming the role of “transitional individuals”in the process of modernization transformation. The modernization-oriented evolution of their lifestyles and social mentalities serves as a “weathervane” for the overall shift in societal values (Deng, 2018). With the progression of industrialization, China’s social structure has undergone marked changes, and the pace of social differentiation has accelerated. As sociologist Zygmunt Bauman has observed, modernity in today’s society is characterized by liquid modernity, in which traditional patterns of social relations no longer exert binding force and interpersonal connections tend toward fragmentation (Bauman, 2000). Amid such structural transformations, the relationship between individuals and society has itself shifted in fundamental ways. The acceleration of technological advancement, life rhythms, and social change has profoundly shaped the sense of social order and intelligibility, while also altering the ways individuals participate in social development and affirm their own identities. Interpersonal communication has been “enlisted” into this acceleration, with the logic of speed becoming an “invisible normative force”that constrains and regulates how individuals perceive relationships and the world around them (Lei, 2024). For contemporary youth, the traditional “acquaintance society” based on geographical and kinship ties has gradually evolved into an “individualized society” organized around shared interests. In this trend toward individualization, young people are disembedded from traditional social structures and passively swept along by competitive pressures, with social interaction becoming an activity that occupies every moment. The fragmented forms of time and space continuously compress the intervals between moments, while the high-frequency interactions of social media immerse individuals in persistent psychological strain, manifesting chiefly as social overload and relational fatigue. As they are compelled to keep pace with the perceived velocity of change in both the social and technological worlds, people strive not to lose any potentially valuable connections and to maintain competitive opportunities. When social engagement and information saturate one’s life-space, a sense of estrangement often emerges between the pursuit of efficiency and the search for meaning. Individuals become segmented; relationships become spaced apart. In terms of emotional intensity and relational cohesion, online connections remain at the level of “shallow sociality” and “weak ties.” The “always-on” internet environment and perpetually lively online communities cannot fulfill the deepest social needs: the inability to find a stable and communicable Other, and the inability to see a complete self in the Other. In a fluid, transient, and uncertain society, young people’s online social behaviors are split into divergent selves. As Karen Horney noted, “On the one hand, they alienate themselves from others; on the other, they yearn for their love. It is this wholly insoluble conflict that controls our lives” (Horney, 1937).On one hand, young people require the perpetual festivity and constant presence of online communities to sustain their digital existence; on the other hand, they are acutely aware that such liveliness cannot meet their deeper social needs. This bustle lacks relational bonding and never stays for the self—it often brings greater loneliness the closer one gets. In the face of such fragmentation and divergence, many opt for “digital fasting” and “social disconnection” to reconstruct their social space, intentionally reducing or eliminating certain social activities. By “withdrawing” or “cooling down,” they seek to protect their psychological space, thus giving rise to the segmentation of social relationships. RQ1: What are the current conditions faced by contemporary youth in online and offline social interactions? Furthermore, can these experiences be linked to existing psychological theories to propose a new concept that encapsulates the social perceptions of today’s youth? 2.2 Social interaction is an ontological necessity for human existence Human beings are inherently social; they exist as social entities. In Marx’s words, a person is “the total of all social relations,” and such social relations are neither a priori nor pre-formed. Still, they are gradually constructed and developed through human social interaction (Marx & Engels, 1975). “The development of an individual depends on the development of all other individuals with whom he has direct or indirect interaction”. Universal interaction constitutes the generative process of humanity’s integrated social relations, and the comprehensive development of human beings presupposes the universal development of productive forces and global communication. Thus, humans are the product of social interaction, and interaction is the fundamental mode of human existence. Dialogue, in turn, is the primary means by which humanity engages in social interaction and socialization. As Bakhtin has argued, human beings exist in a fundamentally dialogic manner: dialogic communication is not merely a means of sustaining various social relationships—it is those relationships themselves. It is the very essence of humanity. Without dialogue, one cannot truly be called human (Bakhtin, 1981). When human-to-human communication fails to satisfy social needs, interaction and dialogue—as essential attributes of human nature—propel people to seek another outlet for connection through the medium of “language.” Intelligent robots, emerging as new communicative agents, have thus entered the public sphere. 2.3 AI emerges as a new outlet for youth socialization Since the twentieth century, the communicative dilemma between humans and the “nonhuman” has shifted from interaction with organisms to engagement with “beings without flesh and blood” (Peters, 1999). With the rise of computer languages, human–machine relations evolved from interaction to communication and, eventually, to social engagement. During the 1980s and 1990s, humanistic thought reframed computers as social actors, giving rise to the media equation theory and expanding scholarly attention to emotions and behaviors in human–computer interaction (Xiang & Xu, 2023). Computer language thus became the bridge enabling machines to “understand” human intent. Over the past two decades, computing power, large-scale corpora, and machine learning have significantly advanced natural language processing (NLP), enabling applications like Siri and Xiaodu to become integral to daily life (Luo, Wang, & Wang, 2021). The advent of generative AI marks a new stage of symbiosis, from ChatGPT to OpenAI’s “Operator” (2024). Intelligent systems increasingly simulate interpersonal scenarios, collecting users’ emotional cues and shaping embodied communication (Lin & Ye, 2019). Scholars highlight two perspectives on its social function. The substitution hypothesis, grounded in Freud’s theory of displaced needs, views machines as alternative social objects. Turkle (2011) suggests that deep engagement with machines may fulfill social connection while reducing human–to–human interaction. Experimental findings also show AI responses surpass human ones in perceived accuracy, empathy, and emotional support (Yin, Jia, & Wakslak, 2024). Conversely, the compensation hypothesis stresses the irreplaceable social bonds of human interaction: machines lack lived experience, and their empathy falls short of human resonance (Peng, 2022). From this view, AI supplements interpersonal relationships rather than replacing them (Brandtzaeg, Skjuve, & Følstad, 2022). Regardless of one's stance, generative AI provides youth with a novel outlet for social interaction. Its intelligence and emotional responsiveness align with contemporary needs, potentially alleviating problems of social segmentation in an increasingly accelerated society. This leads to the following question: RQ2: Do the interpersonal experiences of contemporary youth persist in human–machine interaction, and what differences in social perception arise across scenarios? How does AI-mediated interaction influence interpersonal relations? 3. Study 1: A Grounded Theory study on youth socialization 3.1 Method and procedure This study adopts the grounded theory approach (Strauss & Corbin, 1990) to link empirical data with theoretical construction. Considering the contextual sensitivity of youth social issues, a methodological triangulation was applied, combining situational observation, social media text analysis, and in-depth interviews. Data came from two primary sources: （1）Weibo topic crawling： By March 2025, over 30 youth-related topics were collected. After filtering out posts with fewer than 30 words or irrelevant content, 807 posts (≈52,000 Chinese characters) with rich narratives remained. （2）Semi-structured interviews： Twenty-eight participants aged 18–35 (14 male, 14 female) were recruited via the researchers’ networks (n=11) and Douban groups (n=17, all under 25). Each 40–60 minute interview explored perceptions, experiences, and behaviors in social relationships, with respondents receiving compensation of 10–15 RMB. Detailed information on the interviewees is provided in Table 1. Table 1 Demographic Information of Interview Participants. Interview ID Gender Age Occupation/Education Level M1 Male 23 Postgraduate student M2 Male 18 Undergraduate student M3 Male 25 Postgraduate student M4 Male 24 Postgraduate student M5 Male 22 Postgraduate student M6 Male 27 PhD student M7 Male 26 Research institute staff M8 Male 27 Political science instructor M9 Male 19 Undergraduate student M10 Male 30 Factory worker M11 Male 24 Full-time postgraduate entrance exam candidate M12 Male 32 Civil servant M13 Male 35 Civil servant M14 Male 23 Undergraduate student F1 Female 25 Education advocate F2 Female 22 Undergraduate student F3 Female 24 Postgraduate student F4 Female 23 Movie marketing specialist F5 Female 24 Bachelor’s degree holder, actively job hunting F6 Female 35 Human resources officer F7 Female 25 Nurse F8 Female 21 On leave from undergraduate program, unemployed F9 Female 26 TV program planner F10 Female 21 Undergraduate student F11 Female 22 Undergraduate student F12 Female 23 Postgraduate student F13 Female 20 Undergraduate student F14 Female 24 Postgraduate student Note: M = Male; F = Female. The sample selection ensured close alignment with research questions and maintained participant balance, enhancing objectivity and representativeness. Before the interviews, informed consent was obtained, and ethical and confidentiality principles were adhered to. Participants could request access to recordings and processed materials. All transcripts were standardized by removing repetitions and fillers, resulting in a corpus of 263,000 characters. Following the grounded theory’s saturation principle, 90% of the data were used for analysis and coding. In comparison, 10% were reserved for saturation testing and iterative comparison until no new concepts emerged, confirming theoretical saturation. 3.2 Results NVivo was used for exhaustive coding of the text, generating 150 open reference points. Thesewere abstracted into 70 initial concepts, then clustered into 22 categories. Further integration yielded six main categories: emotional needs, competence needs, intrusion of self-boundaries, self–other distance, perceived value, and perceived risk. Cluster analysis and constant comparison refined these into three core categories: social purpose, sense of interpersonal distance, and social cognitive evaluation. Coding Framework of Core Categories are summarized in Table 2. Table 2 Coding Framework of Core Categories. Initial Categories (Second-level) Main Categories (Third-level) Core Categories (Fourth-level) Emotional support Emotional needs Social purpose Leisure and entertainment Personal development and growth Competence needs Information and resource acquisition Social functional needs Self-evaluation anxiety Intrusion of self-boundaries Sense of interpersonal distance Privacy-related anxiety Social physical overload Social information overload Social emotional overload Segmentation of social states Self–other distance Alienation of social relationships Alienation of social emotions Cognitive costs Perceived value Social cognitive evaluation Interaction costs Emotional value Competence value Autonomy value Public opinion risk Perceived risk Relationship risk Privacy risk Trust risk To ensure the density, variability, and integration of theoretical concepts, the remaining 10% of the textual data was used for theoretical saturation testing. The study repeated the aforementioned four-level coding process, and comparative analysis revealed that all concepts generated from the new coding could be subsumed under existing categories. No new concepts or categories emerged, indicating that theoretical saturation had been achieved. 3.3 A novel concept （1）Intrusion of self-boundaries A social boundary refers to the physical and psychological limits that guide social behavior, adjusting with relationship closeness and emotional attitudes. It reflects values, beliefs, and personality, serving as both a behavioral code and a self-protection mechanism. Crossing such boundaries—whether through unwanted participation in events or breaches of privacy—can trigger anxiety and discomfort. With increasing individualization, social activities have become more self-oriented. Young people, regardless of the depth of their relationships, construct a “self-boundary” that grants them sovereignty over their personal domain. As George notes, “What’s yours is yours, what’s mine is mine.” For instance, one participant (F5) stated that she would not accompany a friend shopping without a personal need, preferring to conserve her own time. In emotional relationships, youth prioritize self-subjectivity, making a clear sense of boundary a precondition for engagement (Kang, 2021; Wang & Hu, 2022). Individually, this often means withdrawing from “sticky” ties to preserve independence; socially, it leads to low-intensity, self-centered interactions. Such detachment can cause difficulty adapting to others’ boundaries while still feeling pressured to meet social demands. As one participant (F13) described, merely attending group meetings after conflict required extensive mental preparation. These dynamics heighten social anxiety and burnout, producing what Han (2015) terms “divided burnout”, where each person’s exhaustion is isolated. When vulnerable aspects remain unshared and boundaries form unbridgeable gaps, communities and close relationships weaken. Over time, self-boundaries shift from a “social skin” to a “social fortress”, prompting either stricter barriers or avoidance behaviors, deepening the sense of separation from others. （2）Self–other distance Contemporary youth live in an increasingly individualized era. Beyond the anxiety and burnout seen in social interactions, many describe an insurmountable barrier between self and others, a declining sense of belonging, and reduced motivation to connect. Their communication style tends toward politeness, distance, and deliberate control (Giddens, 1991), fostering estrangement from groups and society. Psychology provides established concepts to explain this phenomenon. Berkman (1983) defined social isolation as the irreversible loss of social ties; Lien-Gieschen (1993) described it as wanting but being unable to connect; Nicholson (2009) expanded this definition to include a lack of belonging, minimal interaction, and low-quality relationships. In the intelligent era, isolation gains new spatial dimensions. It is no longer just the inability to connect, but also the prevalence of shallow, fragmented ties in weak-tie networks. As Simmel (2004) observed, conquering physical distance can widen spiritual distance. The internet may appear to solve connection barriers, yet its noisiness without belonging deepens emotional estrangement. Social isolation now encompasses the loneliness stemming from perceived distance—both physical (in terms of frequency of contact) and emotional (due to a lack of identification). Under broader social pressures, this distance amplifies into separation (e.g., digital disconnection) and alienation (e.g., competitive achievement culture). At its core lies the remoteness, detachment, and disharmony between self and others. （3）Sense of interpersonal distance Ultimately, whether concerning self-boundaries or the distance between the self and others, the core issue lies in the delineation between the self and the other. From a philosophical perspective, separation refers to the establishment of boundaries between the subject and object, the individual and society, and the self and other. Levinas (1969)further argued that there is always an inherent heterogeneity in human relations and that otherness constitutes the foundation of human interaction. These ideas suggest that separation is an intrinsic condition of human communication, serving as the basis for the emergence of both subjectivity and the concept of otherness. Accordingly, the term "separation" is appropriate to summarize the relationship between self-boundaries and self–other distance. The notion of a sense of separation specifically emphasizes the self-referential perceptual meaning of youth in interpersonal interactions. Thus, this study defines the sense of interpersonal separation among youth as: a psychological and behavioral state of isolation, estrangement, or maladjustment experienced in interpersonal communication, arising from the incompatibility of self-boundaries and the widening of perceived self–other distance. This state not only shapes the quality of social experiences but may also exert profound effects on individual mental health and the structure of interpersonal relationship networks. (Figure 1) Building on the grounded model developed earlier—specifically the individual-level dimensions of social needs, social experiences, and social cognitive evaluation—and integrating existing theoretical research, this study constructs a comparative model of interpersonal versus human–AI interaction. The aim is to provide a comprehensive and systematic comparison of youths’ interpersonal communication and human–AI interaction, and to analyze the psychological mechanisms underlying their differences. 4. Study 2: A comparative scenario experiment of value, risks, and impact on HHI and HMI According to the Stimulus–Response (S–R) model, social interaction type and social interaction competence serve as the stimulus variables (S), while perceived value and perceived risk—both components of social cognitive evaluation—represent the organism’s (O) internal judgment. Social behavioral intention is positioned as the response (R) within the model. Drawing on the Cognition–Affect–Conation framework, the interpersonal separation sense experienced by young people in real-life interpersonal interactions functions as the affective component that activates human conation, namely, the intention to engage in social behavior. By defining and explaining the constructs of these variables, the study further proposes its hypotheses.(Figure 2) 4.1 Theoretical constructs （1）Independent variable: human-human interaction vs. human- machine (AI)interaction With the increasing prevalence of human–machine interaction, some researchers have begun to examine its impact on human–human interaction. As discussed earlier, perspectives on this issue primarily fall into two categories: the substitution view and the compensation view. According to Social Presence Theory (Epley et al., 2008), human–human interaction tends to offer deeper social satisfaction because it entails authentic emotional empathy and social cues such as facial expressions, tone of voice, and body language (Chen & Park, 2021). Turkle (2017) argues that in-depth and sustained human–machine interaction can substitute for human–human interaction, fulfilling individuals’ sense of relational belonging and reducing their willingness and inclination to engage in interpersonal communication. Other scholars argue that human–machine interaction serves a compensatory function, partially supplementing interpersonal friendships and emotional needs (Brandtzaeg et al., 2022). Mende et al. (2019) also found that human–machine interaction can enhance the willingness to engage in human–human interaction, as it may trigger identity threat, prompting individuals to select identity-symbolizing products or to cooperate with real human groups. Based on Social Response Theory (SRT), even when people are aware that robots do not possess emotions or human motivations, they still respond in accordance with human social interaction rules when robots display human-like attributes or social cues (Nass et al., 2000). Thus, treating human–human interaction and human–AI interaction as a binary manipulation of social interaction type is conceptually justified. In this study, the operational definitions are as follows: Human–Human Interaction refers to social activities in which individuals or group members exchange and transmit information bidirectionally through verbal or nonverbal communication (Beebe et al., 2002). Human–AI Interaction refers to “human–machine–human” bidirectional communication behaviors mediated by machines supported by internet technologies (Lin & Ye, 2019), which may further evolve into interaction with socially capable intelligent agents (e.g., AI systems, robots). （2）Independent variable: Competence scenario vs. emotional scenario According to self-determination theory, behavior originates from fulfilling three basic needs: competence, relatedness, and autonomy. In social contexts, autonomy is often constrained because relationship formation depends on the willingness of both parties to engage in a mutually beneficial way. Grounded theory results indicate that youth engage in social behavior mainly for emotional and competence needs. Since current human–machine interactions typically involve user-initiated “one-click” activation, and to ensure comparability with human–human interactions, this study focuses on interpersonal communication contexts where individuals proactively seek connections driven by emotional or competence motives. This framework aligns with the Stereotype Content Model (SCM), which posits that social cognition is structured along two universal dimensions—warmth and competence (Fiske et al., 2007; Fiske, 2018). Warmth corresponds to perceived friendliness and trustworthiness, while competence relates to capability and effectiveness. Both dimensions shape rapid judgments and stereotype formation in interpersonal encounters. Accordingly, two social scenarios are defined: Competence scenario: Individuals seek interaction to enhance their skills or obtain resources, evaluating value primarily in terms of utility. Emotional scenario: Individuals seek emotional comfort, recognition, and a sense of belonging when experiencing loneliness or distress, using social interactions to regulate their emotions and restore resilience. （3）Dependent variable: Social behavioral intention Behavioral intention is a key research domain that involves individuals’ attitudes, values, decision-making processes, and willingness to choose and use products or services. In international studies on behavioral intention, American psychologist Ajzen (1985) and colleagues proposed a classic theoretical framework, the Theory of Planned Behavior (TPB), which posits that three factors—personal attitude, subjective norms, and perceived behavioral control—can effectively predict and explain an individual’s behavioral intention. When individuals hold a positive attitude, agree with others’ expectations, and believe they can perform a given behavior, they are more likely to develop behavioral intention and ultimately take corresponding action. In interpersonal relationships, when young people lose their social behavioral intention, they may fall into self-determined loneliness, characterized by a lack of communication and deep emotional connection, which detaches them from society, rendering them an “isolated island” (Zhang & Lin, 2025). As groups gradually fragment and communities cease to exist, the consequences can be devastating for both young individuals and society as a whole. Therefore, social behavioral intention is of great significance to young people. In discussions of human–machine relationships, whether in instrumental use or emotional reliance, intention remains central—its formation marks the beginning of action and the start of a shift in social interaction. Hence, social behavioral intention is included in this study as the dependent variable, and the following hypotheses are proposed: H1: There is an interaction effect between social type and social scenario. H1a: In human–human interaction, the competence (vs. emotional) scenario affects the social behavioral intention of young people. H1b: In human–machine interaction, the competence (vs. emotional) scenario affects the social behavioral intention of young people. Given the accelerating logic of social interaction among contemporary youth, efficiency and value have become essential evaluation criteria. In online social scenarios, artificial intelligence offers unique advantages in response speed, knowledge storage, and continuous availability, which can enhance efficiency and value in social interactions. Therefore, the following hypotheses are proposed: H2a: Compared to human–human interaction in the competence scenario, human–machine interaction leads to higher social behavioral intention. H2b: Compared to human–human interaction in the emotional scenario, human–machine interaction leads to higher social behavioral intention. （4）Mediating variable: Perceived Value Zeithaml’s Perceived Value Theory defines value as the overall evaluation of a product or service based on the trade-off between perceived benefits and perceived costs (Zeithaml, 1988). Benefits include extrinsic gains (e.g., knowledge, information) and intrinsic rewards (e.g., pleasure, achievement), while costs encompass the resources users expend. Drawing on the grounded theory model and Self-Determination Theory, this study conceptualizes perceived value in two dimensions:perceived benefitsandinteraction costs. Compared with human–human interaction, human–machine interaction often yields higher perceived value. With “super-brain” capabilities, AI can condense vast knowledge into dialogue, solve practical problems instantly, and meet competence needs more efficiently than many professional human exchanges. Human–human interaction typically requires temporal synchrony, with mismatched schedules forcing one party to wait (Deng, 2024). In contrast, AI allows instant, on-demand dialogue with complete control over topics, sharply reducing time costs. Monetary costs are also lower in many human–machine interactions, as AI bypasses some material and social expenses inherent in human exchanges. However, advanced AI functions—such as memory storage or premium features—still involve fees, making monetary cost a relevant factor in both contexts. H3a: In competence scenarios, human–machine interaction yields higher perceived value than human–human interaction. H3b: In emotional scenarios, human–machine interaction yields higher perceived value than human–human interaction. Perceived value has a positive effect on behavioral intention. Cronin et al. (2000) identified perceived value as a crucial determinant of customer satisfaction, which in turn influences consumer behavioral intention. He et al. (2022), focusing on community-dwelling older adults, found that enhancing internal factors of perceived value can effectively increase their acceptance and willingness to use companion robots. During interaction, users not only care about whether the interaction partner meets their practical needs, but also about the pleasure and emotional support gained in the process. Perceived value, therefore, psychologically strengthens the closeness of the relationship between users and their interaction partners, thereby increasing the willingness to engage. Hence, the following hypotheses are proposed: H4: Perceived value mediates the effect of the interaction between social type and social scenario on social behavioral intention. H4a: In human–human interaction, perceived value mediates the effect of social scenario on social behavioral intention. H4b: In human–machine interaction, perceived value mediates the effect of social scenario on social behavioral intention. （5）Mediating Variable: Perceived Risk Giddens (2024) argued that modernity has disembedded individuals from traditional social orders. While digital media expands connections, it also fosters unfamiliarity and uncertainty, prompting people to make risk assessments that can trigger negative emotions (Yuan, 2023). Perceived risk—a subjective judgment formed from personal experience and situational context—often involves privacy and trust in social contexts. Individuals assess such risks before deciding whether to engage in interaction (Chen, 2020). Based on this, we propose: H5: Perceived risk mediates the effect of the interaction between social type and social scenario on social behavioral intention. H5a: In human–human interaction, perceived risk mediates the effect of social scenario on social behavioral intention. H5b: In human–machine interaction, perceived risk mediates the effect of social scenario on social behavioral intention. In the privacy dimension, AI’s detachment from existing human networks may alleviate privacy concerns, but algorithmic “black boxes” and data collection raise concerns about leakage (Fan & Gao, 2025). In the trust dimension, human–machine trust exhibits a “human-like” hybrid form, combining interpersonal and system trust. While the externalization of AIGC logic enhances transparency and fosters trust, algorithmic opacity and occasional misinformation still undermine it. Given these differences, human–human and human–machine interactions yield heterogeneous perceived risk outcomes. Therefore: H6a: In competence scenarios, human–machine interaction produces lower perceived risk than human–human interaction. H6b: In emotional scenarios, human–machine interaction produces lower perceived risk than human–human interaction. （6）Moderating Variable：Sense of interpersonal distance The CASA paradigm (Nass & Moon, 2000) posits that humans project emotions onto anthropomorphized, socially capable agents. Individuals with social anxiety or avoidant tendencies may prefer human–machine interaction, as social robots can mitigate anxiety (Rasouli et al., 2022). This preference is influenced by social connectedness needs: when high, perceiving robots as partners can produce a substitution effect, reducing real-life engagement. A sense of interpersonal distance—characterized by incompatible self-boundaries and perceived self–other distance—affects this preference. Those with high interpersonal distance often turn to technological mediation to satisfy social needs while avoiding interpersonal anxiety, burnout, and alienation, thereby reshaping boundaries, recalibrating distance, and reconstructing social motivation and behavior. Conversely, those with low interpersonal distance, experiencing little distress in human interaction, view AI as a tool lacking genuine social meaning; their interaction intentions are driven solely by instrumental needs, showing no significant variation across social types or scenarios. This study, therefore, examines the moderating role of interpersonal distance in the social type × social scenario framework: H7: Interpersonal distance moderates the interaction effect between social type and social scenario on social behavioral intention. H7a: For high-distance youth, in both competence and emotional scenarios, human–machine interaction is more effective than human–human interaction in enhancing intention. H7b: For low-distance youth, in both scenarios, human–human interaction is more effective than human–machine interaction in enhancing intention. 4.2 Operationalized variables （1）Sense of interpersonal distance scale Building on established standardized scales, this study refined the construct of Sense of Interpersonal Distance into two higher-order dimensions: Self-Boundary and Self–Other Distance. The Self-Boundary dimension comprises Evaluation Anxiety (EA1–EA3), Interactive Anxiety (IA1–IA4), and Privacy Anxiety (PA1–PA3) (Fenigstein et al., 1975; Alkis et al., 2017), as well as Emotional Fatigue (EF1–EF3) and Physical Fatigue (BF1–BF3) (Lin et al., 2020; Eng et al., 2021). The Self–Other Distance dimension includes Environmental Disconnectedness (ED1–ED3), Relational Disconnectedness (RD1–RD3), and Self-Emotional Disconnectedness (SD1–SD3) (Üngüren & Tekin, 2023). In the item analysis phase (n = 54), all but one item (SD3) demonstrated significant discriminative power. SD3, with a t-value of 1.359 and a discrimination coefficient of 0.63, was removed. Spearman’s rank correlation identified EF1 as correlating 0.30 with the total score, leading to its exclusion. The remaining 23 items yielded a Cronbach’s alpha of 0.953, indicating excellent internal consistency. Exploratory factor analysis (EFA) was conducted with 152 valid responses after excluding cases failing lie-detection and reverse-coded checks. The KMO measure was 0.931, and Bartlett’s test was significant (p < .001). Principal component analysis indicated component correlations exceeding 0.4, warranting oblique rotation. Three components had eigenvalues greater than 1, with factor correlations above 0.3. Parallel analysis (Figure 3)revealed that the real-data eigenvalue curve intersected the simulated-data curve between the second and third factors, indicating that the variance explained by the first two factors exceeded the random error variance. Both theoretical considerations and statistical criteria supported a two-factor model as the most parsimonious and optimal solution. According to statistical standards, an item with a factor loading ≥ 0.30 is considered salient. If an item exhibits a factor loading < 0.30 or an absolute value of cross-loading < 0.10, it should be removed from further analysis (Zhou et al., 2017). Consequently, items B3, D4, D5, and D1 were sequentially removed, and an exploratory factor analysis (EFA) was conducted on the remaining 19 items. The Kaiser–Meyer–Olkin (KMO) value was 0.922, and Bartlett’s test of sphericity yielded p < .001, indicating sampling adequacy and factorability. The two extracted common factors had a cumulative variance contribution rate of 61.786%, exceeding the 60% benchmark (Shi et al., 2012). These two factors were subsequently named Sense of Self-Boundary and Distance Between Self and Others.The factor loading matrix after oblique rotation is presented in Table 3. Table 3. Factor Loading Matrix After Oblique Rotation. Item Component 1 Component 2 B1 0.829 B2 0.837 B4 0.682 B5 0.893 B6 0.852 B7 0.745 B8 0.762 B9 0.780 B10 0.770 B11 0.770 B12 0.645 B13 0.739 B14 0.688 B15 0.458 D2 0.608 D3 0.797 D6 0.911 D7 0.742 D8 0.943 Data were recollected via wjx.comand subjected to confirmatory factor analysis (CFA) using AMOS. In accordance with statistical guidelines, the sample size should be at least 10 times the number of scale items to ensure statistical validity. Therefore, 250 responses were gathered, and 227 valid cases remained after excluding invalid responses. Based on the standardized path coefficients, items B1, B2, B4, B5, B6, and B7—although meeting the 0.50 threshold—were at borderline values and thus were removed. The remaining eight items were reanalyzed via CFA. All error variances reached statistical significance, and all standardized path coefficients were significantly above 0.50. The model demonstrated good fit indices: χ²/df = 2.041, CFI = 0.970, TLI = 0.963, RMSEA = 0.068. Other variable scales and their adaptations The other variables—perceived value (Zhang et al., 2021; Hew & Kadir, 2016; Li et al., 2021), perceived risk (Rempel et al., 1986; Ding & Peng, 2020; Li et al., 2021), and social behavioral intention (Venkatesh et al., 2000)—were all adapted from well-established measurement scales. The Cronbach’s α coefficients for all scales exceeded 0.80, indicating satisfactory internal consistency reliability. According to the model fit indices, the structural models for the three scales exhibited good fit, confirming their structural validity. As social behavioral intention was modeled as a saturated model, no structural validity analysis was required. For each scale, the factor loading of every item on its corresponding latent factor was greater than 0.40 (Hair et al., 2013), and all factor loadings met the minimum threshold of 0.50 (Brown et al., 1993). Additionally, the composite reliability values for all models exceeded 0.80, indicating satisfactory convergent validity of the measurement instruments used in this study. 4.3 Pre-experiment With technology increasingly blurring the line between interpersonal and human–machine interaction, this study employs a scenario-based experiment to examine the psychological mechanisms involved, focusing on perceived value, perceived risk (mediators), and sense of interpersonal distance (moderator). Participants first completed demographic and emotional state measures (control variables) and were then randomly assigned to one of four conditions: interpersonal competence, interpersonal emotional, human–machine competence, or human–machine emotional. Using a scenario priming method, participants recalled and described past events matching the assigned condition. This strengthened recall accuracy and allowed cross-validation between event elements and manipulation items. Priming materials were uniform in structure and length to avoid measurement bias (Appendix A). A pretest with 10 participants confirmed accurate recall and understanding. The main pre-experiment recruited 205 participants via Credamo; after excluding 25 participants (due to attention check failures or lack of prior AI experience), 180 remained (47 per group). Manipulation checks included a single-choice interaction type item and open-ended contextual details, with 98% recalling events from the past week. Interaction context was measured with adapted Stereotype Content Model scales (Aaker et al., 2010), with high reliability (Cronbach’s α > .90). One-sample t-tests (5-point Likert, test value = 3) confirmed successful manipulation: emotional needs (M = 4.28, SD = 0.66, t = 18.56, p < .001) and competence needs (M = 3.97, SD = 0.73, t = 12.70, p < .001). No group mood differences were found, ruling out mood as a confounding factor. 4.4 Formal-experiment （1） Experiment design and procedure In the formal experiment, the measurement of the mediator and moderator variables was incorporated into the procedure. Both the mediator and moderator variables were treated as continuous variables. All other experimental procedures were identical to those employed in the pilot study. （2）Data collection and manipulation check The sample size was estimated using G*Power 3.1, assuming a medium effect (f = 0.30), α = 0.05, and 80% power, which required 237 participants (Faul et al., 2007). To offset attrition, 300 individuals were recruited via Credamo; after failing checks and withdrawals, 288 remained (72 per group; 124 males and 164 females). Most were 18–26 (80.5%), and students (65.3%). Manipulation checks (single-choice and open-ended) confirmed accurate recall, with 80% of participants describing events that occurred within the past week, ensuring vivid responses. All scales showed high reliability (Cronbach’s α > .9). One-sample t-tests indicated emotional (M = 4.25, SD = 0.68, t = 22.02, p < .001) and competence needs (M = 4.20, SD = 0.65, t = 22.09, p < .001) exceeded the midpoint, confirming effective manipulations. No mood differences were observed, and measures of value, risk, distance, and behavioral intention demonstrated strong psychometric properties. （3）Results Before analyzing variance (ANOVA), Levene’s test for equality of variances was performed. Based on the p-value (p < .05), the null hypothesis of homogeneity was rejected, indicating heterogeneity of variances. Examination of residual and data distributions indicated approximate normality with low skewness. Consequently, variable transformation and robust standard error estimation were considered. After applying a square root transformation to the dependent variable, heterogeneity persisted. Therefore, the HC3 heteroskedasticity-consistent estimator was employed to address the potential underestimation of standard errors caused by heteroscedasticity. The advantage of this method lies in preserving the original model structure while adjusting the inferential results (i.e., p-values and confidence intervals) (MacKinnon & White, 1985). The results revealed that both social interaction type (p < .001, η² = .065) and social interaction scenario (p < .05, η² = .015) exerted significant main effects on social behavioral intention. Furthermore, the interaction effect between interaction type and scenario was significant (p < .01, η² = .037), indicating that the influence of interaction type differed across scenarios. Among the control variables, neither gender nor pre-experiment mood had a significant effect on social behavioral intention, thereby eliminating potential confounding effects. H1 was thus supported. Given the significant interaction, simple effects analyses were conducted to explore between-group differences. Participants in the human–machine group reported significantly higher social behavioral intention (M = 3.90, SD = 0.65, 95% CI [3.43, 3.69]) than those in the human–human group (M = 2.93, SD = 0.65, 95% CI [2.80, 3.69]), F(1, 176) = 66.48, p < .001, η² = .14. Participants in the competence scenario (M = 3.56, SD = 0.65, 95% CI [3.23, 3.49]) also reported significantly higher social behavioral intention than those in the warmth scenario (M = 3.13, SD = 0.65, 95% CI [3.00, 3.26]), F(1, 284) = 6.86, p < .01, η² = .024. Independent samples t tests further indicated that, in the competence scenario, social behavioral intention in the human–human group (M = 2.89, SD = 0.69) was significantly lower than that in the human–machine group (M = 3.84, SD = 0.45), t = –10.01, p < .001, 95% CI [–1.14, –0.76]. In the warmth scenario, social behavioral intention in the human–human group (M = 2.96, SD = 1.00) was also significantly lower than that in the human–machine group (M = 3.28, SD = 0.88), t = –2.03, p = .044, 95% CI [–0.63, –0.01]. H2a and H2b were thus supported (see Figure 4). A bootstrapping analysis was conducted using the PROCESS macro (Hayes, 2014) with 5,000 resamples and 95% confidence intervals. Given that the independent variables in the model were categorical with two factors (two levels each), Model 8 of PROCESS was selected. One factor, with two levels, was entered as the moderator, and dummy coding was applied to the independent variables (interpersonal = 0, human–machine = 1; competence = 0, emotion = 1). Social behavioral intention served as the dependent variable, while perceived value and perceived risk were included as mediators to test the mediating effects. The results indicated that under the interaction between social scenario and social interaction type, the confidence intervals for both perceived value and perceived risk did not contain zero,suggesting that the mediating effects were significant. The mediation test pathways are presented in the corresponding table. The results of the mediation model test are summarized in Table 4.Both H4and H5 were supported. Table 4. Results of the Mediation Model Test. Dependent Variable:Social Behavioral Intention Variable Coefficient SE t 95% CI Social interaction type 0.52 0.08 6.61*** [0.36, 0.67] Social scenario 0.47 0.08 6.00*** [0.31, 0.62] Social interaction type × Social scenario -0.44 0.11 -3.93*** [-0.66, -0.22] Perceived value 0.89 0.04 18.42*** [0.73, 0.91] Perceived risk -0.13 0.04 -3.20** [-0.21, -0.05] F 156.42*** R² 0.74 Note: p < .05 , p< .01 , p< .001 . SE = Standard Error; CI = Confidence Interval. Notably, unlike traditional complete mediation, the dummy coding of social interaction type and social scenario in the mediation analysis results in different signs for categorical variables. The opposite signs of perceived value reflect a non-additive effect across combinations of interaction type and scenario. Specifically, the interaction term between social interaction type and social scenario (β = -0.34, p < .01) exerts an adverse effect on perceived value, which contrasts with the positive effect of perceived value on behavioral intention (β = 0.89, p < .001) (Hayes, 2018). Data analysis treating social interaction type and social scenario as independent variables further reveals that, in human–machine interaction contexts (vs. interpersonal), users’ perceived value increases significantly; however, this advantage is attenuated in emotional scenarios (β = -0.34, p < .01). Although emotional scenarios directly promote behavioral intention (β = 0.47, p < .001), their indirect effect via reducing perceived value partially offsets this positive influence. A further analysis of the mediating effects of perceived value and perceived risk across different interaction types shows that, within the interpersonal interaction group, the mediation effect of perceived value is significant (LLCI = -0.52, ULCI = -0.09, excluding 0). Within the human–machine interaction group, the mediation effect of perceived value is also significant (LLCI = -0.76, ULCI = -0.42, excluding 0), thereby supporting H4a and H4b. For perceived risk, the mediation effect in the interpersonal interaction group is significant (LLCI = -0.15, ULCI = -0.03, excluding 0), but not significant in the human–machine interaction group (LLCI = -0.04, ULCI = 0.04, including 0), indicating that perceived risk has virtually no influence on behavioral intention in human–machine interactions. Thus, H5a is supported, whereas H5b is not. Compared with interpersonal interaction in ability scenarios, human–machine interaction yields higher perceived value (M interpersonal–ability= 3.30, M human–machine–ability= 3.83) and lower perceived risk (M interpersonal–ability= 2.96, M human–machine–ability= 2.94), supporting H3a and H6a. In emotional scenarios, human–machine interaction also leads to higher perceived value (M interpersonal–emotional = 2.92, M human–machine–emotional = 3.12); however, H3b is not supported. It further produces lower perceived risk (M interpersonal–emotional = 3.61, M human–machine–emotional= 2.96), supporting H6b. To examine the moderating effect of sense of interpersonal distance on the path from the independent variables to the dependent variable, the two categorical variables were entered as moderators. Based on 5,000 bootstrap resamples in PROCESS Model 3, the interaction term among social interaction type, social scenario, and sense of interpersonal distance had a confidence interval excluding zero (LLCI = 0.50, ULCI = 1.09), confirming a significant moderation effect and supporting H7. Finally, differences in behavioral intention were compared between participants with varying levels of sense of interpersonal distance. Given the normal distribution of interpersonal distance scores (M = 3.02, SD = 0.87), and following prior research, the top 27% and bottom 27% were selected to form the high- and low-score groups, respectively. Scores below 2.31 were classified as a low sense of interpersonal distance, and scores above 3.54 as a high sense of interpersonal distance. As shown in the figures(figure 5 and figure 6), in the competence scenario, for youth with a high sense of interpersonal distance, human–machine interaction is more effective than human–human interaction in enhancing willingness to engage in social interaction. For youth with a low sense of interpersonal distance, there is a difference between human–human and human–machine interaction in promoting social interaction willingness, though the gap is relatively small, with human–machine interaction showing a slight advantage. In the emotion scenario, for youth with a high sense of interpersonal distance, human–machine interaction again outperforms human–human interaction in promoting willingness to engage in social interaction. However, for youth with a low sense of interpersonal distance, there is a significant difference between the two interaction types: when the interaction partner is an artificial intelligence, social interaction willingness is significantly weakened. Thus, H7a is supported, while H7b is not supported. 5. Conclusions and Limitations This study examined youth socialization from a macro perspective and introduced the construct of “Sense of Interpersonal Distance.” The findings suggest that AI-mediated interaction may help reduce some of the negative experiences associated with human–human interaction. For individuals with higher interpersonal distance, AI appeared to provide perceived value and reduced risk, thereby potentially fostering openness toward others. At the same time, participants with lower interpersonal distance tended to prefer human–human interaction in emotionally oriented contexts but were more open to AI in competence-oriented tasks. These patterns indicate that AI has the potential to function as an enabling mediator in specific circumstances, though such effects should be interpreted cautiously and within the limitations of the study’s design. Two important risks should also be noted. First, the absence of embodiment constrains multisensory engagement, limiting the depth of interaction compared with face-to-face communication. Second, because AI systems are not yet embedded in everyday social networks, their supportive role may be temporary, potentially leading to mismatches between virtual experiences and real-world relationships. Several limitations should be acknowledged. Human–human interaction was simplified to dyadic exchanges, and competence- and emotion-oriented needs were treated separately, despite their overlap in practice. The study relied on memory-based, cross-sectional data, which may restrict causal interpretations. In addition, the sample was skewed toward young adults (18–26 years) with a higher proportion of female respondents, which limits generalizability. These limitations highlight the importance of future research employing more diverse samples, longitudinal or multimethod approaches, and more ecologically valid social settings to further assess the role of AI in youth socialization. Declarations Acknowledgments We would like to thank all the young participants who generously shared their experiences during the study. Funding This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Competing interests The authors declare no competing interests. Ethical approval This study was reviewed and approved by the institutional ethics review body on December 9, 2024 (Approval No. 20250930), prior to the commencement of any data collection or informed consent procedures. All procedures were conducted in accordance with institutional guidelines and the principles of the Declaration of Helsinki. Informed consent Informed consent was obtained from all participants after ethical approval and prior to data collection. For the interview study, oral consent was obtained before each session conducted via Tencent Meeting, beginning on January 9, 2025. For the survey study, written informed consent was obtained electronically via Wenjuanxing and Credamo, beginning on March 10, 2025. Participants were informed of the study purpose, procedures, voluntary nature of participation, and their right to withdraw at any time without penalty. Data availability The interview materials involved in this project contain personal privacy information and therefore cannot be made publicly available. Aggregated anonymized data may be available from the corresponding author upon reasonable request. Author Contribution These authors contributed equally to this work.QH and BJ contributed to the conceptualization and design of the study. QH was responsible for data collection, data analysis, and drafting of the manuscript. Both authors read and approved the final manuscript. References Aaker J, Vohs KD, Mogilner C (2010) Nonprofits are seen as warm and for-profits as competent: Firm stereotypes matter. J Consum Res 37(2):224–237 Alkis Y, Kadirhan Z, Sat M (2017) Development and validation of social anxiety scale for social media users. Comput Human Behav 72:296–303 Ajzen I (1985) From intentions to actions: A theory of planned behavior. In: Kuhl J, Beckmann J (eds) Action control: From cognition to behavior. Springer, Berlin, pp 11–39 Bakhtin M (1981) The dialogic imagination: Four essays (ed: Holquist M; trans: Emerson C, Holquist M). University of Texas Press, Austin Bauman Z (2000) Liquid modernity. PolityPress, Cambridge Beebe SA, Beebe SJ, Redmond MV, Salem-Wiseman L (2002) Interpersonal communication: Relating to others. Allyn & Bacon, Boston Berkman L (1983) The assessment of social networks and social support in older people. J Am Geriatr Soc 31(12):743–749 Brandtzaeg PB, Skjuve M, Følstad A (2022) My AI friend: How users of a social chatbot understand their human–AI friendship. Human Communication Research 48(3):404–429 Brown TJ, Churchill GA, Peter JP (1993) Improving the measurement of service quality. J Retail 69(1):127–139 Chen QQ, Park HJ (2021). How Anthropomorphism Affects Trust in Intelligent Personal Assistants Industrial Management & Data Systems 121(12):2722–2737 Chen R (2020) An analysis of the relationship between perceived risk in WeChat Moments and self-disclosure. News Frontline 3:50–51 Cronin JJ, Brady MK, Hult GTM (2000). Assessing the effects of quality, value, and customer satisfaction on consumer behavioral intentions in service environments. J Retail 76(2):193–218 Deng Y (2024) Waiting: A sociological analysis of a universal temporal experience. Sociological Studies 39(06):176–199, 230 Deng Z (2018) The modernization-oriented transformation of Chinese youth’s social mentality since the reform and opening up. China Youth Studies (4):5 Ding W, Peng K (2020). Correction and revision of the Chinese version of the Interpersonal Trust Scale. Psychology Monthly 15(6):4–5, 7 Eng L, Nordström P, Strandberg G (2021). Incorporating compassion into compassion fatigue: The development of a new scale. J Clin Psychol 77(9):2109–2129 Epley N, Akalis S, Waytz A, Cacioppo JT (2008) Creating social connection through inferential reproduction: Loneliness and perceived agency in gadgets, gods, and greyhounds. Psychol Sci 19(2):114–120 Fan H, Gao X (2025) Analysis of trust issues and construction of trust mechanisms in complex social systems of human–machine interaction. Jiangsu Social Sciences:65–70 Faul F, Erdfelder E, Lang A-G, Buchner A (2007) G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39(2):175–191 Fenigstein A, Scheier MF, Buss AH (1975) Public and private self-consciousness: Assessment and theory. J Consult Clin Psychol 43(4):522–527 Fiske ST, Cuddy AJC, Glick P (2007) Universal dimensions of social cognition: Warmth and competence. Trends in CognitiveScience11(2):77–83 Fiske ST (2018) Stereotype content: Warmth and competence endure. Curr Dir Psychol Sci 27(2):67–73 Giddens A (1991) The consequences of modernity. Stanford University Press, Stanford Han B-C (2015) The burnout society (trans: Butler E). Stanford University Press, Stanford Hayes AF (2018) Introduction to mediation, moderation, and conditional process analysis: A regression-based approach, 2nd edn. Guilford Press, New York Hair JF, Black WC, Babin BJ, Anderson RE (2013) Multivariate data analysis, 7th edn. Pearson, Harlow He Q, He Y, Li G, Chen X, Li L, Wang Y, Liu J, Ma D, Liu Q (2022) Qualitative study on the willingness of community-dwelling older adults to use companion robots. Journal of Nursing 29(19):6–10 Hew T-S, Kadir SLSA (2016) Predicting the acceptance of cloud-based virtual learning environment: The roles of Self-Determination and Channel Expansion Theory. Telematics Inform 33(4):990–1013 Horney K (1937) The neurotic personality of our time. W. W. Norton, New York Kang Y (2021) The formation mechanism of “lifelong singlehood” under individualization: A case study of urban youth. China Youth Studies (12):33–39 Lei B (2024) The temporal dimension of relationships: A study on contemporary youth’s values in interpersonal communication. Youth Exploration (3):70–79 Levinas E (1969) Totality and infinity: An essay on exteriority (trans: Lingis A). Duquesne University Press, Pittsburgh Li H, Zhou Y, Li J (2021) Are users leaving social media? A study on the influencing factors of social media fatigue based on perceived value. Chinese Journal of Journalism & Communication 43(12):120–141 Lien-Gieschen T (1993) Validation of social isolation related to maturational age: Elderly. Nursing Diagnosis 4(1):37–44 Lin J, Lin S, Turel O, Xu F (2020) The buffering effect of flow experience on the relationship between overload and social media users’ discontinuance intentions. Telematics Inform 49:101374 Lin S, Ye L (2019) Human–machine interaction and reshaping: Intelligent robots as the “sixth medium.” Journal of Journalism and Communication Studies 26(10):87–104, 128 Luo L, Wang B, Wang X (2021) Function, relationship, and philosophy: A study of user–smart speaker interaction from the perspective of human–machine communication. Global Media Journal 8(3):102–118 MacKinnon JG, White H (1985) Some heteroskedasticity-consistent covariance matrix estimators with improved finite sample properties. J Econometrics 29(3):305–325 Marx K, Engels F (1975) Collected Works of Karl Marx and Friedrich Engels, Vol. 3 (March 1843–August 1844). London: Lawrence & Wishart. ISBN 0853152934. Mende M, Scott ML, van Doorn J, Grewal D, Shanks I (2019) Service robots rising: How humanoid robots influence service experiences and elicit compensatory consumer responses. J Mark Res 56(4):535–556 Nass CI, Moon Y, Carney P (2000). Machines and mindlessness: Social responses to computers. J Soc Issues 56(1):81–103 Nicholson NR (2009) Social isolation in older adults: An evolutionary concept analysis. J Adv Nurs 65(6):1342–1352 Peng L (2022) The future of human–machine communication and interaction. Journal of Social Sciences of Hunan Normal University 51(5):12–22 Peters JD (1999) Speaking into the air: A history of the idea of communication. University of Chicago Press, Chicago Rasouli S, Gupta G, Nilsen E, Dautenhahn K (2022) Potential applications of social robots in robot-assisted interventions for social anxiety. Int J Soc Robotics 14(5):1167-1198 Rempel JK, Holmes JG, Zanna MP (1986) How do I trust thee? Psychology Today 20(2):28-34 Shi J, Mo X, Sun Z (2012) Application of the content validity index in scale development. Journal of Central South University (Medical Sciences) 37(2):152–155 Simmel G (2004) The philosophy of money, 3rd enlarged edn (trans: Frisby D, Bottomore T). Routledge, London Strauss A, Corbin J (1990). Basics of qualitative research: Grounded theory procedures and techniques. Sage, Newbury Park Turkle S (2017) Alone together: Why we expect more from technology and less from each other. Basic Books, New York Üngüren Ü, Tekin ÖA (2023). The effects of social disconnectedness, social media addiction, and social appearance anxiety on tourism students’ career intentions: The moderating role of self-efficacy and physical activity. J Hospitality Leisure Sport Tourism Educ 33:100463 Venkatesh V, Davis FD, Morris MG (2000) A theoretical extension of the technology acceptance model: Four longitudinal field studies. Manage Sci 46(2):186–204 Wang X, Hu P (2022) Sense of boundaries: The social needs of modern youth and their construction. China Youth Studies (10):72–79 Xiang A, Xu K (2023). Why and how humans and machines interact: Theoretical origins, paradigm shifts, and future trends. Global Media Journal 10(5):88–105 Yin Y, Jia N, Wakslak CJ (2024) AI can help people feel heard, but an AI label diminishes this impact. Proc. Natl. Acad. Sci. USA 121(14):e2319112121 Yuan G (2023) Digital media, uncertainty, and emotional governance in risk communication. Theory and Reform 3:134–143, 160 Zeithaml VA (1988) Consumer perceptions of price, quality, and value: A means–end model and synthesis of evidence. J Mark 52(3):2–22 Zhang J, Lin B (2025) “Self-determined loneliness”: Reflection on the “self-logic” behind the social patterns of young people in the era of individualization. Inner Mongolia Social Sciences 46(1):173–181 Zhang M, Ye Y, Xu P (2021) Scale development and validation of social media perceived value. Journalism and Communication Review 74(5):28–42 Zhou H, Zhang L, Luo X, et al. (2017) Modifying the Autism Spectrum Rating Scale (6–18 years) to a Chinese context: An exploratory factor analysis. Neurosci Bull 33(2):175–182 Additional Declarations No competing interests reported. Supplementary Files AppendixAScenarioPrimingMaterials.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7495693","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":557818989,"identity":"3c9a4fb9-f694-4e18-95a5-28f966858180","order_by":0,"name":"Qiyin Hu","email":"","orcid":"","institution":"Jilin University","correspondingAuthor":false,"prefix":"","firstName":"Qiyin","middleName":"","lastName":"Hu","suffix":""},{"id":557818991,"identity":"8fffdded-3d17-49ff-b8f6-8886ada9e8ae","order_by":1,"name":"Bao Jiao","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAx0lEQVRIiWNgGAWjYBACxmYGBmYgLcfYDuKykaDFGMQgTgsIgLQkNjATq4W5nfnh44Iam/TmZh4Dhg9lhxn4ZzcQchibsfGMY2m5jUAtjDPOHWaQuHOAkBYGM2ketsNgLcy8bYcZDCQSCGlh/ybN8+9wOiNIy1/itPCYSQMNTwBrYSRSS7HxzL40w8ZmtoKDPefSeSRuENBi2H984+OCbzbyhu3NGx/8KLOW459BSEsDEuMAkObBrx4I5DEYo2AUjIJRMArQAQCMVjts92a73wAAAABJRU5ErkJggg==","orcid":"","institution":"Jilin University","correspondingAuthor":true,"prefix":"","firstName":"Bao","middleName":"","lastName":"Jiao","suffix":""}],"badges":[],"createdAt":"2025-08-30 13:53:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7495693/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7495693/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97888056,"identity":"965acb52-3cba-49ac-b54e-18cff3fc15a9","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1421580,"visible":true,"origin":"","legend":"","description":"","filename":"BreakingEmotionalBarriersHowAIMitigatesInterpersonalDistanceandFostersSocialCapabilityGrowth.docx","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/fa43c8857c640b750801cc45.docx"},{"id":97888054,"identity":"0ae2a9ac-2a0f-41eb-90c6-07b3d315cbf1","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":3953,"visible":true,"origin":"","legend":"","description":"","filename":"85ff22fd01ed459985328c06df0a3182.json","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/06c82d9974168a1bfd833cb7.json"},{"id":97888060,"identity":"cda95411-0fc2-4d00-87dd-bbcd3ca01155","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":45550,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixAScenarioPrimingMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/6fb5f2a65dce32cf6dc54bb8.docx"},{"id":97900797,"identity":"18fa329a-9562-4be4-80bc-a36ca6d446c7","added_by":"auto","created_at":"2025-12-10 15:45:54","extension":"xml","order_by":3,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":154889,"visible":true,"origin":"","legend":"","description":"","filename":"85ff22fd01ed459985328c06df0a31821enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/1da89c669ba49153cc786c19.xml"},{"id":97888064,"identity":"ac2d1c83-c003-46ff-87d4-f48decc56c61","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"png","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":726280,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/9574963d204bcd17073a903c.png"},{"id":97899440,"identity":"58ae1a5e-88da-4f65-9819-9df32ce4ee69","added_by":"auto","created_at":"2025-12-10 15:44:31","extension":"png","order_by":5,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":60138,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/7a3f6a5bbe1fde82b41f238b.png"},{"id":97900965,"identity":"8b2555e4-1353-4f02-a2be-94abedc33af2","added_by":"auto","created_at":"2025-12-10 15:46:13","extension":"png","order_by":6,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":121170,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/77c87d87c5b424c279d54951.png"},{"id":97888069,"identity":"6fba910b-f923-4b9c-8c3e-c512aaa01415","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":249411,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/0e55080f2be1c3142211e893.png"},{"id":97888067,"identity":"88dd3f15-b604-4621-a944-63de85c38651","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":106324,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/fd029a65e640047e54e8a983.png"},{"id":97888074,"identity":"6225d304-31a6-40b0-9cf4-df746ef037c1","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":89544,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/0fc55097332cc381d83a4b39.png"},{"id":97900955,"identity":"5e9344bb-8837-4683-b86e-32116dcdbbb6","added_by":"auto","created_at":"2025-12-10 15:46:11","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":38367,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/cb58b94c21e7f8e67739bead.png"},{"id":97888072,"identity":"b3e11099-a931-4739-bade-d6a7f408d3af","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"png","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":23570,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/28c58595f5422ff7fa94ecd1.png"},{"id":97888073,"identity":"d3fe11ac-8889-449d-ba45-36114a7adf0b","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"png","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21607,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/7c338e07218eb1f788ce6c1a.png"},{"id":97888063,"identity":"83bb7fe3-acc1-44ec-b871-655d72103c81","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"png","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":31854,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/6d634785ec87a5286d788a4d.png"},{"id":97900843,"identity":"516d16cf-1d1b-4875-9f66-bfa77b071b64","added_by":"auto","created_at":"2025-12-10 15:46:00","extension":"png","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":19866,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/c2dbc3c6ff05ec0d3f155787.png"},{"id":97900809,"identity":"d9e9401d-e071-4542-81e3-581da0b0c139","added_by":"auto","created_at":"2025-12-10 15:45:55","extension":"png","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":17843,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/9d76940debff7efe51629d80.png"},{"id":97888075,"identity":"d56c2edc-233a-467e-b2f7-caea29584a91","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"xml","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":151315,"visible":true,"origin":"","legend":"","description":"","filename":"85ff22fd01ed459985328c06df0a31821structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/0efed9fe39ccb328c9f4b1c8.xml"},{"id":97888077,"identity":"aea11b68-ee02-4fb4-824f-1994189d68d7","added_by":"auto","created_at":"2025-12-10 14:04:01","extension":"html","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":160760,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/4e349464b9dd9e7131d8de18.html"},{"id":97888058,"identity":"be962847-c0dc-4b80-8446-e09729643521","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":665575,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eRepresentation of the sense of Interpersonal Distance\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/cad2c7ec7006a5792684fd56.png"},{"id":97888052,"identity":"e7ff29d5-3095-42ba-a822-96463e3236cd","added_by":"auto","created_at":"2025-12-10 14:04:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":60138,"visible":true,"origin":"","legend":"\u003cp\u003eComparative Model of Human-Human vs. Human–AI Interaction\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/461d69c3226a0e127df7ddfe.png"},{"id":97900437,"identity":"bb17c3fa-9078-430d-bc3b-8fce10971862","added_by":"auto","created_at":"2025-12-10 15:45:30","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":121170,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eParallel Analysis of the Scale’s Exploratory Factors\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/fba75f998e91ddda600caa2f.png"},{"id":97900793,"identity":"d475b927-a4de-4f4c-90a4-471faf9283e8","added_by":"auto","created_at":"2025-12-10 15:45:54","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":249411,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSocial Behavior Intention Scores for Different Social Interaction Types Across Social Scenarios\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/86733db8eb56bfaa086b5c5a.png"},{"id":97900876,"identity":"af1349d0-6f7a-4167-bf07-0bbe31aed51d","added_by":"auto","created_at":"2025-12-10 15:46:01","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":106324,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMean Scores of Social Interaction Willingness in Different Social Interaction Types among High and Low Sense of Interpersonal Distance Groups in the Competence Scenario\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/28aea27074bfddb8f534927b.png"},{"id":97900003,"identity":"22ad84fa-aba0-4da2-975c-4fbb67320473","added_by":"auto","created_at":"2025-12-10 15:45:09","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":89544,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eMean Scores of Social Interaction Willingness in Different Social Interaction Types among High and Low Sense of Interpersonal Distance Groups in the Emotion Scenario\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage6.png","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/44259af9961728f4d771e53c.png"},{"id":102747323,"identity":"8bfdba15-87ba-4be0-a26b-4a36e6d6de51","added_by":"auto","created_at":"2026-02-16 09:04:31","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2015443,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/2560cf48-f1f2-4eb2-a318-23fa7a9d4234.pdf"},{"id":97900812,"identity":"b793efc9-4d58-4c4f-a091-b3c6596b4050","added_by":"auto","created_at":"2025-12-10 15:45:56","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":45550,"visible":true,"origin":"","legend":"","description":"","filename":"AppendixAScenarioPrimingMaterials.docx","url":"https://assets-eu.researchsquare.com/files/rs-7495693/v1/63c08c532f036727ddaaee62.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Breaking Emotional Barriers: How AI Mitigates Interpersonal Distance and Fosters Social Capability Growth","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eWith industrialization accelerating social differentiation in China, traditional relationship patterns have lost their binding force, giving way to an individualized society (Bauman, 2000). As Han (2015) notes, the \u0026ldquo;achievement society\u0026rdquo; compels self-discipline and intensifies individualization, fragmenting interpersonal ties. The rise of internet technologies has reshaped interaction, merging virtual and physical spaces. Online exchanges now dominate daily communication, while offline meetings have become rare, often planned rituals. Smart devices fragment attention, creating a state of being \u0026ldquo;alone together\u0026rdquo;.\u003c/p\u003e\n\u003cp\u003eAlgorithmic communities transcend geography and foster tailored connections, yet hyperconnectivity often deepens feelings of alienation. Under conditions of time scarcity, social interaction follows an efficiency logic, with partner choice and time allocation treated as cost\u0026ndash;benefit decisions. Constant connectivity can lead to cognitive overload, emotional exhaustion, and social media fatigue. Many young people respond with \u0026ldquo;digital fasting\u0026rdquo; or \u0026ldquo;social downgrading\u0026rdquo; to regain balance.\u003c/p\u003e\n\u003cp\u003eThis reflects a more profound contradiction: when promised \u0026ldquo;free connectivity\u0026rdquo; becomes \u0026ldquo;forced connectivity,\u0026rdquo; social interaction turns into labor that is both consumptive and exploitative. True freedom lies in regaining control over life rhythms and creating space for meaningful connection. Intelligent technologies, particularly AI, may offer more controllable interaction partners.\u003c/p\u003e\n\u003cp\u003eHuman\u0026ndash;machine relations have evolved from subordination, opposition, and equivalence to a fourth stage\u0026mdash;human\u0026ndash;AI symbiosis (Xiang \u0026amp; Xu, 2023). In today\u0026rsquo;s liquid modernity, many young people under academic or work stress find AI an appealing outlet\u0026mdash;meeting conversational needs, enabling emotional resonance, and easing the burdens of human-to-human interaction.\u003c/p\u003e"},{"header":"2. Literature review","content":"\u003cp\u003e2.1 The social dynamics of contemporary youth groups\u003c/p\u003e\n\u003cp id=\"_Toc1577738152\"\u003eAs one of the most dynamic groups in modern society, youth constitute a core driving force in social development, and their value orientations reflect the social mentality of different historical periods. In empirical research, the age boundaries for defining youth are typically set between 13 and 45, yet the range of 18 to 35 years old is the most commonly adopted and broadly compatible. Contemporary youth have grown up in a stage of deep modernization, assuming the role of \u0026ldquo;transitional individuals\u0026rdquo;in the process of modernization transformation. The modernization-oriented evolution of their lifestyles and social mentalities serves as a \u0026ldquo;weathervane\u0026rdquo; for the overall shift in societal values (Deng, 2018).\u003c/p\u003e\n\u003cp\u003eWith the progression of industrialization, China\u0026rsquo;s social structure has undergone marked changes, and the pace of social differentiation has accelerated. As sociologist Zygmunt Bauman has observed, modernity in today\u0026rsquo;s society is characterized by liquid modernity, in which traditional patterns of social relations no longer exert binding force and interpersonal connections tend toward fragmentation (Bauman, 2000). Amid such structural transformations, the relationship between individuals and society has itself shifted in fundamental ways. The acceleration of technological advancement, life rhythms, and social change has profoundly shaped the sense of social order and intelligibility, while also altering the ways individuals participate in social development and affirm their own identities. Interpersonal communication has been \u0026ldquo;enlisted\u0026rdquo; into this acceleration, with the logic of speed becoming an \u0026ldquo;invisible normative force\u0026rdquo;that constrains and regulates how individuals perceive relationships and the world around them (Lei, 2024).\u003c/p\u003e\n\u003cp\u003eFor contemporary youth, the traditional \u0026ldquo;acquaintance society\u0026rdquo; based on geographical and kinship ties has gradually evolved into an \u0026ldquo;individualized society\u0026rdquo; organized around shared interests. In this trend toward individualization, young people are disembedded from traditional social structures and passively swept along by competitive pressures, with social interaction becoming an activity that occupies every moment. The fragmented forms of time and space continuously compress the intervals between moments, while the high-frequency interactions of social media immerse individuals in persistent psychological strain, manifesting chiefly as social overload and relational fatigue. As they are compelled to keep pace with the perceived velocity of change in both the social and technological worlds, people strive not to lose any potentially valuable connections and to maintain competitive opportunities.\u003c/p\u003e\n\u003cp\u003eWhen social engagement and information saturate one\u0026rsquo;s life-space, a sense of estrangement often emerges between the pursuit of efficiency and the search for meaning. Individuals become segmented; relationships become spaced apart. In terms of emotional intensity and relational cohesion, online connections remain at the level of \u0026ldquo;shallow sociality\u0026rdquo; and \u0026ldquo;weak ties.\u0026rdquo; The \u0026ldquo;always-on\u0026rdquo; internet environment and perpetually lively online communities cannot fulfill the deepest social needs: the inability to find a stable and communicable Other, and the inability to see a complete self in the Other. In a fluid, transient, and uncertain society, young people\u0026rsquo;s online social behaviors are split into divergent selves. As Karen Horney noted, \u0026ldquo;On the one hand, they alienate themselves from others; on the other, they yearn for their love. It is this wholly insoluble conflict that controls our lives\u0026rdquo; (Horney, 1937).On one hand, young people require the perpetual festivity and constant presence of online communities to sustain their digital existence; on the other hand, they are acutely aware that such liveliness cannot meet their deeper social needs. This bustle lacks relational bonding and never stays for the self\u0026mdash;it often brings greater loneliness the closer one gets. In the face of such fragmentation and divergence, many opt for \u0026ldquo;digital fasting\u0026rdquo; and \u0026ldquo;social disconnection\u0026rdquo; to reconstruct their social space, intentionally reducing or eliminating certain social activities. By \u0026ldquo;withdrawing\u0026rdquo; or \u0026ldquo;cooling down,\u0026rdquo; they seek to protect their psychological space, thus giving rise to the segmentation of social relationships.\u003c/p\u003e\n\u003cp\u003eRQ1: What are the current conditions faced by contemporary youth in online and offline social interactions? Furthermore, can these experiences be linked to existing psychological theories to propose a new concept that encapsulates the social perceptions of today\u0026rsquo;s youth?\u003c/p\u003e\n\u003cp\u003e2.2\u0026nbsp;Social interaction is an ontological necessity for human existence\u003c/p\u003e\n\u003cp\u003eHuman beings are inherently social; they exist as social entities. In Marx\u0026rsquo;s words, a person is \u0026ldquo;the total of all social relations,\u0026rdquo; and such social relations are neither a priori nor pre-formed. Still, they are gradually constructed and developed through human social interaction (Marx \u0026amp; Engels, 1975). \u0026ldquo;The development of an individual depends on the development of all other individuals with whom he has direct or indirect interaction\u0026rdquo;.\u003c/p\u003e\n\u003cp\u003eUniversal interaction constitutes the generative process of humanity\u0026rsquo;s integrated social relations, and the comprehensive development of human beings presupposes the universal development of productive forces and global communication. Thus, humans are the product of social interaction, and interaction is the fundamental mode of human existence. Dialogue, in turn, is the primary means by which humanity engages in social interaction and socialization. As Bakhtin has argued, human beings exist in a fundamentally dialogic manner: dialogic communication is not merely a means of sustaining various social relationships\u0026mdash;it is those relationships themselves. It is the very essence of humanity. Without dialogue, one cannot truly be called human (Bakhtin, 1981).\u003c/p\u003e\n\u003cp\u003eWhen human-to-human communication fails to satisfy social needs, interaction and dialogue\u0026mdash;as essential attributes of human nature\u0026mdash;propel people to seek another outlet for connection through the medium of \u0026ldquo;language.\u0026rdquo; Intelligent robots, emerging as new communicative agents, have thus entered the public sphere.\u003c/p\u003e\n\u003cp\u003e2.3 AI emerges as a new outlet for youth socialization\u003c/p\u003e\n\u003cp\u003eSince the twentieth century, the communicative dilemma between humans and the \u0026ldquo;nonhuman\u0026rdquo; has shifted from interaction with organisms to engagement with \u0026ldquo;beings without flesh and blood\u0026rdquo; (Peters, 1999). With the rise of computer languages, human\u0026ndash;machine relations evolved from interaction to communication and, eventually, to social engagement. During the 1980s and 1990s, humanistic thought reframed computers as social actors, giving rise to the media equation theory and expanding scholarly attention to emotions and behaviors in human\u0026ndash;computer interaction (Xiang \u0026amp; Xu, 2023). Computer language thus became the bridge enabling machines to \u0026ldquo;understand\u0026rdquo; human intent.\u003c/p\u003e\n\u003cp\u003eOver the past two decades, computing power, large-scale corpora, and machine learning have significantly advanced natural language processing (NLP), enabling applications like Siri and Xiaodu to become integral to daily life (Luo, Wang, \u0026amp; Wang, 2021). The advent of generative AI marks a new stage of symbiosis, from ChatGPT to OpenAI\u0026rsquo;s \u0026ldquo;Operator\u0026rdquo; (2024). Intelligent systems increasingly simulate interpersonal scenarios, collecting users\u0026rsquo; emotional cues and shaping embodied communication (Lin \u0026amp; Ye, 2019).\u003c/p\u003e\n\u003cp\u003eScholars highlight two perspectives on its social function. The substitution hypothesis, grounded in Freud\u0026rsquo;s theory of displaced needs, views machines as alternative social objects. Turkle (2011) suggests that deep engagement with machines may fulfill social connection while reducing human\u0026ndash;to\u0026ndash;human interaction. Experimental findings also show AI responses surpass human ones in perceived accuracy, empathy, and emotional support (Yin, Jia, \u0026amp; Wakslak, 2024). Conversely, the compensation hypothesis stresses the irreplaceable social bonds of human interaction: machines lack lived experience, and their empathy falls short of human resonance (Peng, 2022). From this view, AI supplements interpersonal relationships rather than replacing them (Brandtzaeg, Skjuve, \u0026amp; F\u0026oslash;lstad, 2022).\u003c/p\u003e\n\u003cp\u003eRegardless of one\u0026apos;s stance, generative AI provides youth with a novel outlet for social interaction. Its intelligence and emotional responsiveness align with contemporary needs, potentially alleviating problems of social segmentation in an increasingly accelerated society. This leads to the following question:\u003c/p\u003e\n\u003cp\u003eRQ2: Do the interpersonal experiences of contemporary youth persist in human\u0026ndash;machine interaction, and what differences in social perception arise across scenarios? How does AI-mediated interaction influence interpersonal relations?\u003c/p\u003e"},{"header":"3. Study 1: A Grounded Theory study on youth socialization","content":"\u003cp\u003e3.1 Method and procedure\u003c/p\u003e\n\u003cp\u003eThis study adopts the grounded theory approach (Strauss \u0026amp; Corbin, 1990) to link empirical data with theoretical construction. Considering the contextual sensitivity of youth social issues, a methodological triangulation was applied, combining situational observation, social media text analysis, and in-depth interviews.\u003c/p\u003e\n\u003cp\u003eData came from two primary sources:\u003c/p\u003e\n\u003cp\u003e（1）Weibo topic crawling：\u0026nbsp;By March 2025, over 30 youth-related topics were collected. After filtering out posts with fewer than 30 words or irrelevant content, 807 posts (\u0026asymp;52,000 Chinese characters) with rich narratives remained.\u003c/p\u003e\n\u003cp\u003e（2）Semi-structured interviews：\u0026nbsp;Twenty-eight participants aged 18\u0026ndash;35 (14 male, 14 female) were recruited via the researchers\u0026rsquo; networks (n=11) and Douban groups (n=17, all under 25). Each 40\u0026ndash;60 minute interview explored perceptions, experiences, and behaviors in social relationships, with respondents receiving compensation of 10\u0026ndash;15 RMB.\u003c/p\u003e\n\u003cp\u003eDetailed information on the interviewees is provided in Table 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 1 Demographic Information of Interview Participants.\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"3\" cellpadding=\"0\" width=\"522\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003e\u003cstrong\u003eInterview ID\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003e\u003cstrong\u003eGender\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e\u003cstrong\u003eAge\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003e\u003cstrong\u003eOccupation/Education Level\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM1\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e23\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM2\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e18\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM3\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e25\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM4\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e24\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM5\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e22\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM6\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e27\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePhD student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM7\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e26\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eResearch institute staff\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM8\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e27\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePolitical science instructor\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM9\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e19\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM10\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e30\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eFactory worker\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM11\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e24\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eFull-time postgraduate entrance exam candidate\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM12\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e32\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eCivil servant\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM13\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e35\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eCivil servant\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eM14\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eMale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e23\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF1\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e25\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eEducation advocate\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF2\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e22\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF3\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e24\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF4\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e23\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eMovie marketing specialist\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF5\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e24\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eBachelor\u0026rsquo;s degree holder, actively job hunting\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF6\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e35\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eHuman resources officer\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF7\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e25\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eNurse\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF8\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e21\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eOn leave from undergraduate program, unemployed\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF9\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e26\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eTV program planner\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF10\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e21\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF11\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e22\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF12\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e23\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF13\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e20\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003eUndergraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 102px;\"\u003eF14\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 68px;\"\u003eFemale\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 45px;\"\u003e24\u003cbr\u003e\u003c/td\u003e\n \u003ctd style=\"width: 297px;\"\u003ePostgraduate student\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNote: M = Male; F = Female.\u003c/p\u003e\n\u003cp\u003eThe sample selection ensured close alignment with research questions and maintained participant balance, enhancing objectivity and representativeness. Before the interviews, informed consent was obtained, and ethical and confidentiality principles were adhered to. Participants could request access to recordings and processed materials.\u003c/p\u003e\n\u003cp\u003eAll transcripts were standardized by removing repetitions and fillers, resulting in a corpus of 263,000 characters. Following the grounded theory\u0026rsquo;s saturation principle, 90% of the data were used for analysis and coding. In comparison, 10% were reserved for saturation testing and iterative comparison until no new concepts emerged, confirming theoretical saturation.\u003c/p\u003e\n\u003cp id=\"_Toc1302107319\"\u003e3.2 Results\u003c/p\u003e\n\u003cp\u003eNVivo was used for exhaustive coding of the text, generating 150 open reference points. Thesewere abstracted into 70 initial concepts, then clustered into 22 categories. Further integration yielded six main categories: emotional needs, competence needs, intrusion of self-boundaries, self\u0026ndash;other distance, perceived value, and perceived risk. Cluster analysis and constant comparison refined these into three core categories: social purpose, sense of interpersonal distance, and social cognitive evaluation. Coding Framework of Core Categories are summarized in Table 2.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2 \u0026nbsp; Coding Framework of Core Categories.\u003c/strong\u003e\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"3\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\u003cstrong\u003eInitial Categories\u0026nbsp;\u003c/strong\u003e\u003cbr\u003e\u003cem\u003e\u003cstrong\u003e(Second-level)\u003c/strong\u003e\u003c/em\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eMain Categories \u003cem\u003e(Third-level)\u003c/em\u003e\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eCore Categories \u003cem\u003e(Fourth-level)\u003c/em\u003e\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003eEmotional support\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"2\"\u003eEmotional needs\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"5\"\u003eSocial purpose\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eLeisure and entertainment\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003ePersonal development and growth\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 173px;\"\u003eCompetence needs\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eInformation and resource acquisition\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSocial functional needs\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSelf-evaluation anxiety\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"5\" style=\"width: 173px;\"\u003eIntrusion of self-boundaries\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"8\" style=\"width: 179px;\"\u003eSense of interpersonal distance\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003ePrivacy-related anxiety\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSocial physical overload\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSocial information overload\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSocial emotional overload\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eSegmentation of social states\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"3\" style=\"width: 173px;\"\u003eSelf\u0026ndash;other distance\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eAlienation of social relationships\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eAlienation of social emotions\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eCognitive costs\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"5\" style=\"width: 173px;\"\u003ePerceived value\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"9\" style=\"width: 179px;\"\u003eSocial cognitive evaluation\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eInteraction costs\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eEmotional value\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eCompetence value\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eAutonomy value\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003ePublic opinion risk\u003cbr\u003e\u003c/td\u003e\n \u003ctd rowspan=\"4\" style=\"width: 173px;\"\u003ePerceived risk\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eRelationship risk\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003ePrivacy risk\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 200px;\"\u003eTrust risk\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eTo ensure the density, variability, and integration of theoretical concepts, the remaining 10% of the textual data was used for theoretical saturation testing. The study repeated the aforementioned four-level coding process, and comparative analysis revealed that all concepts generated from the new coding could be subsumed under existing categories. No new concepts or categories emerged, indicating that theoretical saturation had been achieved.\u003c/p\u003e\n\u003cp id=\"_Toc1659347503\"\u003e3.3 A novel concept\u003c/p\u003e\n\u003cp\u003e（1）Intrusion of self-boundaries\u003c/p\u003e\n\u003cp\u003eA social boundary refers to the physical and psychological limits that guide social behavior, adjusting with relationship closeness and emotional attitudes. It reflects values, beliefs, and personality, serving as both a behavioral code and a self-protection mechanism. Crossing such boundaries\u0026mdash;whether through unwanted participation in events or breaches of privacy\u0026mdash;can trigger anxiety and discomfort.\u003c/p\u003e\n\u003cp\u003eWith increasing individualization, social activities have become more self-oriented. Young people, regardless of the depth of their relationships, construct a \u0026ldquo;self-boundary\u0026rdquo; that grants them sovereignty over their personal domain. As George notes, \u0026ldquo;What\u0026rsquo;s yours is yours, what\u0026rsquo;s mine is mine.\u0026rdquo; For instance, one participant (F5) stated that she would not accompany a friend shopping without a personal need, preferring to conserve her own time.\u003c/p\u003e\n\u003cp\u003eIn emotional relationships, youth prioritize self-subjectivity, making a clear sense of boundary a precondition for engagement (Kang, 2021; Wang \u0026amp; Hu, 2022). Individually, this often means withdrawing from \u0026ldquo;sticky\u0026rdquo; ties to preserve independence; socially, it leads to low-intensity, self-centered interactions. Such detachment can cause difficulty adapting to others\u0026rsquo; boundaries while still feeling pressured to meet social demands. As one participant (F13) described, merely attending group meetings after conflict required extensive mental preparation.\u003c/p\u003e\n\u003cp\u003eThese dynamics heighten social anxiety and burnout, producing what Han (2015) terms \u0026ldquo;divided burnout\u0026rdquo;, where each person\u0026rsquo;s exhaustion is isolated. When vulnerable aspects remain unshared and boundaries form unbridgeable gaps, communities and close relationships weaken. Over time, self-boundaries shift from a \u0026ldquo;social skin\u0026rdquo; to a \u0026ldquo;social fortress\u0026rdquo;, prompting either stricter barriers or avoidance behaviors, deepening the sense of separation from others.\u003c/p\u003e\n\u003cp\u003e（2）Self\u0026ndash;other distance\u003c/p\u003e\n\u003cp\u003eContemporary youth live in an increasingly individualized era. Beyond the anxiety and burnout seen in social interactions, many describe an insurmountable barrier between self and others, a declining sense of belonging, and reduced motivation to connect. Their communication style tends toward politeness, distance, and deliberate control (Giddens, 1991), fostering estrangement from groups and society.\u003c/p\u003e\n\u003cp\u003ePsychology provides established concepts to explain this phenomenon. Berkman (1983) defined social isolation as the irreversible loss of social ties; Lien-Gieschen (1993) described it as wanting but being unable to connect; Nicholson (2009) expanded this definition to include a lack of belonging, minimal interaction, and low-quality relationships.\u003c/p\u003e\n\u003cp\u003eIn the intelligent era, isolation gains new spatial dimensions. It is no longer just the inability to connect, but also the prevalence of shallow, fragmented ties in weak-tie networks. As Simmel (2004) observed, conquering physical distance can widen spiritual distance. The internet may appear to solve connection barriers, yet its noisiness without belonging deepens emotional estrangement.\u003c/p\u003e\n\u003cp\u003eSocial isolation now encompasses the loneliness stemming from perceived distance\u0026mdash;both physical (in terms of frequency of contact) and emotional (due to a lack of identification). Under broader social pressures, this distance amplifies into separation (e.g., digital disconnection) and alienation (e.g., competitive achievement culture). At its core lies the remoteness, detachment, and disharmony between self and others.\u003c/p\u003e\n\u003cp\u003e（3）Sense of interpersonal distance\u003c/p\u003e\n\u003cp\u003eUltimately, whether concerning self-boundaries or the distance between the self and others, the core issue lies in the delineation between the self and the other. From a philosophical perspective, separation refers to the establishment of boundaries between the subject and object, the individual and society, and the self and other. Levinas (1969)further argued that there is always an inherent heterogeneity in human relations and that otherness constitutes the foundation of human interaction. These ideas suggest that separation is an intrinsic condition of human communication, serving as the basis for the emergence of both subjectivity and the concept of otherness.\u003c/p\u003e\n\u003cp\u003eAccordingly, the term \u0026quot;separation\u0026quot; is appropriate to summarize the relationship between self-boundaries and self\u0026ndash;other distance. The notion of a sense of separation specifically emphasizes the self-referential perceptual meaning of youth in interpersonal interactions. Thus, this study defines the sense of interpersonal separation among youth as: a psychological and behavioral state of isolation, estrangement, or maladjustment experienced in interpersonal communication, arising from the incompatibility of self-boundaries and the widening of perceived self\u0026ndash;other distance. This state not only shapes the quality of social experiences but may also exert profound effects on individual mental health and the structure of interpersonal relationship networks. (Figure 1)\u003c/p\u003e\n\u003cp\u003eBuilding on the grounded model developed earlier\u0026mdash;specifically the individual-level dimensions of social needs, social experiences, and social cognitive evaluation\u0026mdash;and integrating existing theoretical research, this study constructs a comparative model of interpersonal versus human\u0026ndash;AI interaction. The aim is to provide a comprehensive and systematic comparison of youths\u0026rsquo; interpersonal communication and human\u0026ndash;AI interaction, and to analyze the psychological mechanisms underlying their differences.\u003c/p\u003e"},{"header":"4. Study 2: A comparative scenario experiment of value, risks, and impact on HHI and HMI","content":"\u003cp\u003eAccording to the Stimulus\u0026ndash;Response (S\u0026ndash;R) model, social interaction type and social interaction competence serve as the stimulus variables (S), while perceived value and perceived risk\u0026mdash;both components of social cognitive evaluation\u0026mdash;represent the organism\u0026rsquo;s (O) internal judgment. Social behavioral intention is positioned as the response (R) within the model. Drawing on the Cognition\u0026ndash;Affect\u0026ndash;Conation framework, the interpersonal separation sense experienced by young people in real-life interpersonal interactions functions as the affective component that activates human conation, namely, the intention to engage in social behavior. By defining and explaining the constructs of these variables, the study further proposes its hypotheses.(Figure 2)\u003c/p\u003e\n\u003cp\u003e4.1 Theoretical constructs\u003c/p\u003e\n\u003cp\u003e（1）Independent variable: human-human interaction vs. human- machine (AI)interaction\u003c/p\u003e\n\u003cp\u003eWith the increasing prevalence of human\u0026ndash;machine interaction, some researchers have begun to examine its impact on human\u0026ndash;human interaction. As discussed earlier, perspectives on this issue primarily fall into two categories: the substitution view and the compensation view.\u003c/p\u003e\n\u003cp\u003eAccording to Social Presence Theory (Epley et al., 2008), human\u0026ndash;human interaction tends to offer deeper social satisfaction because it entails authentic emotional empathy and social cues such as facial expressions, tone of voice, and body language (Chen \u0026amp; Park, 2021). Turkle (2017) argues that in-depth and sustained human\u0026ndash;machine interaction can substitute for human\u0026ndash;human interaction, fulfilling individuals\u0026rsquo; sense of relational belonging and reducing their willingness and inclination to engage in interpersonal communication. Other scholars argue that human\u0026ndash;machine interaction serves a compensatory function, partially supplementing interpersonal friendships and emotional needs (Brandtzaeg et al., 2022). Mende et al. (2019) also found that human\u0026ndash;machine interaction can enhance the willingness to engage in human\u0026ndash;human interaction, as it may trigger identity threat, prompting individuals to select identity-symbolizing products or to cooperate with real human groups.\u003c/p\u003e\n\u003cp\u003eBased on Social Response Theory (SRT), even when people are aware that robots do not possess emotions or human motivations, they still respond in accordance with human social interaction rules when robots display human-like attributes or social cues (Nass et al., 2000). Thus, treating human\u0026ndash;human interaction and human\u0026ndash;AI interaction as a binary manipulation of social interaction type is conceptually justified. In this study, the operational definitions are as follows:\u003c/p\u003e\n\u003cp\u003eHuman\u0026ndash;Human Interaction refers to social activities in which individuals or group members exchange and transmit information bidirectionally through verbal or nonverbal communication (Beebe et al., 2002). Human\u0026ndash;AI Interaction refers to \u0026ldquo;human\u0026ndash;machine\u0026ndash;human\u0026rdquo; bidirectional communication behaviors mediated by machines supported by internet technologies (Lin \u0026amp; Ye, 2019), which may further evolve into interaction with socially capable intelligent agents (e.g., AI systems, robots).\u003c/p\u003e\n\u003cp\u003e（2）Independent variable: Competence scenario vs. emotional scenario\u003c/p\u003e\n\u003cp\u003eAccording to self-determination theory, behavior originates from fulfilling three basic needs: competence, relatedness, and autonomy. In social contexts, autonomy is often constrained because relationship formation depends on the willingness of both parties to engage in a mutually beneficial way. Grounded theory results indicate that youth engage in social behavior mainly for emotional and competence needs. Since current human\u0026ndash;machine interactions typically involve user-initiated \u0026ldquo;one-click\u0026rdquo; activation, and to ensure comparability with human\u0026ndash;human interactions, this study focuses on interpersonal communication contexts where individuals proactively seek connections driven by emotional or competence motives.\u003c/p\u003e\n\u003cp\u003eThis framework aligns with the Stereotype Content Model (SCM), which posits that social cognition is structured along two universal dimensions\u0026mdash;warmth and competence (Fiske et al., 2007; Fiske, 2018). Warmth corresponds to perceived friendliness and trustworthiness, while competence relates to capability and effectiveness. Both dimensions shape rapid judgments and stereotype formation in interpersonal encounters.\u003c/p\u003e\n\u003cp\u003eAccordingly, two social scenarios are defined:\u003c/p\u003e\n\u003cp\u003eCompetence scenario: Individuals seek interaction to enhance their skills or obtain resources, evaluating value primarily in terms of utility.\u003c/p\u003e\n\u003cp\u003eEmotional scenario: Individuals seek emotional comfort, recognition, and a sense of belonging when experiencing loneliness or distress, using social interactions to regulate their emotions and restore resilience.\u003c/p\u003e\n\u003cp\u003e（3）Dependent variable: Social behavioral intention\u003c/p\u003e\n\u003cp\u003eBehavioral intention is a key research domain that involves individuals\u0026rsquo; attitudes, values, decision-making processes, and willingness to choose and use products or services. In international studies on behavioral intention, American psychologist Ajzen (1985) and colleagues proposed a classic theoretical framework, the Theory of Planned Behavior (TPB), which posits that three factors\u0026mdash;personal attitude, subjective norms, and perceived behavioral control\u0026mdash;can effectively predict and explain an individual\u0026rsquo;s behavioral intention. When individuals hold a positive attitude, agree with others\u0026rsquo; expectations, and believe they can perform a given behavior, they are more likely to develop behavioral intention and ultimately take corresponding action.\u003c/p\u003e\n\u003cp\u003eIn interpersonal relationships, when young people lose their social behavioral intention, they may fall into self-determined loneliness, characterized by a lack of communication and deep emotional connection, which detaches them from society, rendering them an \u0026ldquo;isolated island\u0026rdquo; (Zhang \u0026amp; Lin, 2025). As groups gradually fragment and communities cease to exist, the consequences can be devastating for both young individuals and society as a whole. Therefore, social behavioral intention is of great significance to young people.\u003c/p\u003e\n\u003cp\u003eIn discussions of human\u0026ndash;machine relationships, whether in instrumental use or emotional reliance, intention remains central\u0026mdash;its formation marks the beginning of action and the start of a shift in social interaction. Hence, social behavioral intention is included in this study as the dependent variable, and the following hypotheses are proposed:\u003c/p\u003e\n\u003cp\u003eH1: There is an interaction effect between social type and social scenario.\u003c/p\u003e\n\u003cp\u003eH1a: In human\u0026ndash;human interaction, the competence (vs. emotional) scenario affects the social behavioral intention of young people.\u003c/p\u003e\n\u003cp\u003eH1b: In human\u0026ndash;machine interaction, the competence (vs. emotional) scenario affects the social behavioral intention of young people.\u003c/p\u003e\n\u003cp\u003eGiven the accelerating logic of social interaction among contemporary youth, efficiency and value have become essential evaluation criteria. In online social scenarios, artificial intelligence offers unique advantages in response speed, knowledge storage, and continuous availability, which can enhance efficiency and value in social interactions. Therefore, the following hypotheses are proposed:\u003c/p\u003e\n\u003cp\u003eH2a: Compared to human\u0026ndash;human interaction in the competence scenario, human\u0026ndash;machine interaction leads to higher social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH2b: Compared to human\u0026ndash;human interaction in the emotional scenario, human\u0026ndash;machine interaction leads to higher social behavioral intention.\u003c/p\u003e\n\u003cp\u003e（4）Mediating variable: Perceived Value\u003c/p\u003e\n\u003cp\u003eZeithaml\u0026rsquo;s Perceived Value Theory defines value as the overall evaluation of a product or service based on the trade-off between perceived benefits and perceived costs (Zeithaml, 1988). Benefits include extrinsic gains (e.g., knowledge, information) and intrinsic rewards (e.g., pleasure, achievement), while costs encompass the resources users expend.\u003c/p\u003e\n\u003cp\u003eDrawing on the grounded theory model and Self-Determination Theory, this study conceptualizes perceived value in two dimensions:perceived benefitsandinteraction costs. Compared with human\u0026ndash;human interaction, human\u0026ndash;machine interaction often yields higher perceived value. With \u0026ldquo;super-brain\u0026rdquo; capabilities, AI can condense vast knowledge into dialogue, solve practical problems instantly, and meet competence needs more efficiently than many professional human exchanges.\u003c/p\u003e\n\u003cp\u003eHuman\u0026ndash;human interaction typically requires temporal synchrony, with mismatched schedules forcing one party to wait (Deng, 2024). In contrast, AI allows instant, on-demand dialogue with complete control over topics, sharply reducing time costs. Monetary costs are also lower in many human\u0026ndash;machine interactions, as AI bypasses some material and social expenses inherent in human exchanges. However, advanced AI functions\u0026mdash;such as memory storage or premium features\u0026mdash;still involve fees, making monetary cost a relevant factor in both contexts.\u2028H3a: In competence scenarios, human\u0026ndash;machine interaction yields higher perceived value than human\u0026ndash;human interaction.\u2028H3b: In emotional scenarios, human\u0026ndash;machine interaction yields higher perceived value than human\u0026ndash;human interaction.\u003c/p\u003e\n\u003cp\u003ePerceived value has a positive effect on behavioral intention. Cronin et al. (2000) identified perceived value as a crucial determinant of customer satisfaction, which in turn influences consumer behavioral intention. He et al. (2022), focusing on community-dwelling older adults, found that enhancing internal factors of perceived value can effectively increase their acceptance and willingness to use companion robots. During interaction, users not only care about whether the interaction partner meets their practical needs, but also about the pleasure and emotional support gained in the process. Perceived value, therefore, psychologically strengthens the closeness of the relationship between users and their interaction partners, thereby increasing the willingness to engage. Hence, the following hypotheses are proposed:\u003c/p\u003e\n\u003cp\u003eH4: Perceived value mediates the effect of the interaction between social type and social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH4a: In human\u0026ndash;human interaction, perceived value mediates the effect of social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH4b: In human\u0026ndash;machine interaction, perceived value mediates the effect of social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003e（5）Mediating Variable: Perceived Risk\u003c/p\u003e\n\u003cp\u003eGiddens (2024) argued that modernity has disembedded individuals from traditional social orders. While digital media expands connections, it also fosters unfamiliarity and uncertainty, prompting people to make risk assessments that can trigger negative emotions (Yuan, 2023). Perceived risk\u0026mdash;a subjective judgment formed from personal experience and situational context\u0026mdash;often involves privacy and trust in social contexts. Individuals assess such risks before deciding whether to engage in interaction (Chen, 2020). Based on this, we propose:\u003c/p\u003e\n\u003cp\u003eH5: Perceived risk mediates the effect of the interaction between social type and social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH5a: In human\u0026ndash;human interaction, perceived risk mediates the effect of social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH5b: In human\u0026ndash;machine interaction, perceived risk mediates the effect of social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eIn the privacy dimension, AI\u0026rsquo;s detachment from existing human networks may alleviate privacy concerns, but algorithmic \u0026ldquo;black boxes\u0026rdquo; and data collection raise concerns about leakage (Fan \u0026amp; Gao, 2025). In the trust dimension, human\u0026ndash;machine trust exhibits a \u0026ldquo;human-like\u0026rdquo; hybrid form, combining interpersonal and system trust. While the externalization of AIGC logic enhances transparency and fosters trust, algorithmic opacity and occasional misinformation still undermine it. Given these differences, human\u0026ndash;human and human\u0026ndash;machine interactions yield heterogeneous perceived risk outcomes. Therefore:\u003c/p\u003e\n\u003cp\u003eH6a: In competence scenarios, human\u0026ndash;machine interaction produces lower perceived risk than human\u0026ndash;human interaction.\u003c/p\u003e\n\u003cp\u003eH6b: In emotional scenarios, human\u0026ndash;machine interaction produces lower perceived risk than human\u0026ndash;human interaction.\u003c/p\u003e\n\u003cp\u003e（6）Moderating Variable：Sense of interpersonal distance\u003c/p\u003e\n\u003cp\u003eThe CASA paradigm (Nass \u0026amp; Moon, 2000) posits that humans project emotions onto anthropomorphized, socially capable agents. Individuals with social anxiety or avoidant tendencies may prefer human\u0026ndash;machine interaction, as social robots can mitigate anxiety (Rasouli et al., 2022). This preference is influenced by social connectedness needs: when high, perceiving robots as partners can produce a substitution effect, reducing real-life engagement.\u003c/p\u003e\n\u003cp\u003eA sense of interpersonal distance\u0026mdash;characterized by incompatible self-boundaries and perceived self\u0026ndash;other distance\u0026mdash;affects this preference. Those with high interpersonal distance often turn to technological mediation to satisfy social needs while avoiding interpersonal anxiety, burnout, and alienation, thereby reshaping boundaries, recalibrating distance, and reconstructing social motivation and behavior. Conversely, those with low interpersonal distance, experiencing little distress in human interaction, view AI as a tool lacking genuine social meaning; their interaction intentions are driven solely by instrumental needs, showing no significant variation across social types or scenarios.\u003c/p\u003e\n\u003cp\u003eThis study, therefore, examines the moderating role of interpersonal distance in the social type \u0026times; social scenario framework:\u003c/p\u003e\n\u003cp\u003eH7: Interpersonal distance moderates the interaction effect between social type and social scenario on social behavioral intention.\u003c/p\u003e\n\u003cp\u003eH7a: For high-distance youth, in both competence and emotional scenarios, human\u0026ndash;machine interaction is more effective than human\u0026ndash;human interaction in enhancing intention.\u003c/p\u003e\n\u003cp\u003eH7b: For low-distance youth, in both scenarios, human\u0026ndash;human interaction is more effective than human\u0026ndash;machine interaction in enhancing intention.\u003c/p\u003e\n\u003cp id=\"_Toc1369748531\"\u003e4.2 Operationalized variables\u003c/p\u003e\n\u003cp\u003e（1）Sense of interpersonal distance scale\u003c/p\u003e\n\u003cp\u003eBuilding on established standardized scales, this study refined the construct of Sense of Interpersonal Distance into two higher-order dimensions: Self-Boundary and Self\u0026ndash;Other Distance. The Self-Boundary dimension comprises Evaluation Anxiety (EA1\u0026ndash;EA3), Interactive Anxiety (IA1\u0026ndash;IA4), and Privacy Anxiety (PA1\u0026ndash;PA3) (Fenigstein et al., 1975; Alkis et al., 2017), as well as Emotional Fatigue (EF1\u0026ndash;EF3) and Physical Fatigue (BF1\u0026ndash;BF3) (Lin et al., 2020; Eng et al., 2021). The Self\u0026ndash;Other Distance dimension includes Environmental Disconnectedness (ED1\u0026ndash;ED3), Relational Disconnectedness (RD1\u0026ndash;RD3), and Self-Emotional Disconnectedness (SD1\u0026ndash;SD3) (\u0026Uuml;ng\u0026uuml;ren \u0026amp; Tekin, 2023).\u003c/p\u003e\n\u003cp\u003eIn the item analysis phase (n = 54), all but one item (SD3) demonstrated significant discriminative power. SD3, with a t-value of 1.359 and a discrimination coefficient of 0.63, was removed. Spearman\u0026rsquo;s rank correlation identified EF1 as correlating 0.30 with the total score, leading to its exclusion. The remaining 23 items yielded a Cronbach\u0026rsquo;s alpha of 0.953, indicating excellent internal consistency.\u003c/p\u003e\n\u003cp\u003eExploratory factor analysis (EFA) was conducted with 152 valid responses after excluding cases failing lie-detection and reverse-coded checks. The KMO measure was 0.931, and Bartlett\u0026rsquo;s test was significant (p \u0026lt; .001). Principal component analysis indicated component correlations exceeding 0.4, warranting oblique rotation. Three components had eigenvalues greater than 1, with factor correlations above 0.3. Parallel analysis (Figure 3)revealed that the real-data eigenvalue curve intersected the simulated-data curve between the second and third factors, indicating that the variance explained by the first two factors exceeded the random error variance. Both theoretical considerations and statistical criteria supported a two-factor model as the most parsimonious and optimal solution.\u003c/p\u003e\n\u003cp\u003eAccording to statistical standards, an item with a factor loading \u0026ge; 0.30 is considered salient. If an item exhibits a factor loading \u0026lt; 0.30 or an absolute value of cross-loading \u0026lt; 0.10, it should be removed from further analysis (Zhou et al., 2017). Consequently, items B3, D4, D5, and D1 were sequentially removed, and an exploratory factor analysis (EFA) was conducted on the remaining 19 items. The Kaiser\u0026ndash;Meyer\u0026ndash;Olkin (KMO) value was 0.922, and Bartlett\u0026rsquo;s test of sphericity yielded p \u0026lt; .001, indicating sampling adequacy and factorability. The two extracted common factors had a cumulative variance contribution rate of 61.786%, exceeding the 60% benchmark (Shi et al., 2012). These two factors were subsequently named Sense of Self-Boundary and Distance Between Self and Others.The factor loading matrix after oblique rotation is presented in Table 3.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3. Factor Loading Matrix After Oblique Rotation.\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"3\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd\u003e\u003cstrong\u003eItem\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eComponent 1\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eComponent 2\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003eB1\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.829\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB2\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.837\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB4\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.682\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB5\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.893\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB6\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.852\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB7\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.745\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB8\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.762\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB9\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.780\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB10\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.770\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB11\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.770\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB12\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.645\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB13\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.739\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB14\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.688\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eB15\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.458\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eD2\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.608\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eD3\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.797\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eD6\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.911\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eD7\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.742\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eD8\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.943\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eData were recollected via wjx.comand subjected to confirmatory factor analysis (CFA) using AMOS. In accordance with statistical guidelines, the sample size should be at least 10 times the number of scale items to ensure statistical validity. Therefore, 250 responses were gathered, and 227 valid cases remained after excluding invalid responses.\u003c/p\u003e\n\u003cp\u003eBased on the standardized path coefficients, items B1, B2, B4, B5, B6, and B7\u0026mdash;although meeting the 0.50 threshold\u0026mdash;were at borderline values and thus were removed. The remaining eight items were reanalyzed via CFA. All error variances reached statistical significance, and all standardized path coefficients were significantly above 0.50. The model demonstrated good fit indices: \u0026chi;\u0026sup2;/df = 2.041, CFI = 0.970, TLI = 0.963, RMSEA = 0.068.\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003eOther variable scales and their adaptations\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eThe other variables\u0026mdash;perceived value (Zhang et al., 2021; Hew \u0026amp; Kadir, 2016; Li et al., 2021), perceived risk (Rempel et al., 1986; Ding \u0026amp; Peng, 2020; Li et al., 2021), and social behavioral intention (Venkatesh et al., 2000)\u0026mdash;were all adapted from well-established measurement scales. The Cronbach\u0026rsquo;s \u0026alpha; coefficients for all scales exceeded 0.80, indicating satisfactory internal consistency reliability. According to the model fit indices, the structural models for the three scales exhibited good fit, confirming their structural validity. As social behavioral intention was modeled as a saturated model, no structural validity analysis was required.\u003c/p\u003e\n\u003cp\u003eFor each scale, the factor loading of every item on its corresponding latent factor was greater than 0.40 (Hair et al., 2013), and all factor loadings met the minimum threshold of 0.50 (Brown et al., 1993). Additionally, the composite reliability values for all models exceeded 0.80, indicating satisfactory convergent validity of the measurement instruments used in this study.\u003c/p\u003e\n\u003cp id=\"_Toc338864677\"\u003e4.3 Pre-experiment\u003c/p\u003e\n\u003cp id=\"_Toc171994495\"\u003eWith technology increasingly blurring the line between interpersonal and human\u0026ndash;machine interaction, this study employs a scenario-based experiment to examine the psychological mechanisms involved, focusing on perceived value, perceived risk (mediators), and sense of interpersonal distance (moderator).\u003c/p\u003e\n\u003cp\u003eParticipants first completed demographic and emotional state measures (control variables) and were then randomly assigned to one of four conditions: interpersonal competence, interpersonal emotional, human\u0026ndash;machine competence, or human\u0026ndash;machine emotional. Using a scenario priming method, participants recalled and described past events matching the assigned condition. This strengthened recall accuracy and allowed cross-validation between event elements and manipulation items. Priming materials were uniform in structure and length to avoid measurement bias (Appendix A).\u003c/p\u003e\n\u003cp\u003eA pretest with 10 participants confirmed accurate recall and understanding. The main pre-experiment recruited 205 participants via Credamo; after excluding 25 participants (due to attention check failures or lack of prior AI experience), 180 remained (47 per group). Manipulation checks included a single-choice interaction type item and open-ended contextual details, with 98% recalling events from the past week. Interaction context was measured with adapted Stereotype Content Model scales (Aaker et al., 2010), with high reliability (Cronbach\u0026rsquo;s \u0026alpha; \u0026gt; .90).\u003c/p\u003e\n\u003cp\u003eOne-sample t-tests (5-point Likert, test value = 3) confirmed successful manipulation: emotional needs (M = 4.28, SD = 0.66, t = 18.56, p \u0026lt; .001) and competence needs (M = 3.97, SD = 0.73, t = 12.70, p \u0026lt; .001). No group mood differences were found, ruling out mood as a confounding factor.\u003c/p\u003e\n\u003cp\u003e4.4 Formal-experiment\u003c/p\u003e\n\u003cp\u003e（1）\u0026nbsp;Experiment design and procedure\u003c/p\u003e\n\u003cp\u003eIn the formal experiment, the measurement of the mediator and moderator variables was incorporated into the procedure. Both the mediator and moderator variables were treated as continuous variables. All other experimental procedures were identical to those employed in the pilot study.\u003c/p\u003e\n\u003cp\u003e（2）Data collection and manipulation check\u003c/p\u003e\n\u003cp\u003eThe sample size was estimated using G*Power 3.1, assuming a medium effect (f = 0.30), \u0026alpha; = 0.05, and 80% power, which required 237 participants (Faul et al., 2007). To offset attrition, 300 individuals were recruited via Credamo; after failing checks and withdrawals, 288 remained (72 per group; 124 males and 164 females). Most were 18\u0026ndash;26 (80.5%), and students (65.3%). Manipulation checks (single-choice and open-ended) confirmed accurate recall, with 80% of participants describing events that occurred within the past week, ensuring vivid responses. All scales showed high reliability (Cronbach\u0026rsquo;s \u0026alpha; \u0026gt; .9). One-sample t-tests indicated emotional (M = 4.25, SD = 0.68, t = 22.02, p \u0026lt; .001) and competence needs (M = 4.20, SD = 0.65, t = 22.09, p \u0026lt; .001) exceeded the midpoint, confirming effective manipulations. No mood differences were observed, and measures of value, risk, distance, and behavioral intention demonstrated strong psychometric properties.\u003c/p\u003e\n\u003cp\u003e（3）Results\u003c/p\u003e\n\u003cp\u003eBefore analyzing variance (ANOVA), Levene\u0026rsquo;s test for equality of variances was performed. Based on the p-value (p \u0026lt; .05), the null hypothesis of homogeneity was rejected, indicating heterogeneity of variances. Examination of residual and data distributions indicated approximate normality with low skewness. Consequently, variable transformation and robust standard error estimation were considered. After applying a square root transformation to the dependent variable, heterogeneity persisted. Therefore, the HC3 heteroskedasticity-consistent estimator was employed to address the potential underestimation of standard errors caused by heteroscedasticity. The advantage of this method lies in preserving the original model structure while adjusting the inferential results (i.e., p-values and confidence intervals) (MacKinnon \u0026amp; White, 1985).\u003c/p\u003e\n\u003cp\u003eThe results revealed that both social interaction type (p \u0026lt; .001, \u0026eta;\u0026sup2; = .065) and social interaction scenario (p \u0026lt; .05, \u0026eta;\u0026sup2; = .015) exerted significant main effects on social behavioral intention. Furthermore, the interaction effect between interaction type and scenario was significant (p \u0026lt; .01, \u0026eta;\u0026sup2; = .037), indicating that the influence of interaction type differed across scenarios. Among the control variables, neither gender nor pre-experiment mood had a significant effect on social behavioral intention, thereby eliminating potential confounding effects. H1 was thus supported.\u003c/p\u003e\n\u003cp\u003eGiven the significant interaction, simple effects analyses were conducted to explore between-group differences. Participants in the human\u0026ndash;machine group reported significantly higher social behavioral intention (M = 3.90, SD = 0.65, 95% CI [3.43, 3.69]) than those in the human\u0026ndash;human group (M = 2.93, SD = 0.65, 95% CI [2.80, 3.69]), F(1, 176) = 66.48, p \u0026lt; .001, \u0026eta;\u0026sup2; = .14. Participants in the competence scenario (M = 3.56, SD = 0.65, 95% CI [3.23, 3.49]) also reported significantly higher social behavioral intention than those in the warmth scenario (M = 3.13, SD = 0.65, 95% CI [3.00, 3.26]), F(1, 284) = 6.86, p \u0026lt; .01, \u0026eta;\u0026sup2; = .024. Independent samples t tests further indicated that, in the competence scenario, social behavioral intention in the human\u0026ndash;human group (M = 2.89, SD = 0.69) was significantly lower than that in the human\u0026ndash;machine group (M = 3.84, SD = 0.45), t = \u0026ndash;10.01, p \u0026lt; .001, 95% CI [\u0026ndash;1.14, \u0026ndash;0.76]. In the warmth scenario, social behavioral intention in the human\u0026ndash;human group (M = 2.96, SD = 1.00) was also significantly lower than that in the human\u0026ndash;machine group (M = 3.28, SD = 0.88), t = \u0026ndash;2.03, p = .044, 95% CI [\u0026ndash;0.63, \u0026ndash;0.01]. H2a and H2b were thus supported (see Figure 4).\u003c/p\u003e\n\u003cp\u003eA bootstrapping analysis was conducted using the PROCESS macro (Hayes, 2014) with 5,000 resamples and 95% confidence intervals. Given that the independent variables in the model were categorical with two factors (two levels each), Model 8 of PROCESS was selected. One factor, with two levels, was entered as the moderator, and dummy coding was applied to the independent variables (interpersonal = 0, human\u0026ndash;machine = 1; competence = 0, emotion = 1). Social behavioral intention served as the dependent variable, while perceived value and perceived risk were included as mediators to test the mediating effects.\u003c/p\u003e\n\u003cp\u003eThe results indicated that under the interaction between social scenario and social interaction type, the confidence intervals for both perceived value and perceived risk did not contain zero,suggesting that the mediating effects were significant. The mediation test pathways are presented in the corresponding table. The results of the mediation model test are summarized in\u0026nbsp;Table 4.Both H4and H5 were supported.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4. Results of the Mediation Model Test.\u003c/strong\u003e\u003c/p\u003e\n\u003cdiv align=\"\"\u003e\n \u003ctable border=\"1\" cellspacing=\"3\" cellpadding=\"0\"\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003ctd colspan=\"5\" style=\"width: 556px;\"\u003e\u003cstrong\u003eDependent Variable:Social Behavioral Intention\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\u003cstrong\u003eVariable\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eCoefficient\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003eSE\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003et\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u003cstrong\u003e95% CI\u003c/strong\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003eSocial interaction type\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.52\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.08\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e6.61***\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e[0.36, 0.67]\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eSocial scenario\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.47\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.08\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e6.00***\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e[0.31, 0.62]\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eSocial interaction type \u0026times; Social scenario\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e-0.44\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.11\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e-3.93***\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e[-0.66, -0.22]\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003ePerceived value\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.89\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.04\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e18.42***\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e[0.73, 0.91]\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003ePerceived risk\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e-0.13\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.04\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e-3.20**\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e[-0.21, -0.05]\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eF\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e156.42***\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003eR\u0026sup2;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e0.74\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003ctd\u003e\u0026nbsp;\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n \u003c/table\u003e\n\u003c/div\u003e\n\u003cp\u003eNote: p \u0026lt; .05 , p\u0026lt; .01 , p\u0026lt; .001 . SE = Standard Error; CI = Confidence Interval.\u003c/p\u003e\n\u003cp\u003eNotably, unlike traditional complete mediation, the dummy coding of social interaction type and social scenario in the mediation analysis results in different signs for categorical variables. The opposite signs of perceived value reflect a non-additive effect across combinations of interaction type and scenario. Specifically, the interaction term between social interaction type and social scenario (\u0026beta; = -0.34, p \u0026lt; .01) exerts an adverse effect on perceived value, which contrasts with the positive effect of perceived value on behavioral intention (\u0026beta; = 0.89, p \u0026lt; .001) (Hayes, 2018).\u003c/p\u003e\n\u003cp\u003eData analysis treating social interaction type and social scenario as independent variables further reveals that, in human\u0026ndash;machine interaction contexts (vs. interpersonal), users\u0026rsquo; perceived value increases significantly; however, this advantage is attenuated in emotional scenarios (\u0026beta; = -0.34, p \u0026lt; .01). Although emotional scenarios directly promote behavioral intention (\u0026beta; = 0.47, p \u0026lt; .001), their indirect effect via reducing perceived value partially offsets this positive influence.\u003c/p\u003e\n\u003cp\u003eA further analysis of the mediating effects of perceived value and perceived risk across different interaction types shows that, within the interpersonal interaction group, the mediation effect of perceived value is significant (LLCI = -0.52, ULCI = -0.09, excluding 0). Within the human\u0026ndash;machine interaction group, the mediation effect of perceived value is also significant (LLCI = -0.76, ULCI = -0.42, excluding 0), thereby supporting H4a and H4b. For perceived risk, the mediation effect in the interpersonal interaction group is significant (LLCI = -0.15, ULCI = -0.03, excluding 0), but not significant in the human\u0026ndash;machine interaction group (LLCI = -0.04, ULCI = 0.04, including 0), indicating that perceived risk has virtually no influence on behavioral intention in human\u0026ndash;machine interactions. Thus, H5a is supported, whereas H5b is not.\u003c/p\u003e\n\u003cp\u003eCompared with interpersonal interaction in ability scenarios, human\u0026ndash;machine interaction yields higher perceived value (M interpersonal\u0026ndash;ability= 3.30, M human\u0026ndash;machine\u0026ndash;ability= 3.83) and lower perceived risk (M interpersonal\u0026ndash;ability= 2.96, M human\u0026ndash;machine\u0026ndash;ability= 2.94), supporting H3a and H6a. In emotional scenarios, human\u0026ndash;machine interaction also leads to higher perceived value (M interpersonal\u0026ndash;emotional = 2.92, M human\u0026ndash;machine\u0026ndash;emotional = 3.12); however, H3b is not supported. It further produces lower perceived risk (M interpersonal\u0026ndash;emotional = 3.61, M human\u0026ndash;machine\u0026ndash;emotional= 2.96), supporting H6b.\u003c/p\u003e\n\u003cp\u003eTo examine the moderating effect of sense of interpersonal distance on the path from the independent variables to the dependent variable, the two categorical variables were entered as moderators. Based on 5,000 bootstrap resamples in PROCESS Model 3, the interaction term among social interaction type, social scenario, and sense of interpersonal distance had a confidence interval excluding zero (LLCI = 0.50, ULCI = 1.09), confirming a significant moderation effect and supporting H7.\u003c/p\u003e\n\u003cp\u003eFinally, differences in behavioral intention were compared between participants with varying levels of sense of interpersonal distance. Given the normal distribution of interpersonal distance scores (M = 3.02, SD = 0.87), and following prior research, the top 27% and bottom 27% were selected to form the high- and low-score groups, respectively. Scores below 2.31 were classified as a low sense of interpersonal distance, and scores above 3.54 as a high sense of interpersonal distance.\u003c/p\u003e\n\u003cp\u003eAs shown in the figures(figure 5 and figure 6), in the competence scenario, for youth with a high sense of interpersonal distance, human\u0026ndash;machine interaction is more effective than human\u0026ndash;human interaction in enhancing willingness to engage in social interaction. For youth with a low sense of interpersonal distance, there is a difference between human\u0026ndash;human and human\u0026ndash;machine interaction in promoting social interaction willingness, though the gap is relatively small, with human\u0026ndash;machine interaction showing a slight advantage.\u003c/p\u003e\n\u003cp\u003eIn the emotion scenario, for youth with a high sense of interpersonal distance, human\u0026ndash;machine interaction again outperforms human\u0026ndash;human interaction in promoting willingness to engage in social interaction. However, for youth with a low sense of interpersonal distance, there is a significant difference between the two interaction types: when the interaction partner is an artificial intelligence, social interaction willingness is significantly weakened. Thus, H7a is supported, while H7b is not supported.\u003c/p\u003e"},{"header":"5. Conclusions and Limitations","content":"\u003cp\u003eThis study examined youth socialization from a macro perspective and introduced the construct of \u0026ldquo;Sense of Interpersonal Distance.\u0026rdquo; The findings suggest that AI-mediated interaction may help reduce some of the negative experiences associated with human\u0026ndash;human interaction. For individuals with higher interpersonal distance, AI appeared to provide perceived value and reduced risk, thereby potentially fostering openness toward others. At the same time, participants with lower interpersonal distance tended to prefer human\u0026ndash;human interaction in emotionally oriented contexts but were more open to AI in competence-oriented tasks. These patterns indicate that AI has the potential to function as an enabling mediator in specific circumstances, though such effects should be interpreted cautiously and within the limitations of the study\u0026rsquo;s design.\u003c/p\u003e\n\u003cp\u003eTwo important risks should also be noted. First, the absence of embodiment constrains multisensory engagement, limiting the depth of interaction compared with face-to-face communication. Second, because AI systems are not yet embedded in everyday social networks, their supportive role may be temporary, potentially leading to mismatches between virtual experiences and real-world relationships.\u003c/p\u003e\n\u003cp\u003eSeveral limitations should be acknowledged. Human\u0026ndash;human interaction was simplified to dyadic exchanges, and competence- and emotion-oriented needs were treated separately, despite their overlap in practice. The study relied on memory-based, cross-sectional data, which may restrict causal interpretations. In addition, the sample was skewed toward young adults (18\u0026ndash;26 years) with a higher proportion of female respondents, which limits generalizability. These limitations highlight the importance of future research employing more diverse samples, longitudinal or multimethod approaches, and more ecologically valid social settings to further assess the role of AI in youth socialization.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgments\u003c/p\u003e\n\u003cp\u003eWe would like to thank all the young participants who generously shared their experiences during the study.\u003c/p\u003e\n\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThis research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.\u003c/p\u003e\n\u003cp\u003eCompeting interests\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e\n\u003cp\u003eEthical approval\u003c/p\u003e\n\u003cp\u003eThis study was reviewed and approved by the institutional ethics review body on December 9, 2024 (Approval No. 20250930), prior to the commencement of any data collection or informed consent procedures. All procedures were conducted in accordance with institutional guidelines and the principles of the Declaration of Helsinki.\u003c/p\u003e\n\u003cp\u003eInformed consent\u003c/p\u003e\n\u003cp\u003eInformed consent was obtained from all participants after ethical approval and prior to data collection. For the interview study, oral consent was obtained before each session conducted via Tencent Meeting, beginning on January 9, 2025. For the survey study, written informed consent was obtained electronically via Wenjuanxing and Credamo, beginning on March 10, 2025. Participants were informed of the study purpose, procedures, voluntary nature of participation, and their right to withdraw at any time without penalty.\u003c/p\u003e\n\u003cp\u003eData availability\u003c/p\u003e\n\u003cp\u003eThe interview materials involved in this project contain personal privacy information and therefore cannot be made publicly available. Aggregated anonymized data may be available from the corresponding author upon reasonable request.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eThese authors contributed equally to this work.QH and BJ contributed to the conceptualization and design of the study. QH was responsible for data collection, data analysis, and drafting of the manuscript. Both authors read and approved the final manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eAaker J, Vohs KD, Mogilner C (2010) Nonprofits are seen as warm and for-profits as competent: Firm stereotypes matter. J Consum Res 37(2):224\u0026ndash;237\u003c/li\u003e\n \u003cli\u003eAlkis Y, Kadirhan Z, Sat M (2017) Development and validation of social anxiety scale for social media users. Comput Human Behav 72:296\u0026ndash;303\u003c/li\u003e\n \u003cli\u003eAjzen I (1985) From intentions to actions: A theory of planned behavior. In: Kuhl J, Beckmann J (eds) Action control: From cognition to behavior. Springer, Berlin, pp 11\u0026ndash;39\u003c/li\u003e\n \u003cli\u003eBakhtin M (1981) The dialogic imagination: Four essays (ed: Holquist M; trans: Emerson C, Holquist M). University of Texas Press, Austin\u003c/li\u003e\n \u003cli\u003eBauman Z (2000) Liquid modernity. PolityPress, Cambridge\u003c/li\u003e\n \u003cli\u003eBeebe SA, Beebe SJ, Redmond MV, Salem-Wiseman L (2002) Interpersonal communication: Relating to others. Allyn \u0026amp; Bacon, Boston\u003c/li\u003e\n \u003cli\u003eBerkman L (1983) The assessment of social networks and social support in older people. J Am Geriatr Soc 31(12):743\u0026ndash;749\u003c/li\u003e\n \u003cli\u003eBrandtzaeg PB, Skjuve M, F\u0026oslash;lstad A (2022) My AI friend: How users of a social chatbot understand their human\u0026ndash;AI friendship. Human Communication Research 48(3):404\u0026ndash;429\u003c/li\u003e\n \u003cli\u003eBrown TJ, Churchill GA, Peter JP (1993) Improving the measurement of service quality. J Retail 69(1):127\u0026ndash;139\u003c/li\u003e\n \u003cli\u003eChen QQ, Park HJ (2021). How Anthropomorphism Affects Trust in Intelligent Personal Assistants Industrial Management \u0026amp; Data Systems 121(12):2722\u0026ndash;2737\u003c/li\u003e\n \u003cli\u003eChen R (2020) An analysis of the relationship between perceived risk in WeChat Moments and self-disclosure. News Frontline 3:50\u0026ndash;51\u003c/li\u003e\n \u003cli\u003eCronin JJ, Brady MK, Hult GTM (2000). Assessing the effects of quality, value, and customer satisfaction on consumer behavioral intentions in service environments. J Retail 76(2):193\u0026ndash;218\u003c/li\u003e\n \u003cli\u003eDeng Y (2024) Waiting: A sociological analysis of a universal temporal experience. Sociological Studies 39(06):176\u0026ndash;199, 230\u003c/li\u003e\n \u003cli\u003eDeng Z (2018) The modernization-oriented transformation of Chinese youth\u0026rsquo;s social mentality since the reform and opening up. China Youth Studies (4):5\u003c/li\u003e\n \u003cli\u003eDing W, Peng K (2020). Correction and revision of the Chinese version of the Interpersonal Trust Scale. Psychology Monthly 15(6):4\u0026ndash;5, 7\u003c/li\u003e\n \u003cli\u003eEng L, Nordstr\u0026ouml;m P, Strandberg G (2021). Incorporating compassion into compassion fatigue: The development of a new scale. J Clin Psychol 77(9):2109\u0026ndash;2129\u003c/li\u003e\n \u003cli\u003eEpley N, Akalis S, Waytz A, Cacioppo JT (2008) Creating social connection through inferential reproduction: Loneliness and perceived agency in gadgets, gods, and greyhounds. Psychol Sci 19(2):114\u0026ndash;120\u003c/li\u003e\n \u003cli\u003eFan H, Gao X (2025) Analysis of trust issues and construction of trust mechanisms in complex social systems of human\u0026ndash;machine interaction. Jiangsu Social Sciences:65\u0026ndash;70\u003c/li\u003e\n \u003cli\u003eFaul F, Erdfelder E, Lang A-G, Buchner A (2007) G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39(2):175\u0026ndash;191\u003c/li\u003e\n \u003cli\u003eFenigstein A, Scheier MF, Buss AH (1975) Public and private self-consciousness: Assessment and theory. J Consult Clin Psychol 43(4):522\u0026ndash;527\u003c/li\u003e\n \u003cli\u003eFiske ST, Cuddy AJC, Glick P (2007) Universal dimensions of social cognition: Warmth and competence. Trends in CognitiveScience11(2):77\u0026ndash;83\u003c/li\u003e\n \u003cli\u003eFiske ST (2018) Stereotype content: Warmth and competence endure. Curr Dir Psychol Sci 27(2):67\u0026ndash;73\u003c/li\u003e\n \u003cli\u003eGiddens A (1991) The consequences of modernity. Stanford University Press, Stanford\u003c/li\u003e\n \u003cli\u003eHan B-C (2015) The burnout society (trans: Butler E). Stanford University Press, Stanford\u003c/li\u003e\n \u003cli\u003eHayes AF (2018) Introduction to mediation, moderation, and conditional process analysis: A regression-based approach, 2nd edn. Guilford Press, New York\u003c/li\u003e\n \u003cli\u003eHair JF, Black WC, Babin BJ, Anderson RE (2013) Multivariate data analysis, 7th edn. Pearson, Harlow\u003c/li\u003e\n \u003cli\u003eHe Q, He Y, Li G, Chen X, Li L, Wang Y, Liu J, Ma D, Liu Q (2022) Qualitative study on the willingness of community-dwelling older adults to use companion robots. Journal of Nursing 29(19):6\u0026ndash;10\u003c/li\u003e\n \u003cli\u003eHew T-S, Kadir SLSA (2016) Predicting the acceptance of cloud-based virtual learning environment: The roles of Self-Determination and Channel Expansion Theory. Telematics Inform 33(4):990\u0026ndash;1013\u003c/li\u003e\n \u003cli\u003eHorney K (1937) The neurotic personality of our time. W. W. Norton, New York\u003c/li\u003e\n \u003cli\u003eKang Y (2021) The formation mechanism of \u0026ldquo;lifelong singlehood\u0026rdquo; under individualization: A case study of urban youth. China Youth Studies (12):33\u0026ndash;39\u003c/li\u003e\n \u003cli\u003eLei B (2024) The temporal dimension of relationships: A study on contemporary youth\u0026rsquo;s values in interpersonal communication. Youth Exploration (3):70\u0026ndash;79\u003c/li\u003e\n \u003cli\u003eLevinas E (1969) Totality and infinity: An essay on exteriority (trans: Lingis A). Duquesne University Press, Pittsburgh\u003c/li\u003e\n \u003cli\u003eLi H, Zhou Y, Li J (2021) Are users leaving social media? A study on the influencing factors of social media fatigue based on perceived value. Chinese Journal of Journalism \u0026amp; Communication 43(12):120\u0026ndash;141\u003c/li\u003e\n \u003cli\u003eLien-Gieschen T (1993) Validation of social isolation related to maturational age: Elderly. Nursing Diagnosis 4(1):37\u0026ndash;44\u003c/li\u003e\n \u003cli\u003eLin J, Lin S, Turel O, Xu F (2020) The buffering effect of flow experience on the relationship between overload and social media users\u0026rsquo; discontinuance intentions. Telematics Inform 49:101374\u003c/li\u003e\n \u003cli\u003eLin S, Ye L (2019) Human\u0026ndash;machine interaction and reshaping: Intelligent robots as the \u0026ldquo;sixth medium.\u0026rdquo; Journal of Journalism and Communication Studies 26(10):87\u0026ndash;104, 128\u003c/li\u003e\n \u003cli\u003eLuo L, Wang B, Wang X (2021) Function, relationship, and philosophy: A study of user\u0026ndash;smart speaker interaction from the perspective of human\u0026ndash;machine communication. Global Media Journal 8(3):102\u0026ndash;118\u003c/li\u003e\n \u003cli\u003eMacKinnon JG, White H (1985) Some heteroskedasticity-consistent covariance matrix estimators with improved finite sample properties. J Econometrics 29(3):305\u0026ndash;325\u003c/li\u003e\n \u003cli\u003eMarx K, Engels F (1975) Collected Works of Karl Marx and Friedrich Engels, Vol. 3 (March 1843\u0026ndash;August 1844). London: Lawrence \u0026amp; Wishart. ISBN 0853152934.\u003c/li\u003e\n \u003cli\u003eMende M, Scott ML, van Doorn J, Grewal D, Shanks I (2019) Service robots rising: How humanoid robots influence service experiences and elicit compensatory consumer responses. J Mark Res 56(4):535\u0026ndash;556\u003c/li\u003e\n \u003cli\u003eNass CI, Moon Y, Carney P (2000). Machines and mindlessness: Social responses to computers. J Soc Issues 56(1):81\u0026ndash;103\u003c/li\u003e\n \u003cli\u003eNicholson NR (2009) Social isolation in older adults: An evolutionary concept analysis. J Adv Nurs 65(6):1342\u0026ndash;1352\u003c/li\u003e\n \u003cli\u003ePeng L (2022) The future of human\u0026ndash;machine communication and interaction. Journal of Social Sciences of Hunan Normal University 51(5):12\u0026ndash;22\u003c/li\u003e\n \u003cli\u003ePeters JD (1999) Speaking into the air: A history of the idea of communication. University of Chicago Press, Chicago\u003c/li\u003e\n \u003cli\u003eRasouli S, Gupta G, Nilsen E, Dautenhahn K (2022) Potential applications of social robots in robot-assisted interventions for social anxiety. Int J Soc Robotics 14(5):1167-1198\u003c/li\u003e\n \u003cli\u003eRempel JK, Holmes JG, Zanna MP (1986) How do I trust thee? Psychology Today 20(2):28-34\u003c/li\u003e\n \u003cli\u003eShi J, Mo X, Sun Z (2012) Application of the content validity index in scale development. Journal of Central South University (Medical Sciences) 37(2):152\u0026ndash;155\u003c/li\u003e\n \u003cli\u003eSimmel G (2004) The philosophy of money, 3rd enlarged edn (trans: Frisby D, Bottomore T). Routledge, London\u003c/li\u003e\n \u003cli\u003eStrauss A, Corbin J (1990). Basics of qualitative research: Grounded theory procedures and techniques. Sage, Newbury Park\u003c/li\u003e\n \u003cli\u003eTurkle S (2017) Alone together: Why we expect more from technology and less from each other. Basic Books, New York\u003c/li\u003e\n \u003cli\u003e\u0026Uuml;ng\u0026uuml;ren \u0026Uuml;, Tekin \u0026Ouml;A (2023). The effects of social disconnectedness, social media addiction, and social appearance anxiety on tourism students\u0026rsquo; career intentions: The moderating role of self-efficacy and physical activity. J Hospitality Leisure Sport Tourism Educ 33:100463\u003c/li\u003e\n \u003cli\u003eVenkatesh V, Davis FD, Morris MG (2000) A theoretical extension of the technology acceptance model: Four longitudinal field studies. Manage Sci 46(2):186\u0026ndash;204\u003c/li\u003e\n \u003cli\u003eWang X, Hu P (2022) Sense of boundaries: The social needs of modern youth and their construction. China Youth Studies (10):72\u0026ndash;79\u003c/li\u003e\n \u003cli\u003eXiang A, Xu K (2023). Why and how humans and machines interact: Theoretical origins, paradigm shifts, and future trends. Global Media Journal 10(5):88\u0026ndash;105\u003c/li\u003e\n \u003cli\u003eYin Y, Jia N, Wakslak CJ (2024) AI can help people feel heard, but an AI label diminishes this impact. Proc. Natl. Acad. Sci. USA 121(14):e2319112121\u003c/li\u003e\n \u003cli\u003eYuan G (2023) Digital media, uncertainty, and emotional governance in risk communication. Theory and Reform 3:134\u0026ndash;143, 160\u003c/li\u003e\n \u003cli\u003eZeithaml VA (1988) Consumer perceptions of price, quality, and value: A means\u0026ndash;end model and synthesis of evidence. J Mark 52(3):2\u0026ndash;22\u003c/li\u003e\n \u003cli\u003eZhang J, Lin B (2025) \u0026ldquo;Self-determined loneliness\u0026rdquo;: Reflection on the \u0026ldquo;self-logic\u0026rdquo; behind the social patterns of young people in the era of individualization. Inner Mongolia Social Sciences 46(1):173\u0026ndash;181\u003c/li\u003e\n \u003cli\u003eZhang M, Ye Y, Xu P (2021) Scale development and validation of social media perceived value. Journalism and Communication Review 74(5):28\u0026ndash;42\u003c/li\u003e\n \u003cli\u003eZhou H, Zhang L, Luo X, et al. (2017) Modifying the Autism Spectrum Rating Scale (6\u0026ndash;18 years) to a Chinese context: An exploratory factor analysis. Neurosci Bull 33(2):175\u0026ndash;182\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"human–AI interaction, interpersonal communication, sense of interpersonal distance, technologically mediated intimacy","lastPublishedDoi":"10.21203/rs.3.rs-7495693/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7495693/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eIn today\u0026rsquo;s \u0026ldquo;liquid modernity\u0026rdquo;, rapid technological change and hyper-mediated relationships often leave youth\u0026rsquo;s emotional needs unmet, fostering fatigue, social pressure, and loneliness. Yet, as interaction is intrinsic to human sociality, intelligent agents are emerging as viable alternatives to traditional social partners. Using grounded theory, this study identifies a prevalent \u0026ldquo;sense of interpersonal distance\u0026rdquo; among youth. It examines its moderating role in a 2 (social type: interpersonal vs. human\u0026ndash;AI) \u0026times; 2 (scenario type: competence vs. emotional) experiment. Findings reveal that interpersonal distance significantly shapes evaluations of social types and that AI can function as a social actor in both competence- and emotion-oriented contexts. Such human\u0026ndash;AI connections may offer new forms of technologically mediated intimacy, shifting sociality from co-presence to co-existence.\u003c/p\u003e","manuscriptTitle":"Breaking Emotional Barriers: How AI Mitigates Interpersonal Distance and Fosters Social Capability Growth","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-10 14:03:50","doi":"10.21203/rs.3.rs-7495693/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"fd8072af-5e4d-4f3f-aee4-cc2f7693b42e","owner":[],"postedDate":"December 10th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":59373415,"name":"Business and commerce/Information systems and information technology"},{"id":59373416,"name":"Biological sciences/Psychology"},{"id":59373417,"name":"Social science/Psychology"}],"tags":[],"updatedAt":"2026-02-13T09:56:41+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-10 14:03:50","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7495693","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7495693","identity":"rs-7495693","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}