Supporters of Marcus Endicott’s Patreon can access weekly or monthly video consultations on this topic.
Chen, H., Shao, B., Yang, X., Kang, W., & Fan, W. (2024). Avatars in live streaming commerce: The influence of anthropomorphism on consumers' willingness to accept virtual live streamers. Computers in Human Behavior, 156, Article 108216.
The study examines whether making virtual live streamers more human-like increases consumer acceptance in live-streaming commerce and finds that higher anthropomorphism raises willingness to accept because viewers feel less psychological distance and more trust, with this mediated effect appearing for utilitarian products but not hedonic ones. Its evidential value is an experimental test of a specific psychological pathway rather than a simple correlation, but its main limitation is that the outcome is stated willingness to accept in an online scenario experiment rather than observed purchasing behavior in live commerce.
Chen, J., Pan, L., Zhou, R., & Jiang, Q. (2024). Shaping and optimizing the image of virtual city spokespersons based on factor analysis and entropy weight methodology: A cross-sectional study from China. Systems, 12(2), Article 44.
The study examines which traits most strongly shape the preferred image of virtual city spokespersons in China and concludes that visual design dominates: “design elements” rank first, with realistic appearance, tactile quality, and aesthetically proportioned bodies treated as the highest-priority attributes, followed by anthropomorphism, emotionalization, evolution, culturalism, and narrativity, while reliability and interactivity are comparatively low priorities. Its evidential value is a 512-case empirical ranking of user preferences that yields a concrete design hierarchy, but the conclusions are limited by the lack of age differentiation, the omission of regional variation, and the fact that the proposed design strategies were not tested for real-world effectiveness.
Chen, S., Zhang, D., Shi, W., Ding, X., & Chang, L. (2024). Exploring the efficacy of interactive digital humans in cultural communication. In G. Zhai, J. Zhou, H. Yang, L. Ye, P. An, & X. Yang (Eds.), Digital multimedia communications (pp. 220–239). Springer.
The chapter examines whether interactive virtual digital humans can improve historical and cultural communication and concludes that stronger interactivity increases audience engagement and absorption of cultural content while helping overcome the one-way, monotonous character of traditional dissemination. Its main evidential value is identifying a concrete communicative benefit of digital humans beyond simple novelty, but its conclusions are limited to improved engagement and content uptake rather than demonstrating longer-term retention, interpretation, or broader cultural impact.
Chen, Y., & Li, X. (2024). Expectancy violations and discontinuance behavior in live-streaming commerce: Exploring human interactions with virtual streamers. Behavioral Sciences, 14(10), Article 920.
The study examines why consumers stop engaging with virtual streamers in live-streaming commerce and finds that perceived failures in professionalism, empathy, and responsiveness increase distrust and dissatisfaction, which then drive discontinuance, with dissatisfaction exerting the stronger effect. Its main evidential value is specifying a concrete negative-consumption pathway for virtual streamer failure rather than merely showing reduced acceptance, but its conclusions are limited by self-reported survey data from Chinese users in a single cultural setting, which constrains behavioral and cross-context generalizability.
Chen, Z., & Skey, M. (2024). "I produce songs for her…. In this way, I gradually know her more. The more I know her, the more I like her": Using Collins' model of interactive ritual chains to study the case of virtual idol fandom in China. Convergence, 30(2), 841–859.
The study examines how fans of Luo Tianyi build attachment through digital fan labor and community participation, and concludes that creating songs, videos, and other promotional content functions as an interactive ritual that generates emotional energy, deepens perceived intimacy with the virtual idol, and strengthens solidarity within the fandom. Its main evidential value is showing that virtual-idol fandom depends not only on consumption but also on fan co-production as a source of affective commitment and collective identity, while its central limitation is that conclusions rest on a single Chinese case and therefore have limited generalizability across other virtual idols, fan cultures, or platform environments.
Demeng, W., Shaoxian, Z., & Hongyu, Q. (2024). Exploration and use of large model-driven digital humans for popularizing earthquake science. CT Theory and Applications.
The paper examines a large-model digital human for earthquake science communication, centered on the QuakeGPT system delivered through the WeChat “XiaoQ” interface, and concludes that a domain-specific model can enrich educational content, improve answer accuracy, and raise public engagement and learning around earthquake knowledge and disaster-prevention guidance. Its main evidential value is a concrete applied example of how large-model digital humans can be used in a high-stakes public-education setting, but its claims are limited by the absence of reported controlled evaluation, comparative benchmarking, or observed behavioral outcome data beyond asserted gains in engagement and understanding.
Gao, J., & Wang, Y. (2024). The study of Chinese style virtual digital human in the dissemination of traditional culture. Journal of Education, Humanities and Social Sciences, 41, 100–105.
The paper examines Chinese-style virtual digital humans as vehicles for transmitting traditional culture and concludes that their strongest potential lies in making cultural communication more youth-oriented, internationally legible, and sustainable, but only if developers overcome weak immersive experience, visual and conceptual homogeneity, and the frequent failure to maintain continuous high-quality content output. Its main evidential value is a concrete synthesis of the field’s claimed developmental bottlenecks and proposed strategic directions, while its central limitation is that the argument is primarily conceptual and example-driven rather than supported by systematic empirical testing of audience effects or cultural outcomes.
Gao, J., Zhao, X., Zhai, M., Zhang, D., & Li, G. (2024). AI or human? The effect of streamer types on consumer purchase intention in live streaming. International Journal of Human-Computer Interaction. Advance online publication.
This study examines whether AI versus human live-streaming hosts change consumer purchase intention and finds that human streamers produce higher stated purchase intention because viewers perceive greater intimacy and responsiveness; novelty seeking moderates the effect, with the human advantage weakening among high novelty-seeking consumers. Its main evidential value is showing two specific mechanisms behind the advantage of human hosts, but the conclusions are constrained by survey-based purchase-intention data from 387 users rather than observed purchasing behavior.
Gao, X., Liu, D., & Zhang, J. (2024). 多模态数字人建模、合成与驱动综述 [A survey of multimodal digital human modeling, synthesis, and driving]. 中国图象图形学报 [Journal of Image and Graphics], 29(9), 2494–2512.
The review synthesizes multimodal digital human research across head animation, body animation, and avatar construction, and concludes that implicit neural representations, especially NeRF-based methods, now outperform explicit models in realism for head generation, Transformer architectures are displacing RNNs in body animation, diffusion-based methods are increasingly important for motion and portrait generation, and the field’s main constraints remain real-time 3D rendering, multimodal fusion, limited large-scale 3D datasets, and unresolved ethical and privacy issues. Its main evidential value is in drawing concrete technical conclusions across several subfields rather than merely cataloguing work, but its central limitation is that those conclusions come from narrative synthesis of representative studies rather than a systematic quantitative comparison, so relative performance claims remain indicative rather than decisively benchmarked.
Han, J. (2024). Application research and scenario analysis of virtual digital humans in science popularization and teaching. Knowledge Management Forum, 9(5), 460–476.
The study examines how virtual digital humans can be applied in science popularization and teaching and concludes that their anthropomorphism and interactivity can support more innovative presentation of scientific content and more immersive, participatory learning experiences. Its main evidential value is in linking technical discussion to concrete educational and popularization scenarios rather than treating digital humans only as a general media trend, but its central limitation is that the argument is built mainly from a technical review, case analysis, and limited demonstration, so it does not establish comparative evidence about learning effectiveness, audience understanding, or sustained educational outcomes across settings.
Han, X., Li, J., & Ma, L. (2024). 政务服务数字人：理论诠释、实践困境与因应路径 [Government service digital humans: Theoretical interpretation, practical difficulties, and response paths]. 电子科技大学学报(社科版) [Journal of UESTC (Social Sciences Edition)].
The study interprets government-service digital humans through an information-service-system framework and assesses their role in China’s public-service digitalization; it concludes that such systems offer clear advantages in information richness, service affordance, and interaction synchrony but are constrained in practice by weak construction and management mechanisms, unclear responsibility allocation, insufficient technical standards, poor data sharing, and limited application scenarios, leading the authors to argue for governance and implementation reforms rather than simple expansion. Its evidential value lies in combining a structured conceptual account with an empirical diagnosis of current deployment problems, but its main limitation is that the analysis centers on institutional and design conditions rather than direct measurement of citizen outcomes, service quality, or comparative performance effects.
He, Q., Li, L., Li, D., Peng, T., Zhang, X., & Cai, Y. (2024). From digital human modeling to human digital twin: Framework and perspectives in human factors. Chinese Journal of Mechanical Engineering, 37, Article 9.
The review traces the shift from traditional digital human modeling to human digital twin in human factors and proposes a unified framework linking a physical twin, a virtual twin, and data-driven connections between them; its main conclusion is that adding multi-source real-time data and AI could extend ergonomics from static modeling toward continuous monitoring, timely feedback, and bidirectional human-centered services in Industry 5.0. Its evidential value is as an integrative conceptual synthesis that clarifies how HDT differs from earlier DHM and where the field is heading, but its main limitation is that the framework is not empirically validated, so claims about practical performance and industrial impact remain prospective rather than demonstrated.
Ji, M., Chen, X., Wei, S., Liu, Q., & Sun, J. (2024). Understanding the role of virtual anchor–brand image fit in virtual live streaming.
The study examines how fit between a virtual anchor’s image, voice, and language style and a brand’s image shapes purchase intention in virtual live streaming, and finds that image-fit and voice-fit increase purchase intention, figurative language works better for warm brands, literal language does not improve purchase intention for competent brands, and the effects operate mainly through processing fluency and perceived affinity, with processing fluency not mediating the language-style result. Its evidential value lies in distinguishing multiple fit dimensions and identifying different psychological pathways rather than treating virtual anchor fit as a single construct, but its main limitation is that the outcome is purchase intention rather than observed buying behavior, which restricts claims about actual sales effects.
Ji, X. (2024). Influence of virtual live streamers' credibility on online sales performance. SAGE Open, 14(3).
The study examines whether source-credibility dimensions of virtual live streamers predict e-commerce sales performance and finds that trustworthiness, attractiveness, and expertise are positively associated with sales, whereas interactivity is negatively associated, which the paper attributes to the still limited ability of virtual streamers to interact in a human-like way. Its evidential value lies in using behavioral data from 300 Taobao virtual live-streaming rooms rather than self-reported attitudes, but its main limitation is that credibility is inferred through platform-specific behavioral proxies in an observational design, which constrains causal interpretation and broader generalization beyond that setting.
Jin, J., & Sharudin, S. A. (2024). Research on the design aesthetic characteristics of Chinese digital virtual products in the 21st century. Journal of Social Science and Humanities, 6(9), 94–99.
The paper examines 21st-century Chinese digital virtual products as an aesthetic formation shaped by digital interactivity and illustrated through the animated film Storm Rider Clash of the Evils; it concludes that the strongest design tendency is the fusion of traditional Chinese cultural symbols and philosophies with minimalist, functional, and sometimes futuristic visual design, producing more localized yet globally legible digital forms. Its main evidential value is in naming a concrete cluster of claimed aesthetic traits and linking them to a worked cultural example, but the study’s central limitation is that the argument is largely descriptive and case-based, with broad field-level conclusions drawn from conceptual discussion and a single illustrative example rather than systematic comparison or user-based evidence.
Leo-Liu, J., & Wu-Ouyang, B. (2024). A "soul" emerges when AI, AR, and Anime converge: A case study on users of the new anime-stylized hologram social robot "Hupo." New Media & Society, 26(7), 3810–3832.
Examining user experience around Hupo, an anime-stylized hologram social robot in China, the study finds that anime aesthetics and AR did not solve weak conversational AI but helped users reinterpret those weaknesses through characterization, gamification, and idolization, reducing uncanny-valley discomfort and supporting perceptions that the character had a “soul”; its main value is the concrete claim that pop-cultural framing can temporarily compensate for limited AI affordances in human-machine attachment, while its main limitation is that the evidence comes from a small, self-reported single-case user sample without producer-side evidence or stronger observational depth.
Levy-Landesberg, H., & Cao, X. (2024). Anchoring voices: The news anchor's voice in China from television to AI. Media, Culture & Society. Advance online publication.
Using China’s AI television news anchors to revisit a longer history of broadcast vocal training, the study argues that the anchor voice in China was already constructed decades earlier as measurable, disciplinable, and programmable, making AI anchoring less a break with television than an extension of preexisting technicized voice regimes; its main value is the historically specific linkage it draws between synthetic anchors and earlier broadcasting pedagogy in China, while its main limitation is that the evidence rests mainly on pedagogical standards and training discourse rather than on observed newsroom practice or audience response.
Li, C., & Huang, F. (2024). The impact of virtual streamer anthropomorphism on consumer purchase intention: Cognitive trust as a mediator. Behavioral Sciences, 14(12), Article 1228.
Focusing on how different forms of virtual-streamer human likeness shape buying responses in live e-commerce, the study finds that behavioral, cognitive, and emotional anthropomorphism strengthen cognitive trust, that appearance and emotional anthropomorphism can also raise purchase intention directly, and that cognitive trust is a central pathway linking anthropomorphic design to purchase intention; its main evidential value is that it differentiates four anthropomorphic dimensions instead of treating anthropomorphism as a single trait, but its chief limitation is that the findings come from a cross-sectional questionnaire of Chinese consumers, so they support attitudinal association more clearly than causal or behavioral purchasing effects.
Li, N., Xuan, C., & Chen, R. (2024). Different roles of two kinds of digital coexistence: The impact of social presence on consumers' purchase intention in the live streaming shopping context. Journal of Retailing and Consumer Services, 80, Article 103890.
Distinguishing the social presence created by the streamer from that created by other viewers in live-stream shopping, the study finds that both forms raise purchase intention through better consumer experience, but the streamer’s presence is markedly more influential, while other viewers’ presence can intensify loneliness rather than provide companionship, making social copresence in commerce unequal rather than uniformly beneficial; its main evidential value is that the conclusions are drawn from a large national Douyin dataset rather than a small convenience sample, but its chief limitation is that the analysis is confined to one Chinese platform and a cross-sectional design, which restricts causal inference and broader generalization across retail settings.
Li, Q., & Liu, Y. (2024). 教育虚拟数字人标准体系设计及其路径规划 [Design and path planning of the educational virtual digital human standard system]. 现代教育技术 [Modern Educational Technology], (7), 70–80.
Defining the standardization problem for educational virtual digital humans, the study concludes that effective sector development requires a dedicated standards architecture built from case-derived application needs and organized around six planning principles—top-level design, future orientation, application drive, industry–academia collaboration, cross-disciplinary integration, and international cooperation—rather than ad hoc technical deployment; its main evidential value is that the proposed framework is grounded in both educational use cases and a survey of existing standardization conditions, but its chief limitation is that it remains a prescriptive system design whose practicality and completeness are not validated through implementation or comparative testing.
Li, S., & Chen, J. (2024). Virtual human on social media: Text mining and sentiment analysis. Technology in Society, 78, Article 102666.
Analyzing Chinese social-media discourse around virtual humans, the study finds that public attention concentrates on technological and industrial development, virtual idols and virtual streamers, and related corporate investment and policy, with overall sentiment still more positive than negative but negative sentiment rising over time around service failure, the uncanny valley, ethical risk, and technological unemployment; its main evidential value is that it grounds these conclusions in a large longitudinal corpus of public Weibo discussion rather than small-sample self-report, but its chief limitation is that platform-specific posting behavior captures visible online discourse, not broader population attitudes or actual interaction outcomes.
Li, Y., Lu, K., & Zhang, W. (2024). Research on the application of virtual anchors based on artificial intelligence technology in live streaming e-commerce. Journal of Electrical Systems, 20(3), 1232–1240.
Examining AI-based virtual anchors in live-streaming e-commerce, the study concludes that adoption is being propelled by lower labor cost and greater operational stability than human streamers, but current use is constrained by weak social functionality, a limited core consumer base, low differentiation, and information-security governance risks, so effective deployment depends on better human–virtual streamer coordination, appropriate product-category selection, stronger specialist training, and more plural governance arrangements; its main evidential value is that it identifies concrete commercial and governance bottlenecks in applied use rather than treating virtual anchors as a purely technical novelty, but its chief limitation is that the argument rests on literature review and case analysis rather than direct comparative or behavioral evidence on performance outcomes.
Liao, W., & Wang, W. (2024). 虚拟数字人技术趋势与未来产业生态 [Technology trends and future industrial ecosystem of virtual digital humans]. 中国传媒科技 [Science and Technology for China's Mass Media], (10), 56–59, 76.
Surveying technological and industrial development of virtual digital humans, the paper argues that rapid advances in new information technologies are pushing the field toward stronger AI-driven multimodal interaction, affective intelligence, and greater functional autonomy, with expanding use across entertainment, education, and healthcare, and concludes that the sector’s next stage will be driven chiefly by the joint advance of AI in content generation and VR in interactive presentation; its main evidential value is that it links technical evolution to an emerging application ecosystem rather than treating digital humans as an isolated graphics problem, but its chief limitation is its high-level, forward-looking synthesis, which offers little systematic comparative evidence for sectoral differences or for how far these projected capabilities are already realized in practice.
Liao, X., Liao, Y., She, S., Zeng, Y., & Chen, Y. (2024). AI-driven digital humans: Like humans but not human? A systematic review of research on the uncanny valley.
Reviewing research on uncanny-valley effects in AI-driven digital humans, the paper concludes that human likeness remains a double-edged design feature, because greater anthropomorphism can support engagement while near-human embodiments repeatedly produce discomfort when appearance, motion, voice, or inferred agency fall out of alignment, leaving the phenomenon as a persistent but context-dependent problem rather than a settled single-mechanism rule; the review’s main limitation is that its synthesis rests on heterogeneous prior studies spanning different artificial agents and inconsistent measures of eeriness, affinity, and acceptance, which weakens the force of any unified general conclusion.
Liu, L., Gao, L., Guo, Y., Li, C., Lü, L., Wang, B., Zhang, J., & Chen, X. (2024). 《中国图象图形学报》数字人建模、生成与渲染技术专栏简介 [Introduction to the special column on digital human modeling, generation, and rendering technology]. 中国图象图形学报 [Journal of Image and Graphics], 29(9), 2439–2440.
Framing current digital-human research around the problem of achieving high visual fidelity and natural motion, the paper concludes that the field’s most active lines of progress span 3D face imaging and reconstruction, multimodal head and body animation, facial appearance recovery and rendering, motion generation, and clothed-human reconstruction, while the main unresolved bottlenecks remain realism, efficiency, data quality, editability, and robust reconstruction in unconstrained settings; its evidential value lies in condensing the substantive takeaways of eight papers in the special column, including six reviews and two research articles, into a compact picture of where the field is advancing and where it still struggles, but its main limitation is that it is an editorial synthesis rather than an independent empirical or critical evaluation of those claims.
Liu, S., & Geng, Z. (2024). Ethnic attire exhibition system utilizing digital human technology. In Proceedings of the World Conference on Intelligent and 3-D Technologies.
Addressing digital presentation of ethnic costume, the paper reports that a MetaHuman- and Unreal Engine–based exhibition system can produce highly realistic, visually improved digital displays of traditional attire and therefore has practical promise for cultural presentation, but it also concludes that virtual fitting remains the key unresolved weakness, with further improvement needed in fit precision and interactive functions; the paper’s main evidential value is that it documents an implemented system rather than a purely conceptual proposal, while its chief limitation is that the reported claims center on realism and application potential without comparative performance evidence or demonstrated user-outcome validation.
Luo, L., & Kim, W. (2024). How virtual influencers' identities are shaped on Chinese social media: A case study of Ling. Global Media and China.
Examining how Weibo users construct Ling’s public identity, the study finds that responses cluster around three roles—a CGI figure, a marketable endorser, and, more distinctively, a cultural ambassador—so Ling’s appeal in China depends not only on brand utility and parasocial intimacy but also on the capacity to attract nationalist feeling even while her human likeness still triggers authenticity doubts and uncanny discomfort; its main evidential value is that this conclusion is drawn from a substantial body of audience commentary rather than marketer claims, but the inference remains constrained by a single-influencer, single-platform design that omits visual content and limits cross-platform or cross-case comparison.
Mao, X. (2024). "虚拟数字人"的"今生"和"来世" [The "present life" and "afterlife" of virtual digital humans]. 电子商务评论 [E-Commerce Letters], 13(3), 4884–4889.
The paper examines the current development and projected trajectory of virtual digital humans and concludes that the field’s most important substantive direction is not mere realism but expansion toward more capable intelligent interaction, finer-grained personalization, stronger knowledge and learning capacity, and wider cultural and commercial integration, while privacy, data security, ethical governance, legal regulation, and labor displacement remain the main constraints on sustainable growth; its value is as a compact synthesis of major developmental claims and risk themes across application domains, but because it is a non-empirical narrative overview built largely from secondary discussion, it does not provide direct comparative or causal evidence for the magnitude of those benefits or harms.
Meng, M., Hou, X., Wang, Y., Liu, H., & Zhao, Y. (2024). PolyMotion-7K: A multimodal-driven polyglot avatar motion dataset. In Proceedings of the International Forum on Digital TV and Wireless Multimedia Communications.
This paper introduces PolyMotion-7K as a multilingual, multimodal avatar-motion dataset intended to improve motion diversity and fine-grained expressive quality for avatar generation across languages and cultures, and reports that it supports high-quality multilingual avatar generation in an application case on Beijing Central Axis intangible-cultural-heritage videos, suggesting practical value for cross-cultural communication and digital heritage presentation; its main evidential value is that it contributes a concrete dataset plus a downstream demonstration rather than a purely conceptual proposal, but its most important limitation is that the validation remains narrow and application-specific, with no broad benchmark comparison or external user evaluation to establish general performance across tasks or populations. The source also indicates a bibliographic discrepancy: the chapter appears under IFTC 2024 but was first published online in 2025, and the accessible record lists a sixth author, Qin Yuan.
Min, T., Sun, T., Tan, C., & Lai, F. (2024). Practices and strategies for enhancing video-based dissemination of scientific journal articles through virtual digital human technology. Chinese Journal of Scientific and Technical Periodicals, 35(9), 1178–1185.
The study examines the use of virtual digital human technology to convert scientific-journal paper content into video and concludes that the approach can deliver lower-cost production with stronger dissemination effects, while the main practical requirements are livelier content, closer alignment with technological iteration, stronger editorial review, deeper editor–AI integration, and new-media talent development. Its evidential value is that it reports an applied publishing case rather than a purely conceptual argument, but because the analysis is built around a platform-specific practice context and descriptive evaluation, it offers limited comparative evidence about whether the reported gains would generalize across journals, platforms, or audiences.
Peng, Y., Wang, Y., Li, J., & Yang, Q. (2024). Impact of AI-oriented live-streaming e-commerce service failures on consumer disengagement—Empirical evidence from China. Journal of Theoretical and Applied Electronic Commerce Research, 19, 1580–1598.
The study examines how service failures in AI-oriented live-streaming e-commerce drive consumer disengagement in China and finds that information, functional, system, interaction, and aesthetic failures increase discontinuance behavior, with disappointment and emotional exhaustion acting as key mediating mechanisms and platform type shaping the strength of those effects; its evidential value is a comparatively specific account of the negative pathway from failed virtual-streaming service experience to withdrawal, but its main limitation is that the evidence comes from cross-sectional self-report data, which constrains causal inference and does not show actual long-term disengagement behavior.
Qiu, A. (2024). The influence of virtual anchor characteristics on consumers' purchase intention in the context of e-commerce live broadcasting. In Proceedings of the International Conference on Digital Economy and Computer Application.
The study examines how perceived professionalism, attractiveness, interactivity, and anthropomorphism of virtual anchors shape purchase intention in e-commerce live broadcasting and reports positive effects that operate partly through flow experience and perceived trust, indicating that stronger purchase intention is linked not only to favorable anchor traits themselves but also to the immersion and credibility those traits generate; its main evidential value is a focused empirical account of trait-to-intention mechanisms for virtual anchors, but because it relies on questionnaire-based regression evidence from a single consumption context, it provides limited causal and behavioral validation.
Shao, Z. (2024). Understanding the switching intention to virtual streamers in live streaming commerce: Innovation resistances, shopping motivations and personalities. Journal of Research in Interactive Marketing, 19(3), 333–357.
The study examines why consumers might switch from human to virtual streamers in live-streaming commerce and finds that switching is configuration-dependent, with six distinct combinations of innovation barriers, shopping motivations, and personality factors shaping intention rather than any single universal driver; its main evidential value is showing that lower resistance and stronger motivation can support multiple viable paths to adoption instead of one dominant explanation, but its most important limitation is that it analyzes stated switching intention from an online UK survey rather than observed switching behavior in live commercial settings.
Song, L., Chow, O. W., Na, M., & Gill, S. S. (2024a). Virtual idol and youth identity: The impact of mainstream culture on the virtual idol online fanbase community in China. Studies in Media and Communication, 12(2), 100–115.
This study examines how mainstream culture reshapes Chinese virtual-idol fan communities and concludes that state and mainstream-media incorporation of virtual idols has pushed a formerly subcultural fan formation toward mainstream acceptance and, more specifically, from post-subcultural group identity toward national identity; its evidential value lies in showing a concrete identity shift linked to cultural mainstreaming within an online fan community, but because the analysis is an interpretive virtual ethnography centered on one national context and fan setting, it offers limited causal leverage and uncertain generalizability beyond similar cases.
Song, L., Chow, O. W., Na, M., & Gill, S. S. (2024b). Virtual music idol, the Chinese Leitkultur and the making of the world's first virtual "Red Diva." Asian Social Science, 20(2), 24–34.
This paper examines how Luo Tianyi was transformed from a Japanese-influenced ACGN subcultural virtual singer into a state-legible mainstream “Red Diva,” and concludes that her significance lies less in fandom alone than in her use as a cultural bridge through which official media and commercial actors route mainstream Chinese values toward Generation Z while recoding subcultural participation as national-cultural alignment; its main evidential value is a concrete, detailed account of how one high-profile virtual music idol is positioned within Chinese leitkultur, but because the study is a single interpretive case with little direct audience evidence or comparative testing, its claims about broader effects on youth identity and mainstreaming remain suggestive rather than firmly demonstrated.
Su, C., Luo, X., Liu, Z., Kang, J., Hao, M., Xiong, Z., & Huang, C. (2024). Privacy-preserving pseudonym schemes for personalized 3D avatars in mobile social metaverses. In 2024 6th International Conference on Electronics and Communication, Network and Computer Technology (ECNCT) (pp. 375–380). IEEE.
This study examines privacy-preserving personalization for 3D avatars in mobile social metaverses by introducing avatar pseudonyms intended to protect both profile identity and digital identity during edge-supported social interaction; it reports that the proposed framework, paired with a new Privacy of Personalized Avatars metric and a Stackelberg-game plus deep-reinforcement-learning allocation strategy, is effective and feasible in simulations for improving privacy while managing pseudonym resources; its main limitation is that the evidence is simulation-based, so the study does not establish real-world performance, user behavior effects, or deployment robustness in live metaverse systems.
Sun, Y., Yang, J., & Chen, Z. (2024). 新华社AI主播新闻报道研究 [Research on Xinhua News Agency AI anchor news reporting]. 新闻前哨 [Press Outpost], (3), 32–35.
This study examines Xinhua’s AI anchor as a newsroom reporting tool and argues that its main substantive value lies in faster, continuous, cross-platform, and increasingly cross-scene news output, with the clearest gains in timeliness and production efficiency rather than in deeper, more affectively persuasive journalism; its evidential value is that it documents a concrete mainstream-media deployment at Xinhua, but because it is a brief, single-outlet, largely descriptive case study, it provides limited comparative or audience-effect evidence for how far those advantages generalize.
Tan, C., Lai, Y., & Wu, J. (2024). A study on the developmental process and application mode of AI digital humans in cartoon style: A case study of an offline doll machine store in Shenzhen. In Proceedings of the International Conference on Human-Computer Interaction.
This conference study examines how a cartoon-style AI digital human can be designed for use in a Shenzhen offline claw-machine store and concludes that user satisfaction depends chiefly on natural voice interaction, accurate dialogue understanding, expression recognition, and entertaining performance, whereas presence-triggered greeting has little measurable effect; its main evidential value is that it converts a concrete retail case into specific service-design priorities for cartoon digital humans rather than treating digital-human deployment at a purely conceptual level, while its chief limitation is that a single-site case study centered on one storefront and satisfaction-oriented user-need analysis limits how far the findings can be generalized to other sectors, audiences, or digital-human styles.
Tang, J. (2024). Broadcasting your virtual self: Exploring the authenticity construction and subjectivity of VTubers in China. Global Media and China.
This ethnographic study examines how Chinese VTubers construct authenticity through avatarized live-streaming and concludes that authenticity is produced through sustained alignment of gesture, voice, movement, anime-cultural affect, and technical control with a fictional persona rather than through revelation of an offline “real self”; its main evidential value is the clear argument that VR hardware actively mediates idealized self-expression and virtual intimacy instead of serving as a neutral tool, while its chief limitation is that an interpretive qualitative design constrains generalization beyond the cases observed.
Tu, L. (2024). A virtual future: ODR for virtual idol copyright and fan disputes (SSRN Scholarly Paper No. 5052767).
The paper addresses copyright and fan disputes arising around AI-driven virtual idols and considers online dispute resolution as a governance response to those conflicts; its central conclusion is that disputes in this sector require mechanisms better aligned with digitally mediated, cross-platform, and rapidly scaling creative ecosystems than conventional resolution alone, making ODR a plausible fit for at least part of the problem; its evidential value is chiefly a legal-conceptual framing of an emerging dispute domain, while its main limitation is the apparent absence of direct empirical testing, comparative case analysis, or demonstrated outcomes from actual ODR use in virtual-idol disputes.
Wang, K., Kim, K., & Tang, J. (2024). Analyzing the design of virtual humans in Chinese museums: A case study of "Ai Wenwen." Asia Pacific Journal of Convergence Research Interchange, 10(3), 561–572.
The study examines how virtual humans for Chinese museums should be designed, using Ai Wenwen to ground a framework spanning appearance, media form, and service system; it concludes that museum virtual humans should be defined by more than surface likeness alone and that, relative to human staff, their design value lies in combining these three dimensions into a clearer role-specific model for communication and service in Chinese museums; its evidential value is a concrete design framework tied to an actual museum case, but its main limitation is that the claims are validated through comparative and case analysis rather than direct testing of visitor response, learning, or engagement outcomes.
Wang, K., Wu, J., Sun, Y., Chen, J., Pu, Y., & Qi, Y. (2024). Trust in human and virtual live streamers: The role of integrity and social presence. International Journal of Human-Computer Interaction, 40(23), 8274–8294.
The study compares how trust forms around human versus virtual live streamers in live-streaming commerce; it finds that both can generate trust and purchase intention, but integrity is especially consequential for human streamers, whereas social presence is especially consequential for virtual streamers, with social presence also directly shaping purchase intention more clearly for virtual streamers; its evidential value is stronger than a single-shot study because it reports multi-study evidence on differentiated trust mechanisms, but its main limitation is likely reduced ecological realism because those mechanisms are tested in controlled streamer scenarios rather than in fully natural platform behavior.
Yang, X., Fang, Y., Wang, L., & Raga Jr, R. C. (2024). Digital human intelligent interaction system based on multimodal pre-training mode. Applied Artificial Intelligence, 38(1), 2405953.
The published article presents a digital-human interaction system built around an improved UniLM-based multimodal pre-training approach; it reports consistently lower perplexity across five text lengths, supporting the paper’s core conclusion that the modified model improves text generation and thus the system’s conversational capability relative to comparison models; its evidential value lies in concrete comparative performance results within a working system, but its main limitation is that validation appears to center on model-output metrics in a prototype setting rather than direct evidence of better real-user interaction, broader multimodal performance, or downstream application outcomes.
Wang, Q., Li, B., Tan, C. W., Chong, A. Y. L., Liu, X., Chen, R., Zhang, Y., Xie, X., & Wang, Z. (2024). Virtual idols and virtual streamers: From the perspective of design features. In E. Y. Li, P. Y. K. Chau, & C. M. K. Cheung (Eds.), Proceedings of the 24th International Conference on Electronic Business (pp. 89–97).
The review distinguishes virtual idols from virtual streamers by examining which design features are linked to different marketing outcomes; it finds that both rely chiefly on visual nonverbal cues, especially appearance, but virtual idols more often use celebrity-related cues to build fan relationships while virtual streamers more often emphasize emotional capacity, interactivity, and short-term commercial aims such as purchase intention; its main value is a concrete field-level synthesis that clarifies an often-blurred conceptual boundary, but its conclusions are constrained by a small and heterogeneous underlying literature and by dependence on prior studies rather than direct empirical testing.
Wen, S., & Wang, H. (2024). 人工智能情境下虚拟主播的视觉设计 [Visual design of virtual anchors in the context of artificial intelligence]. 中阿科技论坛(中英文) [China-Arab States Science and Technology Forum], (3), 49–53.
The study considers visual design for AI virtual anchors in media communication, focusing on the relation between visual “inner” and “outer” design elements; it concludes that visual design should be guided holistically and strengthened through specific value-enhancing design recommendations rather than reduced to surface appearance alone, with the stated aim of improving image construction and supporting sustainable digital-economy development; its evidential value is mainly conceptual, because the paper appears to offer design reflection and prescriptive recommendations rather than reported empirical testing of design outcomes or audience effects.
Weng, W. (2024). How anthropomorphic features of virtual influencers affect Generation Z purchase intentions through credibility. In 2024 4th International Conference on Social Development and Media Communication (SDMC 2024) (pp. 55–61). Atlantis Press.
This conference paper examines whether anthropomorphic virtual influencers increase Generation Z purchase intention by strengthening perceived credibility, finding that anthropomorphism positively predicts both credibility and purchase intention and that credibility is the stronger predictor and mediates the relationship. Its evidential value is limited by a quota-sampled self-report survey of 208 respondents who already followed virtual influencers, which narrows generalizability and does not support strong causal inference.
Wu, Y. (2024). Virtual idols: A new cultural industry and fan structure. Transformative Works and Cultures, 43.
This symposium study examines the pandemic-era rise of virtual idols in China and the fan formations built around them, concluding that pandemic-driven social separation helped establish virtual idols as a distinct entertainment model while producing a new fan subculture; its value is as a concise interpretive claim connecting industrial emergence and fandom structure in the Chinese context, but its evidential weight appears limited by the format’s broad, high-level treatment rather than extensive systematic demonstration.
Xiang, J., & Fu, S. (2024). Understanding public opinion and discussion dynamics of digital humans on social media: An analysis of sentiment, themes and user characteristics. Library Hi Tech. Advance online publication.
The study analyzes social-media discourse on digital humans to identify how attention, sentiment, topics, and user characteristics cluster, and concludes that public attention is rising over time but unevenly distributed geographically, with discussion concentrating around live streaming, service use, cultural entertainment, and digital avatars while different user groups show significantly different sentiment levels and topical priorities; its main evidential value is a concrete, platform-based picture of how digital humans are being socially framed and by whom, but its conclusions are limited by reliance on Weibo discourse, which constrains generalization beyond one platform and an observational snapshot of online discussion.
Xie, J., Wu, H., Liu, K., Cui, Y., & Zhang, X. (2024). Is virtual streamer useful? Effect of streamer type on consumer brand forgiveness when streamers make inappropriate remarks. Journal of Retailing and Consumer Services, 79, Article 103899.
The study tests whether human-powered virtual streamers protect brands better than real human streamers after inappropriate remarks, and finds higher consumer brand forgiveness in the virtual-streamer condition because consumers feel less empathy for the offending streamer; the effect is further bounded by context, appearing when brand reputation or streamer influence is high and disappearing when those are low, which gives the paper clear value as evidence for reputation-shielding in livestream crises but limits any broader claim that virtual streamers are generally superior.
Xie, L., Gu, D., Zhang, K., Tan, L., & Tang, L. (2024). A report of research on the digital human and digital medicine in China. Digital Medicine.
This report reviews digital-human and digital-medicine development in China and concludes that technological progress has not only reshaped surgical practice and healthcare service modes but also driven broad application of digital-human research across medical education, clinical medicine, aerospace, automotive manufacturing, sports, and artistic performance, positioning the field as an emerging cross-domain undertaking rather than a narrowly clinical one. Its main value is a concise national overview that ties technical development to concrete application breadth, while its main limitation is its broad descriptive scope, which offers little systematic comparative evidence on effectiveness, adoption depth, or outcome differences across domains.
Xie, M., & Zhou, Z. (2024). 基于CiteSpace的人工智能虚拟数字人发展研究热点与趋势分析 [Analysis of research hotspots and trends in AI virtual digital human development based on CiteSpace]. 科技和产业 [Science Technology and Industry], 24(5), 147–153.
This bibliometric study examines Chinese scholarship on AI virtual digital humans and concludes that the field has developed into a thematically clustered research area with identifiable hotspots and discernible trend lines rather than a scattered set of isolated discussions. Its main value is a compact map of where scholarly attention is concentrating and how the topic is being organized intellectually, while its main limitation is that bibliometric evidence captures publication and keyword patterns, so the analysis can indicate research emphasis but cannot by itself establish the real-world effectiveness, maturity, or social consequences of digital-human systems.
Xie, X., & Zeng, N. (2024). 机遇与挑战：虚拟数字人在传媒业的应用和未来发展 [Opportunities and challenges: The application and future development of virtual digital humans in the media industry]. 新闻爱好者 [Journalism Lover].
This paper examines virtual digital humans across the media production chain and concludes that their main contribution is greater production efficiency, richer immersive presentation, and broader intelligent upgrading of media workflows, but present systems remain limited in genuine intelligence and interactivity and introduce substantial risks around information authenticity, privacy, intellectual property, ethical governance, recommendation-driven narrowing, and uncanny-valley effects. Its main value is a concise industry-level synthesis that links benefits and risks across collection, production, presentation, and distribution rather than isolating a single use case, while its main limitation is its broad conceptual scope and reliance on illustrative examples instead of direct empirical tests of audience effects or organizational outcomes.
Xiong, J., & Jia, J. (2024). 论元宇宙视角下虚拟数字人的法律主体地位 [On the legal subject status of virtual digital humans from the metaverse perspective]. 东方法学 [Oriental Law].
This legal study examines whether and for which kinds of virtual digital humans metaverse-era law should recognize independent subject status, concluding that subjecthood should not extend to all forms uniformly but is most defensible for company- or team-operated performer-type figures and intelligent autonomous figures without a directly attributable natural-person counterpart, so they can act, bear liability, assert rights, and contract in their own name within a supporting framework of legislation, identity authentication, and insurance. Its main value is a differentiated doctrinal argument that ties legal status to type and application scene rather than treating all virtual digital humans as either mere objects or full natural-person equivalents, while its main limitation is that the case for legal subjecthood is primarily normative and anticipatory rather than grounded in systematic judicial practice or empirical evidence of how such arrangements would function in operation.
Xu, B., & Li, D. (2024). The technological evolution of digital humans and their application in Chinese media. In ICADI '24: Proceedings of the 2024 International Conference on Artificial Intelligence, Digital Media Technology and Interaction Design. ACM.
This conference paper traces the development of digital-human technology and its uses in Chinese media, concluding that digital humans have already moved from technical novelty into broad practical deployment across entertainment, service, and education contexts and are becoming a consequential media-production and communication tool rather than a marginal experiment. Its main value is a compact application-level synthesis linking technological evolution to concrete media uptake in China, while its main limitation is its broad descriptive scope, which leaves little systematic comparative evidence on effectiveness, audience response, or sector-specific outcomes.
Yan, H. (2024). 算法驱动型虚拟数字人涉侵权纠纷的规范解决路径 [Normative resolution paths for infringement disputes involving algorithm-driven virtual digital humans]. 重庆大学学报(社会科学版) [Journal of Chongqing University (Social Science Edition)], 30(2), 182–194.
This legal analysis addresses infringement disputes involving algorithm-driven virtual digital humans across training-data acquisition, generated outputs, and interactive deployment, concluding that the disputes should be separated by protected interest and actor role rather than treated as a single AI category, with differentiated attribution rules for copyright, personality-rights, and data harms, including stricter treatment when sensitive personal information is involved. Its main value is a structured doctrinal framework that links distinct stages of the virtual-human pipeline to specific liability principles in Chinese law, while its main limitation is that the argument is primarily normative and case-driven rather than grounded in a large body of adjudicated disputes.
Yan, X., Mo, T., & Zhou, X. (2024). The influence of cultural differences between China and the West on moral responsibility judgments of virtual humans. Acta Psychologica Sinica, 56(2), 161.
This study compares Chinese and Western judgments of virtual humans’ moral responsibility after immoral acts and finds that Chinese participants assigned more responsibility, and greater willingness to punish, to virtual humans than Western participants did, whereas judgments of real humans did not differ in the same way and the cultural gap was mediated by higher perceived mental capacity, especially perceived experience, attributed to virtual humans. Its main value is multi-study evidence that moral evaluation of virtual humans varies systematically by culture rather than simply by whether a human or AI controls the character, while its main limitation is reliance on scenario-based judgments and reported punitive intentions rather than observed real-world behavior.
Yao, L., Tugiman, N., & Sharipudin, M. N. S. (2024). Virtual human influencers in live streaming commerce on social media platforms: Exploring parasocial interactions with consumers in China. SEARCH Journal of Media and Communication Research, 16(4), 47–59.
This study examines how virtual human influencers generate parasocial interaction in Chinese live-streaming commerce and concludes that personalization, interactive narrative, rapid feedback, and cultural localization are the main conditions under which they strengthen familiarity, trust, loyalty, and purchase intention, although participants still generally regarded human influencers as more authentic and emotionally compelling, leaving virtual figures most persuasive in niche or strongly stylized contexts. Its main value is a concrete synthesis of where virtual influence appears effective in live commerce rather than treating appeal as automatic, while its main limitation is its reliance on secondary literature and qualitative focus-group evidence from one market rather than direct behavioral or comparative outcome data.
Yao, R., Qi, G., Wu, Z., Sun, H., & Sheng, D. (2024). Digital human calls you dear: How do customers respond to virtual streamers' social-oriented language in e-commerce livestreaming? A stereotyping perspective. Journal of Retailing and Consumer Services, 79, Article 103872.
This study examines whether social-oriented language used by virtual streamers in e-commerce livestreaming increases purchase intention and concludes that the effect is conditional rather than general: social-oriented phrasing raises purchase intention for experience products, operates through perceived warmth and perceived competence, and is further shaped by streamer type, with the authors treating warmth and competence as jointly important rather than a simple trade-off. Its main value is controlled evidence that language strategy for digital humans should be matched to product category instead of assumed universally effective, while its main limitation is that the evidence comes from experimental manipulations and a very small focus group in a Chinese livestreaming context, which limits external validity for real market behavior across platforms and cultures.
Yin, L., & Kaixing, W. (2024). Research of Guochao style in Chinese virtual idol design. International Journal of Internet, Broadcasting and Communication, 16(3), 85–97.
This paper examines how Guochao aesthetics are incorporated into Chinese virtual idol design and concludes that combining traditional Chinese cultural elements with contemporary visual styling gives virtual idols a more distinctive identity and greater cultural value while serving as a differentiation strategy in a competitive market and as a vehicle for digitally mediated cultural promotion. Its main value is a concrete design-level synthesis of the features the authors treat as characteristic of Guochao virtual idols, while its main limitation is that the evidence is interpretive and conceptual rather than based on direct testing of audience response, market impact, or comparative performance outcomes.
Yuan, J. E. (2024). Making virtual celebrity: Platformization and intermediation in digital cultural production. International Journal of Cultural Studies. Advance online publication.
This study analyzes cultural intermediation in China’s virtual celebrity sector across multiple platforms, arguing against a simple disintermediation model and concluding from two cases that platformized cultural production is structurally paradoxical: platforms centralize value capture while also enabling more distributed forms of value creation, commodification, and labor division, including prosumption loops and authenticity work against technological standardization and overproduction. Its main value is a concrete field-level account of how virtual celebrity is organized across converging industries rather than by isolated creators, while its main limitation is the narrow evidential base of two cases from one national sector.
Yuan, Q., & Gao, Q. (2024). Being there, and being together: Avatar appearance and peer interaction in VR classrooms for video-based learning. International Journal of Human-Computer Interaction, 40(13), 3313–3333.
This study tests whether avatar appearance and two peer-interaction designs in asynchronous VR classrooms change learners’ sense of togetherness and learning during video-based study, finding that interaction design mattered more than appearance: real-time peer interaction produced the strongest social presence but was associated with poorer learning outcomes, whereas timeline-anchored interaction increased social presence without reducing performance, and avatar appearance showed no significant main effect. Its main evidential value is experimental evidence that social connectedness in VR learning can be improved through interaction structure rather than more human-like visual design, while its main limitation is that the evidence comes from a narrowly defined video-based VR classroom context, which constrains generalization to other subjects, platforms, and live instructional settings.