How real is real enough? Unveiling the diverse power of generative AI‐enabled virtual influencers and the dynamics of human responses

Abstract

Virtual influencers in various forms are capturing a growing share of ad spend from human influencers. Followers respond to virtual influencers much as if they were human, with engagement rates and measures of trust and source credibility that rival their human counterparts. However, there is an acute need for more nuanced un derstanding of the differential characteristics of user engagement with human in fluencers and the many emerging forms, types and interactional characteristics of virtual influencers. We conduct an exploratory study of three parts. First, through an indepth review of the existing literature, we delineate the implications of unsettled taxonomies of virtual influencers by function and form, and we outline a revised typology. Second, our secondary review of virtual influencer literature, trade, and industry sources conceptualizes divergent factors influencing the persuasive capa bility of human and virtual influencers while identifying intersecting research themes. From this synthesis we induce suggestions for future research and practice. Finally, we assess, refine and adjust our framework through depth interviews with leading expert practitioners to generate six key findings to guide future researched‐ backed virtual influencer practice and research. KEYWORDS artificial intelligence, generative AI, influencer marketing, metaverse, social media, virtual influencer 1 | INTRODUCTION Recent precipitous growth in the use of virtual influencers by brands reflects their rising commercial power. Some virtual influencers generate audience reach and engagement rates that exceed those of their human counterparts (Muniz et al., 2023) and brands from the fashion, beauty, automotive, and entertainment industries, among others, have turned to virtual influencers as a solution for their influencer marketing needs (Sands, Campbell, et al., 2022). For ex ample, virtual influencer Lu do Magalu has more than 6 million fol lowers and earns an estimated $33,000 per Instagram post (AIT News Desk, 2024). In just 6 months, more than 60 brands sponsored a virtual influencer called Lil Miquela (de Brito Silva et al., 2022). According to our assessment of recent industry reports (Supporting Information S1: Appendix 1), chief marketing officers are allocating up to 30% of their marketing budget to virtual influencers to cater to 3124 | wileyonlinelibrary.com/journal/mar Psychol Mark. 2024;41:3124–3143. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2024 The Author(s). Psychology & Marketing published by Wiley Periodicals LLC. the 58% of consumers (and the 75% of Generation Z consumers) who follow at least one virtual influencer (Emplifi, 2023). This begs several questions for researchers and practitioners. At present, brands have little research‐backed information to inform the efficacy of virtual influencers and the contexts or under lying processes in which they can generate more or less engagement and persuasion than human counterparts. There are some self‐ evident advantages to using virtual influencers over human influen cers (Meng et al., 2025). A virtual influencer is fully controllable (Lee & Yuan, 2023) and will not deviate from the script or from the agreed contract. Virtual influencers do not face the same risks as human influencers when it comes to unpredictable behavior, potential scandals, accusations of being inauthentic (Claeys et al., 2024), vio lations of advertising rules, or other personal issues (AlRabiah et al., 2022; Shen et al., 2022). However, two important considera tions remain. First, the creators behind virtual influencers must thoroughly understand and comply with current regulations when developing scripts and storylines. Second, virtual influencers en hanced by generative artificial intelligence may pose new risks through potentially inappropriate real‐time interactions. The use of virtual influencers offers vast creative and techno logical possibilities. A virtual influencer offers a broad palette of possibilities to keep the content engaging and interesting to users by adjusting the virtual influencer visually, changing scripts and story lines, and mixing virtual and real imagery. One of the most intriguing aspects of virtual influencers operationally is that they are not merely computer‐generated imagery versions of human influencers. Virtual influencers can be designed to be hyper‐realistic and humanlike, looking and acting like human influencers (Appel et al., 2020), but they can be equally successful in the form of nonhuman (Yan et al., 2024), yet engagingly funny abstract characters, such as Nobody Sausage “who” has more than 30 million followers across several social media platforms (Koles et al., 2024) and monetizes this popularity with lucrative brand partnerships (Bersola, 2023). Despite the growing number of partnerships between brands and virtual influencers, research on this phenomenon is still in its early stages (Sands, Campbell, et al., 2022). There are still misconceptions about the operational mechanisms in play in user engagement. A clear terminology can help prevent misinterpretations of virtual in fluencer research among academics and marketing practitioners. The fact that virtual influencers subsist merely as lines of code does not seem to inhibit user responses that imply trust, authenticity and credibility can be invoked in the human user (Wang et al., 2019; Wu et al., 2022). However, the fundamental difference between human and virtual influencers is often overlooked or disregarded by users, brands, and even some researchers. Thus, theories applied in the context of human influencers cannot be unproblematically extended to the realm of virtual influencers, even though, in commercial practice, brands' use of virtual influencers often parallels their use of human influencers. For example, Prada uses a virtual influencer called Candy to promote its perfumes, while KFC has created a younger, virtual version of Colonel Sanders to act as the face of the brand (Sands, Ferraro, et al., 2022). While the latter example is not of an influencer as such, it does illustrate that brands are sometimes using virtual characters in the same way as they might use humans. Virtual char acters can update and modernize the brand for younger audiences (Franke et al., 2023). However, using virtual influencers can also lead to controversy. For instance, Calvin Klein's campaign featuring Lil Miquela and a human model was criticized for being insincere or “queerbaiting,” and creators of black virtual influencers have been accused of commodifying and exploiting race (Taylor, 2023). Some virtual influencers, such as Shudu and Kami, have been specifically designed to represent minority groups, achieving notable success in this regard. Such instances highlight the extent to which using virtual influencers in the same way as using human influencers remains an experimental area for brands. Anthropomorphism is the perception of a nonhuman artifact (e.g., a brand, object, or animal [Han, 2021]) as humanlike in its appearance and/ or behavior (Nowak & Rauh, 2005). These humanlike qualities—similarities to humans in behavior, form or appearance and communication style— can influence how relatable and engaging the user perceives the artifact to be. This concept has been used to understand human responses to product types, brands (Huaman‐Ramirez et al., 2022), virtual agents, and chatbots (Munnukka et al., 2022). When users respond to virtual influ encers by trusting them, finding them a credible source of information or simply by liking them, they often respond as if the virtual influencer is human, since it would be nonsensical to like or trust a machine as if it were human and capable of human virtues such as charm or integrity. Therefore, the anthropomorphic dimension is important for under standing the ways in which virtual influencers generate engagement and solicit commercial success. Accordingly, our study explores and delineates the varieties of anthropomorphic virtual influencer design. One note of caution should be sounded, as we refer to the “uncanny valley” phe nomenon, whereby virtual entities that are regarded by users as too human‐like can be perceived negatively by users as creepy and alienating. This is an important factor to consider with differential audience responses to virtual influencers. Although the literature on virtual influencers is growing in response to the increased interest from brands, much of it derives theoretical frameworks from the human influencer domain and reveals conflicting findings concerning their efficacy (Supporting Information S1: Appen dix 2). Studies on virtual influencers have mainly focused on three areas: (1) how virtual influencers' characteristics such as the uncanny valley, anthropomorphism (Gutuleac et al., 2024), homophily (Wan et al., 2024), attractiveness (Kim & Park, 2023), and sensory deficiencies (Li et al., 2023) impact audience engagement, credibility, and persuasion (Ozdemir et al., 2023) compared to human influencers: (2) the fit of product/con text with virtual influencers (Belanche et al., 2024): and (3) the realism or degree of anthropomorphic mimesis of virtual influencers' behaviors, including emotional messaging (Quach et al., 2024), social presence (Yan et al., 2024), and faulty endorsements (Zhao et al., 2024). A systematic literature review by Byun and Ahn (2023) revealed a scarcity of peer‐ reviewed studies and empirical publications in this area. The limited availability of high‐impact studies and key informants' insights constrained their review, hindering deeper discussions on the ethical issues and potential value creation in the future development of virtual influencers. Thus, the nascent and fragmented scholarship of virtual influencer mar keting needs advancement, especially in assessing and mapping how users' engagement with virtual influencers mirrors or diverges from that with human influencers. Additionally, virtual influencers enhanced by generative artificial intelligence necessitate more research and real‐world data or industry expert input to understand how advanced virtual influ encers can create value for brands and consumers across virtual spaces, and what concerns should be addressed when the younger generations tend to be the target audience. This study aims to address the lack of clarity and specificity in virtual influencer research and practice by refining the conceptualization of vir tual influencers and their typology, delineating the differences between virtual and human influencers, and identifying future research avenues. These goals are achieved through comprehensive literature reviews and interviews with senior experts in the field. Specifically, our research questions are: How do human and virtual influencers differ in characteristics, strategies, engagement, and outcomes within the persuasive communication framework? What typology of virtual influencers emerges, how can they be categorized based on form, content, agency, and function, and what are their marketing implications? What are the value creation or value erosion out comes that could be brought about by the future phase of virtual influencers in the immersive medium era? This research offers three contributions to the virtual influencer marketing literature. First, we engage in a deep review of the research literature to detail conflicting approaches to the taxonomy of virtual influencers and the resultant lack of clarity for research and practice. Hence, we advance the nascent understanding of the emerging definitions and classifications of virtual influencers to minimize misinterpretations of virtual influencer research. Second, we evaluate research‐based conceptualizations of virtual influencer through secondary sources including virtual influencer trade press, industry reports and an investigation of virtual influencers on social media platforms. From this synthesis we offer suggestions for future research and practice (see Table 2). Accordingly, this study explicates distinctions and commonalities of human and virtual influencers, and identifies their implications concerning theoretical challenges, factors influencing their efficacy, and research themes through a refined conceptual framework. Finally, we conduct a qualitative study based on indepth interviews with leading expert practitioners to refine and evaluate the practical traction of our induced suggestions from Table 2. We offer six key propositions for future research and practice and conclude with the implications and limitations of the study. 2 | CONCEPTUAL BACKGROUND The term ‘virtual’ serves as a useful umbrella concept for computer‐ generated entities that do not exist in the physical world, while ‘in fluencer’ denotes the primary commercial activity of both virtual entities and their human counterparts. Hence, a computer‐generated entity that is created to mimic the activities of human influencers can be termed a virtual influencer. We adopt a working definition of a virtual influencer as a digital or virtual character that is created and managed by human operators through advanced technologies (e.g., computer‐generated imagery, deepfaking, some artificial intelligent ele ments, and machine learning) that enable the performance of human influencer‐like tasks in a virtual environment. Currently, both researchers and practitioners use different terms which carry different connotations. Words such as robot, avatar, AI or artificial intelligence, computer‐generated imagery or CGI, and metaverse influencers are used to emphasize different types of vir tual influencer functionality and design. This diversity in terms is understandable given the rapid evolution of the virtual influencer scene and its highly diverse manifestations, but it also creates potential ambiguity or confusion (Miao et al., 2022). Virtual influen cers are seldom representations of actual living people (although some human influencers/celebrities do occasionally use computer‐ generated self‐imagery), hence they are not avatars (Franke et al., 2023). They are not robots since they have no physical mass and are not automated (Igarashi et al., 2024). Some are artificial intelligence enhanced, but many are not, and some operate in the metaverse, but most do not, at least not yet. Virtual influencers have more complex technological integrations, purpose, and functionality than computer‐generated imagery (Koles et al., 2024). Virtual influ encers' purpose is operationalized by their creative team which designs their visual appearance, scripts and storylines (Lee & Yuan, 2023). Hence, virtual influencers do not have the autonomy that is normally associated with the term influencer. While “virtual influencer” effectively captures a broad category of entities, explor ing the diverse types and functions of these digital personalities is crucial to establish a more robust taxonomic foundation for advancing our understanding. We summarize different labels and taxonomies used for virtual influencers in the research literature, and their respective limitations (Supporting Information S1: Appendix 3). Issues such as over simplification and unclear distinctions between categories persist in previous typologies such as “humanlike” or “non‐humanlike,” since virtual influencers might look human, but they might not behave or communicate in human‐like ways, or vice versa. Additionally, content strategies and diverse forms beyond the humanlike perspective have not been incorporated into previous categorizations, despite influ ential storylines conveyed by anime‐like and nonhumanlike virtual influencers (Yan et al., 2024). To address the limitations of existing taxonomies, we assimilate previous research alongside an empirical examination of virtual influencers' Instagram accounts to present a revised typology based on four dimensions: form‐relatedness, content‐ relatedness, agency, and function. These dimensions are explicated through our analysis of existing studies (Supporting Information S1: Appendix 3 and Table 1) and our Instagram‐derived typology of ex emplars of each type of virtual influencer (Figure 1 and Supporting Information S1: Appendix 4). These dimensions can also be under stood in terms of degrees of anthropomorphism, as form‐relatedness TABLE 1 Typology definitions and examples of analyzed articles. Author(s) Typology and examples of virtual influencers Summary of definitions* Low form‐relatedness, low content‐relatedness (Mascot) Branded nonhumanlike virtual influencer Branded nonhumanlike virtual influencers possess limited humanlike qualities, and their content typically features less life story which tends to be flat. This type of virtual influencer functions similarly to a brand mascot. Xie‐Carson, Benckendorff, et al. (2023); Xie‐ Carson, Magor, et al. (2023) For example, Geico, Janky, and Guggimon Zoomorphic or nonhuman virtual influencers, appearing in the forms of various living or nonliving artifacts such as animals, plants, objects, and toys, can serve as mascots or licensed characters. Medium form‐relatedness, Shallow content‐relatedness (Evangelist) Branded anime‐like virtual influencer and Free‐agent anime‐like virtual influencer Anime‐like virtual influencers, whether owned by a brand (branded) or creators (free‐agent), have an animated or cartoonish appearance, either in two dimensions (2D) or three dimensions (3D). Their content is likely to show a superficial, undeveloped personal life story. Even a free‐ agent anime‐like virtual influencer tends to be positioned as a promoter of products, services, or social issues. Baudier et al. (2023); Lou et al. (2023) For example, Casas Bahia, Qai Qai, Noonoouri Doll‐like or anime‐like virtual influencer—a nonhumanoid character appearing in 2D or 3D, can yield information to target customers and enhance the management of customer relationships. Mouritzen et al. (2023) An unrealistic humanoid in a cartoon form sharing content specialized in specific areas like music. High form‐relatedness, Low content‐relatedness (Spokesperson) Branded humanlike virtual influencer Branded humanlike virtual influencers are used as spokespersons or ambassadors for a brand. Designed to promote the brand owning them, branded humanlike virtual influencers' content often involves products/services of the brand and less focus on their personal life‐like storytelling. Franke et al. (2023); Koles et al. (2024) For example, Lu do Magalu, Candy Prada, Colonel Sanders Humanoid virtual influencers created by brands serve as spokespersons. These humanlike virtual influencers advocate for or promote the brand or company to which they belong as opposed to a free‐agent humanlike virtual influencer. Low form‐relatedness, High content‐relatedness (Storyteller) Free‐agent nonhumanlike virtual influencer Even though free‐agent virtual influencers are low human‐ likeness in forms, their content seems accessible. To draw attention and engagement, free‐agent nonhumanlike virtual influencers tend to share relatable content, such as humorous posts related to daily human circumstances and personal experiences. Koles et al. (2024); Mouritzen et al. (2023) For example, Nobody Sausage, John Pork, Bee influencer Nonhumanlike virtual influencers' appearances are limitless, ranging from 2D to 3D forms in various shapes or objects. Their content tends to be theme‐oriented (e.g., hedonic, entertaining, social issues), often mimicking lifelike lifestyles or occasionally presenting them as part of the real world. High form‐relatedness, High content‐relatedness (Influencer) Free‐agent Humanlike virtual influencer Equipped with hyper‐realistic attributes and operated by creators, free‐agent humanlike virtual influencers mirror human influencers by sharing personal and multidimensional narrative stories. Their evolving storylines are crafted for fostering engagement and relationship with their audiences to shape audiences' attitude/behaviors. (Continues) relates to the extent to which the virtual influencer looks human, and content‐relatedness relates to the human‐like nature of the content, for example with personal revelations about the virtual influencer's daily “life” and personal routines or images with human “friends.” Agency refers to the extent to which the virtual influencer is scripted to appear to have human‐like autonomy either as a free‐ agent virtual influencer (brand‐independent or typically owned by its developer) or a branded virtual influencer (brand‐owned). Function refers to the strategic purpose(s) which the virtual influencer appears to be fulfilling or in‐character role of the virtual influencer as a brand ambassador, as a presenter, or as a content‐creator. Our secondary review of Instagram virtual influencers (see Figure 1) synthesized with our literature review shows how a selection of virtual influencer types would fit into a proposed framework based on these two dimensions: form‐relatedness and content‐relatedness. Within these dimensions, virtual influencers are further differentiated by sub‐categories of branded or free‐ agent virtual influencers and flat, shallow, or multi‐dimensional content. In addition, we note the putative function or in‐ character role of the virtual influencer as a presenter, content‐ creator, or ambassador. In Table 1, we elaborate on Figure 1 with greater detail and source articles. Accordingly, we argue that a virtual influencer has many nuances depending on form, function, anthropomorphic positioning, product/market context, and con tent delivery substance and style. In particular, the anthropo morphism dimension is many‐layered, since a virtual influencer can imitate a human influencer in visual likeness, or in other ways that are scripted into their storylines. TABLE 1 (Continued) Author(s) Typology and examples of virtual influencers Summary of definitions* Ameen, Cheah, et al. (2023); Miao et al. (2022); Xie‐Carson, Benckendorff, et al. (2023); Zhang et al. (2023) For example, Lil Miquela, Imma A computer‐generated humanoid character, whose fictional persona and ultra‐realistic visuals are crafted by creators using technologies such as artificial intelligence and deepfaking, shares personal experiences, emulates human emotions, and exerts influential power over its followers to support branded campaigns. Note: Authors' summary based on the analysis of related studies and virtual influencer Instagram accounts. FIGURE 1 Typology of virtual influencers. ### 2.2. | Form‐relatedness A virtual influencer with high form‐relatedness is the most humanlike visually (Ferrari & McKelvey, 2023) and their potential for appearing lifelike will increase as the technology develops. We label these as “hu manlike virtual influencers.” Humanlike virtual influencers tend to rely on computer‐generated imagery combined with other technologies, such as machine learning and deepfaking (Campbell et al., 2022) to enhance their realism (Zhang et al., 2023). Their anthropomorphic look heightens the perception of humanness, which elicits positive responses from con sumers (Miao et al., 2022). These positive responses could manifest in enhanced rates of technology evaluation, stronger perceptions of social presence, stronger feelings of connection (Ahn et al., 2022), and being persuaded to buy (Kim et al., 2024). However, consumers' response to a high level of anthropomorphic realism is not always positive. Some users find interacting with humanlike virtual influencers unnerving or “creepy” (Meng et al., 2024), for example, when there is a perceived mismatch between the human‐like appearance of virtual influencers and the detail of their behavior (Ham et al., 2023). For brands, the solution could involve either improving the realism of the behavior, or deliberately rendering the humanlike virtual influencer less realistic to emphasize its nonhuman nature (Moustakas et al., 2020). Virtual influencers with medium form‐relatedness typically appear less realistic, resembling caricatures, animations, or cartoons (Franke et al., 2023) in either two‐dimensional or three‐dimensional forms. Pre vious research has defined these virtual influencers as “anime‐like virtual influencers” (Lou et al., 2023). Anime‐like virtual influencers originated in the 1990s for use in traditional advertising media as an extension of a brand's established cartoon character (Yan et al., 2024). They have the virtue of being less uncanny for some users because they are overtly not humanlike (Koles et al., 2024). However, their lower form‐relatedness can be a disadvantage, except in some cultural backgrounds (Xie‐Carson, Magor, et al., 2023), because it can make them less relatable (Wan et al., 2024) and less likeable (Ahn et al., 2022), factors that often influ ence an influencer's ability to persuade. Nonhumanlike virtual influencers lack form‐relatedness and realism, and they can be drawn in two dimensions or three dimensions (Mouritzen et al., 2023). Some of these influencers have the body or face of an animal or an item of food (Xie‐Carson, Benckendorff, et al., 2023), in common with many traditional brand mascots (Brown & Ponsonby‐ McCabe, 2013). The findings of previous studies on nonhumanlike virtual influencers are conflicted, since some studies suggest that users respond negatively to entities with low levels of relatability or humanness (Muniz et al., 2023) and prefer highly humanlike attributes (de Boissieu & Baudier, 2023), but some nonhumanlike virtual influencers have gener ated remarkably high engagement rates (e.g., Nobody Sausage) and emotional attachment (Yan et al., 2024). Due to the limitations of current technology, virtual influencers cannot be fully automated. A team of designers and programmers is required for their oversight and manipulation (Koles et al., 2024). Consequently, virtual influencers lack agency; they either represent a brand (branded virtual influencers) or are owned by their developers (free‐agent virtual influencers). However, their agency can be mi micked through astute narrative scripting, and this gives them dif fering degrees of “content‐relatedness.” The content can be flat, denoting a lack of nuanced storytelling, or it can be richer and more human‐like (Quach et al., 2024) eliciting multidimensional or intricate lifelike narrations to facilitate virtual influencer roles or functions such as content creator. Conversely, shallow content‐relatedness suggests simplicity and limited depth. Underpinned by agency, content‐relatedness, and function, the three virtual influencer types (humanlike, anime‐like, and non humanlike) can be segmented into five categories: spokesperson humanlike virtual influencers, influencer humanlike virtual influen cers, evangelist anime‐like virtual influencers, mascot nonhumanlike virtual influencers, and storyteller nonhumanlike virtual influencers. The spokesperson humanlike category (high form‐relatedness, flat content‐relatedness) functions as an ambassador or spokesperson for the brand that owns them as they typically feature a bland persona and share limited personal stories. Examples include Lu do Magalu, the Brazilian retailer Magalu's branded humanlike virtual influencer, and Candy, created to promote Prada's fragrance products (Sands, Ferraro, et al., 2022). Some spokesperson humanlike virtual influen cers, such as Lu do Magalu, have a highly humanlike form. However, she achieves an average engagement rate of only 0.2%, which sug gests there may be issues with content strategy in the spokesperson humanlike segment. These virtual influencers present a human‐like appearance, yet their behavior diverges from typical human conduct. Conversely, the influencer humanlike category (high form‐ relatedness, multidimensional content‐relatedness) are free agents with crafted personas that can be limitlessly developed to emulate the personal interests and personalities of human influencers. These humanlike virtual influencers leverage content related to emotions or lifestyles to engage audiences and foster relationships. Examples include Lil Miquela, a virtual influencer that has been recognized as one of Time magazine's “top 25 influential people on the Internet,” (Ozdemir et al., 2023) and Imma, a virtual influencer that was crafted by a Japanese media company and has worked with brands including Ikea and Unilever's Magnum. With their multidimensional content and their realistic appearance, Miquela and Imma generate average engagement rates of 2% and 3%, respectively. This is moderately higher than the 1%–1.3% median benchmark. Irrespective of their agency, evangelist anime‐like virtual influen cers (medium form‐relatedness and shallow content‐relatedness) function as promoters, with semi‐developed storylines and perso nas that lack depth. For instance, the branded anime‐like virtual in fluencer Casas Bahia shares content featuring products and promo tions, tending to elicit emotions only when they are related to the item being promoted. Another example is Noonoouri, a free‐agent anime‐like virtual influencer that claims to be vegan and a fashion lover (Franke et al., 2023). Noonoouri's content tends to promote sponsored products or social issues, without any underlying storylines to build upon. Casas Bahia's engagement rate (0.69%) is comparatively low, while Noonoouri's (1.8%) is slightly above average benchmark. However, anime‐like virtual influencers can generate better metrics when they are given a background storyline, such as a “character biography” (Koles et al., 2024) and more creative narra tives. An example is Qai Qai, a doll anime‐like virtual influencer belonging to Serena Williams' daughter. Williams' fame as a tennis player perhaps transfers to Qai Qai, with the underlying story being that Williams gave this black baby doll to her daughter as her first doll. This, along with Qai Qai's engaging content, has resulted in an engagement rate of 4.25%. This evidence challenges the notion that a virtual influencer's humanlike characteristics are essential for audi ence engagement (Yan et al., 2024). It suggests that even virtual influencers with medium to low form‐relatedness can be engaging when combined with creative content and a well‐developed persona. The mascot nonhumanlike virtual influencers (low form‐ relatedness, flat content‐relatedness) are owned by specific brands. Mascots are designed to embody brand values and humanize the brand, thereby enhancing social engagement between brands and their audiences (Radomskaya & Pearce, 2021). However, mascot nonhumanlike virtual influencers lack complexity in their content and personas, and they function more as brand characters than as true virtual influencers. The absence of ongoing, appealing storylines results in modest engagement rates, with averages of 0.67% for the Geico Gecko and 0.59% for Ryan from the South Korean app Kakao. Nevertheless, when integrated with technologies that facilitate im mersive engagement and social presence, such as the metaverse (Hennig‐Thurau et al., 2023), mascot nonhumanlike and other types of virtual influencers, may be able to enhance brand engagement and positive emotions through innovative and interactive mediums (Dwivedi et al., 2023). Lastly, the storyteller nonhumanlike category (low form‐ relatedness, multidimensional content‐relatedness) refers to brand‐ independent nonhumanlike virtual influencers that perform similarly to those of content creators. They attract audiences for brand part nerships. Despite their low level of realism, storyteller nonhumanlike virtual influencers can still achieve impressive engagement rates. Take Nobody Sausage, a nonhumanlike virtual influencer resembling a sausage, with over 7.5 million Instagram followers, achieving a 67% increase from 2022 to 2023 and a 4.34% engagement rate. This significantly surpasses the 1.2% average for human influencers with over 1 million followers (Statista, 2022). Another example is John Pork, a character with a pig's face on a humanlike body. He has more than 320,000 Instagram followers and an engagement rate of 9.42%. Commonalities in the content shared by these examples of non humanlike virtual influencers include themes of diverse humor and relatable lifestyle content, often becoming memes. They occasionally share amusing content about fictional members of their family, emulating everyday human experiences. When storyteller non humanlike virtual influencers post content that evokes joy and hap piness, it appears to foster high engagement (Lim & Lee, 2023). The exceptional performance marketing metrics for nonhumanlike virtual influencers may explain the discord in the research on both humanlike and nonhumanlike virtual influencers' electronic word‐of‐ mouth effectiveness and likeability determined by form‐relatedness (Mouritzen et al., 2023; Sands, Campbell, et al., 2022). Our analysis suggests that audience engagement with virtual influencers is facili tated by relevant content, interactivity, and positive sentiment, rather than by form similarity (Yan et al., 2024). Accordingly, nonhumanlike virtual influencers offer fertile ground for future research, especially regarding the factors that influence user engagement with characters who are seen as less relatable, such as storyteller nonhumanlike virtual influencers. Virtual influencers are often designed with anthropomorphic qualities to perform human influencer tasks, but there are striking differences in these qualities in type and degree (Lv et al., 2023). Thus, as a further point of analysis, in the next section, we compare features of human influencers and virtual influencers. 3 | KEY DIFFERENCES BETWEEN HUMAN The distinctions between human and virtual influencers primarily reside in (1) their nature and representation, (2) type of engagement with audiences, and (3) forms of content and narrative delivery. We elaborate on these three categories (See Supporting Information S1: Appendix 5). First, human influencers bring their personal life tra jectories (real or invented) to their online personae, which creates a basis for projective identification for their audience target group and determines their distinctions in those three areas. In contrast, virtual influencers' technological mechanisms (visual rendering, pre‐ programmed responses, and designed mixed‐reality content) and creators (expertize, interests, creative content direction) influence these distinctive aspects. Virtual influencers' appearance can be assessed for attractiveness and homophily similar to humans. How ever, technological factors and the role of creators complicate direct applications of human‐centric theories, such as source credibility, to examine virtual influencers' credibility, engagement, and endorse ment efficacy. This complexity arises because virtual influencers inherently lack the personal life experience that would lend authority to their expressed views (Meng et al., 2024). A lack of life experience also renders problematic the applica bility of the brand match‐up hypothesis to brands' relationships with virtual influencers. With human influencers, the efficacy of the match‐up can be influenced by the relationship history with the user (Kim & Kim, 2021), but with virtual influencers the effects are ambiguous across a range of contexts and industries. Additionally, there might be ambiguity in the source of virtual influencers' efficacy and impact attribution (Zhao et al., 2024). Moreover, the explanations of audience‐virtual influencer pseudo‐ or para‐relationships may be distorted due to their creators' influence, preprogrammed interactions, and deficient tailored interactions (Yan et al., 2024). Hence, there are various contextual issues that potentially distort the quasi‐relationship or basis for engagement between user and virtual influencer. There are theorizations that might obviate the potential contra dictions in audience responses to human and virtual influencers. For example, the interaction can be theorized using computers‐as‐social‐ actors theory (Yu et al., 2024), in which users attribute humanlike qualities to inanimate computer‐generated entities to facilitate or mimic social interaction. Perhaps there is a theory of mind in play, whereby the user assumes the consciousness and autonomous cog nitive processes of the nonhuman influencer (Zhang et al., 2023) as a precondition for interaction. However, the presuppositions for the applicability of these theories are unknown. It is unclear whether users consciously suspend reality to engage in a make‐believe rela tionship or interaction with a virtual influencer, or if the interaction is spontaneous with unconscious psychological antecedents. Accord ingly, the potential theoretical application challenges necessitate new indicators or approaches and theoretical integrations from fields like human‐computer interaction or information systems for a more comprehensive elucidation of the impacts and efficacy of virtual in fluencers and the key drivers of these phenomena. The ways in which virtual influencers differ can be subtle, or profound, and all differences have implications for research and practice. Therefore, we further conceptualize and identify nuanced factors influencing the efficacy of human and virtual influencers while constructing overlapping research themes on these influencers in the refined framework. Our research‐based analysis (see boundary of the review and method ology in Supporting Information S1: Appendix 6) explicates factors that similarly or differently limit or enhance the influence, consumer engage ment, and persuasion efficacy of human and virtual influencers. Adapting the framework from the persuasive communication concept (Hovland et al., 1953), we conceptualize those factors encompassing five themes: (1) communicator, (2) communication, (3) audience, (4) responses, and (5) underlying mechanisms. This framework elucidates these dynamics and identifies intersecting research themes (Figure 3). #### 3.2.1. | Communicator Human influencers' credibility is formed through perceived trustworthi ness, expertize, and social identity by sharing personal interests that resonate with their audience (Ye et al., 2021). Leveraging anthropo morphic realism to appear expert and trustworthy (Igarashi et al., 2024), virtual influencers construct their influences through (1) nuanced forms, (2) identity positioning, (3) operational mechanisms, and (4) technological advancements. For nuanced forms, high degrees of realism and homo phily in virtual influencers enhance persuasiveness and credibility (Wan et al., 2024) but can also cause discomfort if overly or unsettlingly hu manlike due to cognitive dissonance. Notably, the high engagement rates and large numbers of followers observed in some virtual influencers with low form‐relatedness (see typology of virtual influencers) suggest that beyond the degree of humanness, unknown factors may contribute to persuasion. For example, recent research found that anime‐ and non humanlike virtual influencers elicit higher emotional attachment and authenticity benefit‐seeking than humanlike virtual influencers (Yan et al., 2024). Moreover, virtual influencers are increasingly being portrayed in nuanced forms, transcending simplistic human or nonhuman ap pearances to include a variety of ethnicities, ages, genders, dis abilities, and sexualities to appeal to minorities (Figure 2): Kim Zulu is a black African virtual influencers or Mosi from Hum.ai.n is a char acter with vitiligo. However, virtual influencer research is still con strained to making simplistic comparisons between human influen cers and humanlike virtual influencers, or between anime‐like and nonhumanlike virtual influencers (Kim et al., 2024; Zhao et al., 2024). Parallel to humans varying in forms, considering particular char acteristics (such as ethnicity and LGBTQ+ status) is important for evaluating user responses related to parasocial interaction, brand attitude, and persuasion (Li, 2022). Regarding identity positioning and operational agencies, human authenticity is perceived through genuine interest in specific topics and autonomy (Leung et al., 2022). Conversely, virtual influencers are mana ged by creators, which can limit their perceived authenticity, personalized responses, and spontaneity. Virtual influencers have restricted sensory capabilities (Li et al., 2023), which reduces their efficacy (Lou et al., 2023) when endorsing products that require a sensory experience. However, exposure to a well‐constructed identity and narrative that fits virtual in fluencers' appearance may evoke their authenticity (Lee & Yuan, 2023). Technological advancement, like a more autonomous or artificial intelli gent enhancement, may enable virtual influencers to cultivate pseudo‐ relationships by posting consistently and responsively engaging with followers (Yan et al., 2024). #### 3.2.2. | Communication Human influencers leverage self‐disclosure, rhetoric, and emotional con tent to build credibility and social presence (Zhou et al., 2021). Emulating humans, virtual influencers employ rational, warm (Gerrath et al., 2024), and flattering messages (Quach et al., 2024) to increase engagement, credibility, prosocial behaviors, and trust (Ameen, Cheah, et al., 2023). These message strategies can enhance virtual influencers' approachability and social presence (Yan et al., 2024) and mitigate perceptions of inauthenticity and lack of humanness (Meng et al., 2024). #### 3.2.3. | Audiences Both human and virtual influencers can enhance persuasion by aligning their self‐concept with target audiences (Wan et al., 2024). Arguably, there are some groups that are more receptive to virtual influencers and their endorsement, including younger consumers, especially younger women, and some cul tures and audiences with certain motivations like promotion‐ focused and personality types (Ham et al., 2023; Muniz et al., 2023). For example, highly empathetic consumers are more willing to follow and engage empathetically with a virtual influ encer as a comforting substitute for the complexities of human interaction (Mirowska & Arsenyan, 2023). Audiences' engage ment is motivated by the virtual influencer's novelty, aesthetic appeal, advanced technological features (Lou et al., 2023), and engaging storytelling (Zelenskaya & Rundle‐Thiele, 2022). #### 3.2.4. | Responses Consumer reactions to virtual influencers are conflictive. Some studies found negative effects of virtual influencer's appearances and unrelatability (Lou et al., 2023; Mouritzen et al., 2023), while others reported no impact in certain contexts (Kim & Park, 2023; Yan et al., 2024). This implies that although the anthropomorphic realism and authenticity of a virtual influencer matter to some consumers in some contexts, it does not matter in all circumstances. However, both human and virtual influencers are scrutinized over ethical issues such as misinformation and unattainable standards (Sands, Ferraro, et al., 2022; Zhao et al., 2024). #### 3.2.5. | Underlying mechanisms Human influencer literature highlighted mechanisms underlying en dorsement efficacy in addition to the four elements indicated in the persuasive communication concept (Chen et al., 2023): (1) the relationship between sources and audiences, and (2) the congruence between sources and their recommendations. Parasocial relationships and attachments stimulate influencers' persuasion (Aw & Chuah, 2021). Virtual influencers, particularly nonhumanlike, can leverage advanced artificial intelligence (Meng et al., 2024) to induce emotional attachment and respond inter actively, fostering connections similar to human influencers (Yan et al., 2024). Concerning endorser‐brand fit, incongruence between the endorser and the brand can damage the human influencer's credibility and worsen consumer attitudes toward the endorsed product (Kim & Kim, 2021). Virtual influencers are more effective for brands that involve innovations or do not require sensory engagement (Franke et al., 2023), and their faulty endorsement elicits lower negative responses than human counterparts due to virtual influencers' low mind perception (i.e. lack of conscionsness and autonomous cognitive processes) (Zhao et al., 2024). From the preceding synthesis and conceptualization of research literature, trade literature, and social media platforms, we induce a number of recommendations or propositions for future research and practice, which we present in Table 2 below. We organize these research directions as we previously framed drivers of human and virtual influencers' efficacy and intersected research themes in Figure 3. FIGURE 2 Examples of virtual influencers representing inclusivity and diversity. TABLE 2 Potential future research directions for virtual influencer marketing. Theme 1 Sub‐themes Potential areas for future research The Communicator (characteristics/ appearance cues) Nuanced forms • How can new tenets or constructs beyond the realm of source credibility theory be identified, and how can longitudinal research help elucidate the factors or distinctive qualities that underpin virtual influencers’ role as persuasive communicators? • Which forms of virtual influencers are preferable for performing endorsement tasks, and in which context should they be utilized? • How do virtual influencers with diverse identities on audiences' attitudes, particularly when conveying endorsements or messages related to gender equality, ageism, or inclusivity? • How does perceived similarity or homophily affect engagement among specific target audiences who share resemblances with virtual influencers' characteristics and values, such as stances on veganism or sustainability? Positioning of virtual influencers • What agencies or factors can compensate virtual influencers' lack of physical presence and sensory abilities to establish credibility and persuasiveness? Operational mechanisms: creators and artificial features • What disclosure approaches should creators employ across virtual spaces to prevent any adverse effects on all parties including virtual influencers, creators, partnered brands, and audiences? • How can policymakers address sensitive designs related to child or minor virtual influencers? • How can the exploitation virtual influencers narration by unethical or immoral creators be prevented? • How do brands partnered with virtual influencers address unethical manipulations instigated by the creators? • What are recovery strategies in situations where creators diffuse misinformation through virtual influencers or do not deliver desirable objectives? • How do consumers perceive virtual influencers performing tasks such as live‐streaming commerce, live commerce or supporting human tasks such as loneliness economy (e.g., always‐on friend) or customer service touchpoints in the Metaverse? Future technology integrations • How do consumers react to hyper‐realistic humanlike virtual influencers integrated with generative artificial intelligence? Does this integration heighten or lower perceptions of the uncanny valley? • How do consumers perceive the appropriateness of virtual influencers as technological entities for fulfilling marketing tasks, including product endorsement? • What potential roles do consumers see for virtual influencer in the metaverse beyond product endorsement? • Develop a new empirical theory or employ theories from interdisciplinary to elucidate the role of virtual influencers as a technology‐enhanced agent to perform human tasks. • What impact does the integration of virtual influencers with generative artificial intelligence have on marketing strategies, including both economic and noneconomic value creation? Theme 2 The Communication (content and messages styles) • How do different content formats (image vs video) affect consumers' perception and virtual influencers' endorsement outcomes? (Continues) In the next section, we use the potential areas for future research that we have posed from our synthesis of research literature and secondary research, to inform a purposive qualitative study. In this way, we apply our proposals for future research deriving from sec ondary sources to an informal process of tentative verification by exposing them to expert practitioners, and we duly elaborate on and/ or adjust our ideas accordingly. 4 | QUALITATIVE METHOD Semi‐structured, in‐depth interviews were conducted from February to March 2024, generating more than 70,000 words of transcripts. We employed a purposive sampling technique to recruit industry senior experts based on two main criteria: (1) being a business owner operating in the realms of virtual influencers and the metaverse, and TABLE 2 (Continued) Theme 1 Sub‐themes Potential areas for future research • How do audiences react to virtual influencers' messages such as self‐disclosure, humblebragging, the use of cohesive or inclusive language styles, assertive language in endorsement, positive and negative messages, message sidedness, and de‐ influencing? And, how do these message affect their partnered brands? • How do audiences react to virtual influencers promoting communication surrounding social causes and justice issues such as cybersecurity, environmental sustainability or responsible tourism? • What levels and types of rhetoric facilitating humanization, increase persuasion ability and influence consumers to follow and continue to engage with virtual influencers, rather than inducing a sense of uncanniness? Theme 3 The Audience (consumers' backgrounds, beliefs, and cognitive abilities) • What and why are certain attributes of audience profiles and contexts more likely to be receptive to virtual influencers? • How do audiences' motivations and psychographic factors For example, cultural differences or personalities impacts engagement with virtual influencers and their persuasion? Theme 4 The Response (consumer reactions) • How nuanced types of virtual influencers influence consumers' responses and decisions such as to follow or not follow, engage with, and trust virtual influencers? And how virtual influencer types impact their actual behaviors? • How do audiences respond to the use of virtual influencers as positive role models or motivators in promoting healthy or constructive responses? Theme 5 Underlying mechanisms (additional drivers of the persuasive communication process) Relationship between sources and audiences • How virtual influencers' interactivity levels and content sharing frequency affect relationships between virtual influencers and audiences, particularly when they are enhanced by generative artificial intelligence? • How do parasocial interaction and parasocial relationships of virtual influencers and audiences change within the context of mixed reality and metaverse? The congruence between sources and their recommendations • Can the meaning transfer and match‐up tenets be applied in virtual influencer marketing and how do these concepts affect the brand image? • Which contexts, such as well‐being, healthcare, tourism, and social causes, virtual influencers can be most effective in? • How do different types of virtual influencers and their roles in creating aspirational and exclusive perceptions for luxury brands and high value products (e.g., real estate or stock/ bitcoin trading) influence consumer reactions and brand perception, and will these outcomes change in the immersive or virtual worlds? (2) having at least 3 years of experience and demonstrable knowledge of how virtual influencers work. To ensure a diverse and represent ative sample, we sought experts from various sectors including marketing agencies, technology firms, content creation studios, and brand management consultancies. Potential participants were iden tified through professional networks, industry conferences, and rec ommendations from initial interviewees (snowball sampling). This approach allowed us to gather insights from a range of perspectives within the virtual influencer ecosystem. The final sample of 10 ex perts represented a mix of backgrounds including creative directors, technology entrepreneurs, digital marketing strategists, and virtual influencer creators. In addition, the experts were based in Asian, Middle Eastern, and Western countries. These were obtained via personal contacts and snowballing. We provide details of the inter viewees (anonymized) in Supporting Information S1: Appendix 7. The purpose of the interviews was to give our suggestions for future research and practice inferred from the synthesis of research litera ture and trade and secondary sources greater traction by running them past experienced professionals for refinement and/or adjust ment. Our aim was to distill our suggestions for future research and practice into a series of concrete findings to guide future research‐ backed practice. The interviews were conducted online in English and lasted from 30 min to 1 h each. They were auto‐transcribed and then manually rechecked. Our semi‐structured interviews were guided by a comprehensive protocol designed to explore key aspects of virtual influencers and user engagement. The interview guide was detailed, drawing on the potential areas for future research that we identified from our syn thesis of research studies (see Supporting Information S1: Appendix 6 for the methodology) and secondary trade sources (Supporting Information S1: Appendix 1). The interview guide covered several main areas: participants' background and experience with virtual in fluencers, perceptions of virtual influencer appearance and typology, the potential for genuine connections between virtual influencers and audiences, industry applications, and the role of virtual influen cers in the metaverse. Examples of the key questions included: “Can you share feedback on your project involving virtual influencers?” “What could lead to their success or failure?” “Does a virtual influ encer's appearance have any implication on their influence?” and “How do you envision the future of virtual influencers evolving within the metaverse and alongside advancements in generative AI?” We also explored topics such as the impact of virtual influencers on cultural norms, societal values, and mental health. The interview guide was structured to allow for open‐ended responses while ensuring coverage of topics relevant to our research questions and conceptual framework. This approach provided rich, detailed data on user engagement with virtual influencers, allowing us to explore the nuances of anthropomorphism and its effects on user responses. We conducted a thematic analysis of the transcripts ac cording to Braun and Clarke (2006) and identified 10 superordinate themes and multiple subordinate themes. The first author conducted the initial coding and thematic analysis of the interview transcripts, and the team of authors then oversaw the data analysis to enable inter‐subjective verification of themes. The data generated by the FIGURE 3 Shared and divergent themes on human and virtual influencers. interviews was deep and broad, and we distilled the findings into those most relevant to the earlier themes we had identified from secondary sources. Our study employed a rigorous qualitative approach to analyze the interview data. The interviews were initially auto‐transcribed and then manually rechecked by the research team to ensure accuracy, providing verbatim transcripts for analysis. We utilized a combination of open and axial coding techniques (Braun & Clarke, 2006; Strauss & Corbin, 1998) and six‐stage thematic analysis to systematically examine the data. The analysis process began with all team members thoroughly familiarizing themselves with the transcripts. We then applied open coding to the data (Miles & Huberman, 1994), generating initial codes based on our interpretations and interview notes. For instance, when analyzing responses related to anthropomorphism, codes such as “human‐like features,” “emotional responses,” and “uncanny valley” emerged. These initial codes were then connected through axial coding to form broader themes, such as “Anthropomorphic Engagement.” Then, the team collectively identified and reviewed emerging themes. Two researchers independently reviewed and triangulated the themes to ensure they were well‐grounded and reliable. We then refined and clearly defined each theme, ensuring they accurately represented the data. Throughout this process, we employed dialectical tracking to compare our emerging patterns with existing literature (Belk et al., 2013), contextualizing our findings within established knowledge. To enhance the credibility and trustworthiness of our analysis, we implemented several strategies. We used investigator triangulation, with multiple re searchers coding independently before comparing results to mitigate individual bias. Member checking was employed, where a sample of participants reviewed our interpretations for accuracy. We maintained a detailed audit trail of our coding decisions and analytical process, which was regularly reviewed by the research team. Additionally, we engaged in peer debriefing through regular team discussions to challenge as sumptions and explore alternative interpretations. To facilitate our coding and theme development process, we utilized NVivo qualitative data analysis software. This tool enhanced our ability to manage and analyze the large volume of data efficiently, contributing to the robustness of our analysis. This comprehensive approach to data analysis ensured a systematic, thorough, and trustworthy examination of our interview data, allowing us to derive robust insights into user en gagement with virtual influencers. By meticulously following these steps and employing various validation strategies, we were able to generate findings that accurately reflect the complexities of user responses to anthropomorphism in virtual influencers. 5 | FINDINGS: PROPOSITIONS In this section, we take the issues summarized in Table 2 that resonated with our interviewees, and we refine, adjust or change our propositions for future research and practice accordingly. We structure our findings using categories from Leung et al. (2022), namely: (1) targeting benefits, (2) positioning benefits, (3) creativity benefits, (4) trust benefits, (5) content control threat, and (6) customer retention threat. ### 5.1. | Targeting benefits The existing literature on human influencers highlights the benefit of leveraging influencers to connect with and acquire new consumer segments from an influencer's fanbase (Zhou et al., 2021). Virtual influencers potentially offer the same benefit if and when they are well‐established, especially regarding young and more diverse audi ences who seek novelty, uniqueness, and emotional escape, as pointed out by some of the interviewees: “Consumers will always engage with virtual influen cers, and I think the numbers will continue to grow because it's so intriguing, and it's so exciting.” (A3) Some consumers, especially those with high cognitive empathy but low emotional disassociation (Mirowska & Arsenyan, 2023), may make little distinction between virtual influencers and human influencers and are happy to interact with virtual influencers as if they were human (Lee & Yuan, 2023). As virtual influencers are infinitely customizable in terms of appearance, voice, and language, they have strong potential for tar geting new consumer segments (Kietzmann et al., 2020). This customi zation allows brands to access international audiences and engage with minority audiences. One of the senior experts stated: “You can have models, multiple models, all shapes and sizes. So you can … Market to a more diverse group of people.” (A4) In the interviews, we asked interviewees how they saw virtual in fluencers fitting into metaverse development. This technological context could help the brand to appear to be part of a technologically innovative future (Franke et al., 2023) and this could appeal to younger consumers (Ameen, Hosany, et al., 2023), as stated by some of the senior experts: “The children being raised today will be the ones who truly experience what the real metaverse is going to be, and to understand the convergence between the virtual and physical worlds. For that generation, it is going to be as regular as having the Internet today or having your mobile device.” (A9) An example in practice is a recent Heinz tomato ketchup advertising campaign, which incorporated an artificial intelligent image generator and exhibited artworks in a metaverse art gallery (Ads of the world, 2023). This campaign revitalized the brand, made it relevant to younger audi ences and (potentially) art lovers, started social conversations, raised engagement rates, and increased sales revenue. With future enhance ment, virtual influencers may be able to generate economic value in digital spaces through live commerce with their audiences (Appel et al., 2020) or support the loneliness economy (Marriott & Pitardi, 2024) and customers within this segment. Assimilating our secondary and research‐paper sources with our interviews we distilled our findings into the first prop osition for future practice thus: P1 Leveraging virtual influencers could help brands to (1) expand their segments by reaching younger and broader audiences across virtual spaces and contexts, and (2) be part of the new wave of advanced technologies. ### 5.2. | Positioning benefits The infinite adaptability of virtual influencers means they can be used in nuanced and differentiated positioning strategies to support diverse brands and their targets. For example, luxury brands can leverage virtual influencers with ideal representation or positioning that res onates with target audiences to elicit perceptions of unique authenticity (Koles et al., 2024) and exclusivity. A virtual influencer's artificial characteristics evoke novelty and futuristic connotations that can be transferred to partnered brands (Zhang et al., 2023). Fur thermore, their customizable positioning can address consumers' “influencer fatigue” by incorporating unexpected details into their storylines. For example, the virtual influencer Imma cloned herself and transformed her shape and appearance to address inclusivity and diversity issues, as indicated by one of the business owners: “I think that was quite an interesting way of incorpo rating it, including people that have Down syndrome. I think they can actually be a really positive way of telling a story incorporating a certain group of people and having a conversation about that group of people. So that everybody kind of understands and potentially can be a lot more compassionate.” (A3) Another virtual influencer, Noonoouri, abruptly transitioned from supermodel to musician to launch a music career. Such radical transformations would be impossible or inauthentic in a human in fluencer, but virtual influencer audiences embrace them without missing a beat. We therefore arrive at our proposition two thus: P2 Leveraging virtual influencers could enable brands to refresh and appeal to diverse target audiences by (1) employing nuanced and differentiated positioning strate gies, and (2) evoking a sense of brand novelty, futurism, or coolness that transfers from the virtual influencer by changing the visual appearance and backstory of the virtual influencer. ### 5.3. | Creativity benefits Virtual influencers present advertising copywriters, scriptwriters, and brand storytellers with a rich palette for developing creative content, including story‐worlds, backstories, and relationships. Instead of cli chéd promotional content, creators of virtual influencers can offer distinctive and boundless content creation as one of the experts stated: “People that manage virtual influencers have to firstly be creative. There's a lot more to it than just posting a pic ture. Every single day you have to push the boundaries because people want to see new things like show us a new trick. It's like going to a magician show.” (A3) The technology allows mixed‐reality presentations (Ham et al., 2023) that can enhance a virtual influencer's relatability and persuasiveness without the constraints of reality, thus making virtual influencers a highly creative and novel medium that can attract au diences who seek novelty and experiences made possible by advanced technology as some of the experts stated: “A lot of companies are now doing work where real life is mixed with 3D objects, and it gives like a kind of wow effect. … they are raising the quality of content… virtual influencers are a very important part of the quality of content.” (A2) “The virtual reality becomes the reality, and this is kind of an interplay between the physical world and the virtual world.” (A9) Moreover, brands can use virtual influencers to engage with audiences in the metaverse while providing a more immersive and real‐time multisensory customer experience and interaction (Hennig‐ Thurau et al., 2023) as one of the business owners suggested: “By allowing people to explore these virtual worlds, which are either digital twins or imaginary worlds, you can really touch people so much deeper. And once you have engaged, when you have higher levels of en gagement, you have higher levels of memory reten tion.” (A6) Creative examples include using virtual influencers as moderators in live product showcases, recommendations, community activities, and even customer cocreation events in the metaverse. We propose: P3 Leveraging virtual influencers enables brands to produce distinct and boundless creative content and engagement, free from the limitations imposed by reality. There are insufficient and mixed research findings surrounding virtual influencers' perceived authenticity, which appears to be a predictor of trust. There might be regional and cultural as well as demographic variations in the extent to which users find virtual influencers authentic and persuasive. For example, a senior expert indicated that: “It might be China or Japan where it booms. With China's technology, live commerce, and social com merce growing, even using virtual characters for live sessions. Or Japan's anime culture makes people like created characters. It might be based there for adop tion.” (A5) This anecdotal evidence aligns with tentative findings (Franke et al., 2023) suggesting that young East Asian consumers may be particularly receptive to forming parasocial relationships with virtual influencers. One possible explanation is that these consumers find reassurance in the fact that virtual influencers, being nonhuman, cannot disappoint their fans through real‐life behavior (Mirowska & Arsenyan, 2023). As a more general point, research suggests that to build authenticity in virtual influencers requires contriving consistent, quality storylines and personas for the virtual influencers and pre senting them in mixed reality or social cues (Gutuleac et al., 2024); for example, as part of real‐life settings (Mouritzen et al., 2023). This makes virtual influencers more relatable, which fosters consumer trust (Kim et al., 2024) as stated by many of the experts: “You need to somehow get a connection with people get the trust and start building audiences.” (A4) “Human are mixed with virtual influencers. That con tent is always more interesting. It has more likes, it has more sense and replies and comments. And I think the more interaction there is between virtual and real humans, the more interesting the post becomes.” (A2) To minimize inauthenticity perceptions, creators and brands should also avoid unsettled (Kim et al., 2024) and excessively unblemished aes thetic designs for virtual influencers (Lv et al., 2023) and should circum vent incongruence between a virtual influencer's persona, messages, and its partnered brands (Quach et al., 2024). Given that the key drivers for virtual influencer authenticity and consumer trust remain ambivalent, however, further empirical research is needed. We postulate: P4 Leveraging virtual influencers still offers brands ambiguous trust benefits, arising from insufficient and conflicting findings on the drivers that stimulate con sumer trust in virtual influencers. Therefore, more focused research is needed on antecedents of trust and authenticity in virtual influencers in different contexts and with differing demographics. For brands working with human influencers, a notorious difficulty is getting the influencer to stick to the script and fulfill their contract. Human influencers build their following by being spontaneous, authentic, and creative: qualities that are stunted when brands try to enforce strict conditions on scripts, contracts, and behaviors. With their fully controllable online personas, vir tual influencers are free from such drawbacks. However, this becomes problematic when a virtual influencer is enhanced with generative artificial intelligence for more automated content production. There have already been cases of artificial intelligence‐enabled chatbots generating untrue, irrelevant, inappropriate, or misleading information in response to inquiries (Niederman & Baker, 2023): “Before creating value, I kind of think it [AI enhance ment] would create fear. Nowadays, with generative AI coming in, there are a lot of issues with intellectual property, fraud, and misuse, so people might have a barrier to believing and adopting these. The more advanced it gets, the more afraid we become.” (A5) “There will be bad actors there as well, so I think we have to be very conscious of that and designing must be very intentional about how we safeguard against those things. … Otherwise, we'll end up with Facebook toxicity across the board.” (A10) This can pose ethical and regulatory problems for brands if, say, vulnerable consumer groups are affected. These issues could harm brand reputation, compromise equity, and reduce the acceptability of virtual influencers (Thomas & Fowler, 2021). Disclosure of or trans parency around the identity of virtual influencers and their creators is needed to mitigate any unethical issues, such as misinformation and manipulation, that arise from virtual influencer content as suggested by one of the experts: “Under the upcoming EU AI act, you will have com panies need to disclose whenever you're interacting exactly with an AI. It was possible that going to future virtual influencers on Instagram, for example, on Tik Tok, YouTube, or the media will have to be clearly labeled as such.” (A4) Accordingly, virtual influencer creators need to be aware of the current regulations and the potential downside risks of enhancing virtual influencers with artificial intelligent interactional capabilities. Another downside risk concerns access to, and security of, consumer data from interactions with virtual influencers: “Virtual reality experiences are going to be tracking things like where I'm looking. The way I move, you know my vital signs and I think that there's actually a lot of data capture going on already that a lot of people don't even know about. … So it is for the benefit and enjoyment of the experience. But then what is happening to that data?” (A6) Therefore, brands should work closely with creators to ensure consumer data privacy is not compromised or unethically exploited. P5 Content created by virtual influencers that are en abled by generative artificial intelligence, coupled with the authorship power of creators, will expose brands to the following risks: (1) misinformation and manipulation, and (2) compromised privacy of consumer data. The enhancement of virtual influencers through artificial intelligence presents extraordinary potential. For example, Leung et al.'s (2022) framework showed that human influencers have limited capability in customer retention. These limitations can be addressed by enhancing virtual influencers with artificial intelligence to enable customer responses, as in a next‐level virtual influencer‐oriented chatbot function (Meng et al., 2024). For example, aided by generative artificial intelli gence, predictive analytics, and customer data, a virtual influencer can be used to detect attrition and enable proactive and responsive communi cations, as confirmed by one of our expert interviewees: “A new generation of virtual influencers backed by AI, like the ones that Meta launched last year, like, I think 24 h personas copied from celebrities, and they have their Instagram accounts and you can actually chat with them. And this is another level of interaction. They actually can communicate with them, and they can be like brand ambassadors. 24/7.” (A1) The enhancement enabled by generative artificial intelligence allows a virtual influencer to deliver data‐driven and optimized solutions auton omously across virtual spaces. Furthermore, virtual influencers with autonomous and large language model abilities may be able to perform customer service tasks within a metaverse (Dwivedi et al., 2023), offering immersive customer service experiences, as one of the experts indicated: “People are becoming more and more disconnected from your Facebook and your Instagram and your everything else. And I think that represents an opportunity for the metaverse. … route to market or that value‐added proposition. And a place to create something new and a bit of a utopian … where you can build an offer, really kind of whatever you want for people, and that's incredibly seductive.” (A8) For instance, branded virtual influencers may handle nonurgent and nonsevere cases at customer service touchpoints in the meta verse, similar to virtual customer service agents in Chinese e‐commerce (Utley, 2024). Compared with experiences provided by typical chatbots, this could result in experiences that are more personalized, with a sense of humanness and psychological closeness (Yan et al., 2024), enhancing the customer journey as elaborated by two of the business owners: “It'd be much more engaging being able to meet someone in an immersive environment… that actually enables people to kind of move more freely, in an immersive space and kind of meet up in the meta verse. It's like they're actually kind of in the room together. And I think that's quite powerful.” (A7) “It's a change in paradigm in terms of how people interact … It provides new ways of people being able to build businesses, being able to communicate across borders, being able to innovate in spaces that they haven't before.” (A10) Currently constrained by sensory and autonomy limitations, vir tual influencers, through the application of generative artificial intelligence in the virtual and mixed reality worlds, would offer brands and creators a broad spectrum of opportunities. We propose: P6 Leveraging virtual influencers enhanced with automated advanced technology and predictive ana lytics across virtual spaces holds powerful potential to enhance customer retention by creating interaction opportunities in virtual spaces. 6 | THEORETICAL CONTRIBUTIONS This study contributes to the existing research on virtual influencers in several ways. First, we unravel unsettled scholarship regarding virtual influencer taxonomies and qualities by clarifying the definitions and proposing a nuanced typology of virtual influencers. Our study extended existing studies focusing on virtual influencer marketing and the impli cations of their nuanced anthropomorphic dimensions (e.g., Franke et al., 2023; Gutuleac et al., 2024; Koles et al., 2024; Yan et al., 2024). By assessing four dimensions—form‐relatedness (the anthropomorphism spectrum), content‐relatedness (depth of storyline), agency, and function—our reframed virtual influencer typology explicates and un derpins the different content strategies that virtual influencers use to engage followers. This typology aids academics and marketers in un derstanding the different roles and functions of virtual influencers while minimizing future misconceptions in virtual influencer research (Miao et al., 2022). It will also help creators and practitioners to strategically position or partner with virtual influencers to meet specific objectives. Second, our research‐based analysis and research theme map ping advance understanding of key perspectives in which human and virtual influencers diverge. This analysis and its implications of virtual influencers' differing qualities from human influencers (Igarashi et al., 2024) identify potential theoretical application challenges arising from relying on human‐centric theories for evaluating the virtual influencer phenomenon. Particularly, we draw attention to the limitations of the source credibility framework as a medium for ex plaining the persuasive power of virtual influencers. Our insights into the dynamic technological attributes and designers of virtual influ encers may encourage academics to develop innovative frameworks and interdisciplinary approaches to empirically examine how virtual influencers engage audiences and perform tasks traditionally assigned to human influencers. Additionally, our refined persuasive communication framework explicates nuanced factors enhancing/ limiting human and virtual influencers' efficacy while revealing intersecting research themes for theorizing in future research. It also highlights that brands must pay detailed attention to the form‐ relatedness level of the virtual influencer along with crafting nuanced creative content suitable for the virtual influencer's form, the brands' specific purpose, and the audience segment (Quach et al., 2024). Third, we proposed future research avenues and six propositions framed within the context of online influencer marketing (Leung et al., 2022) anchored by in‐depth interviews with industry experts in the field and previous literature analysis. The suggested research agenda can serve as a springboard for studies considering the application of virtual influencers in areas such as livestream com merce, generative artificial intelligence, and the metaverse. 7 | MANAGERIAL IMPLICATIONS The findings constructed from our in‐depth interviews with leading industry experts provide guidance through six propositions for marketing practitioners, creators, and policymakers. There is extraordinary potential for virtual influencers to develop as artificial intelligence and immersive virtual environments gain prominence. Their innovative potential enable brands to engage broader and younger audiences across new mediums. Virtual influencers' boundless creative story‐worlds and integration with advanced technologies potentially facilitate engaging real‐time interac tions, customer retention, and service functions, while offering promising benefits such as enhanced brand‐consumer relationship management (Zhao et al., 2024). However, the trust benefits are somewhat conten tious due to virtual influencers' fabricated personas and sensory defi ciencies. To address issues of inauthenticity, brands and creators should ensure consistent, relatable storylines and avoid overly perfect aesthetic designs. They should also ensure congruence between the virtual influ encers' persona, messaging styles, and the associated brand. Moreover, with advancements in technology, brands must remain vigilant about virtual influencer identity, misinformation, and manipulation risks (Lim & Lee, 2023), particularly when targeting vulnerable consumer groups. 8 | CONCLUSIONS AND LIMITATIONS This study advances the body of knowledge of the new form of online influencers, virtual influencers, by analyzing research‐based literature and industry reports to elucidate human and virtual influ encers' distinctions and their implications. It highlighted the importance of the differences between the user engagement process and communication strategies between human and virtual influen cers, including the unique mechanisms virtual influencers rely on to fulfill human influencer tasks. Aside from reframing virtual influencer typology and definitional terms, this study also identified divergent research themes of these influencers while contributing six key findings from qualitative data that provide a direction for future practice‐oriented virtual influencer research. Our study has clear limitations in the sample size and quota of the qualitative element. Although the sample is representative, the findings could be tested and nuanced on larger/different samples. As an interpretive study, our conclusions are not definitive: other researchers might infer dif ferent conclusions from the same synthesis of literature and data. In‐ keeping with the principles of interpretive research (Hackley, 2020), we have tried to make our reasoning as transparent as possible. The ongoing advancement in virtual influencers' visual rendering tech nology and areas like generative artificial intelligence, mixed reality, and metaverse are set to transform virtual influencers into artificial intelligent influencers or AI influencers. These evolutions warrant continuous virtual influencer research into their multifaceted impacts and value creations of such technologies. DATA AVAILABILITY STATEMENT The data that support the findings of this study are available from the corresponding author upon reasonable request. ORCID Tippayanet Sorosrungruang https://orcid.org/0009-0002- 7725-2559 Nisreen Ameen http://orcid.org/0000-0002-1794-9103 Chris Hackley https://orcid.org/0000-0002-8354-8834 Ads of the world. (2023). Heinz: A.I Ketchup. Ads of the World. https:// www.adsoftheworld.com/campaigns/a-i-ketchup Ahn, R. J., Cho, S. Y., & Sunny Tsai, W. (2022). Demystifying Computer‐ Generated imagery (CGI) influencers: the effect of perceived anthropomorphism and social presence on brand outcomes. Journal of Interactive Advertising, 22(3), 327–335. https://doi.org/10.1080/ 15252019.2022.2111242 AIT News Desk. (2024). Virtual Influencers That Earn Their Creators Thou sands of Dollars for a Single Picture. https://aithority.com/ technology/virtual-influencers-that-earn-their-creators-thousands of-dollars-for-a-single-picture/ AlRabiah, S., Marder, B., Marshall, D., & Angell, R. (2022). Too much information: An examination of the effects of social self‐disclosure embedded within influencer eWOM campaigns. Journal of Business Research, 152, 93–105. https://doi.org/10.1016/J.JBUSRES.2022. 07.029 Ameen, N., Cheah, J.‐H., Ali, F., El‐Manstrly, D., & Kulyciute, R. (2023). Risk, trust, and the roles of human versus virtual influencers. Journal of Travel Research, 63, 1370–1394. https://doi.org/10.1177/ Ameen, N., Hosany, S., & Taheri, B. (2023). Generation Z's psychology and new‐age technologies: Implications for future research. Psychology & Marketing, 40(10), 2029–2040. https://doi.org/10. 1002/mar.21868 Appel, G., Grewal, L., Hadi, R., & Stephen, A. T. (2020). The future of social media in marketing. Journal of the Academy of Marketing Science, 48(1), 79–95. https://doi.org/10.1007/s11747-019-00695-1 Aw, E. C.‐X., & Chuah, S. H.‐W. (2021). “Stop the unattainable ideal for an ordinary me!” fostering parasocial relationships with social media influencers: The role of self‐discrepancy. Journal of Business Research, 132, 146–157. https://doi.org/10.1016/j.jbusres.2021. 04.025 Baudier, P., de Boissieu, E., & Duchemin, M.‐H. (2023). Source credibility and emotions generated by robot and human influencers: The per ception of luxury brand representatives. Technological Forecasting and Social Change, 187, 122255. https://doi.org/10.1016/j.techfore. 2022.122255 Belanche, D., Casaló, L. V., & Flavián, M. (2024). Human versus virtual influences, a comparative study. Journal of Business Research, 173, 114493. https://doi.org/10.1016/j.jbusres.2023.114493 Belk, R., Fischer, E., & Kozinets, R. (2013). Qualitative Consumer & Mar keting Research. Sage Publications. Bersola, C. (2023). Lotte Duty Free partnersrtwith virtual influencer Nobody Sausage in retailtainment campaign. https://moodiedavittreport.com/ lotte-duty-free-partners-with-virtual-influencer-nobody-sausage-in retailtainment-campaign/ de Boissieu, E., & Baudier, P. (2023). The perceived credibility of human‐ like social robots: Virtual influencers in a luxury and multicultural context. Journal of Organizational Change Management, 36, 1163–1179. https://doi.org/10.1108/JOCM-05-2023-0182 Braun, V., & Clarke, V. (2006). Using thematic analysis in psychology. Qualitative Research in Psychology, 3(2), 77–101. https://doi.org/10. 1191/1478088706QP063OA de Brito Silva, M. J., de Oliveira Ramos Delfino, L., Alves Cerqueira, K., & de Oliveira Campos, P. (2022). Avatar marketing: A study on the engagement and authenticity of virtual influencers on Instagram. Social Network Analysis and Mining, 12(1), 130. https://doi.org/10. 1007/s13278-022-00966-w Brown, S., & Ponsonby‐McCabe, S. (2013). They're gr‐r‐reat! Introduction to the special issue. Journal of Marketing Management, 29(1–2), 1–4. https://doi.org/10.1080/0267257X.2012.762184 Byun, K. J., & Ahn, S. J. (2023). A systematic review of virtual influencers: similarities and differences between human and virtual influencers in interactive advertising. Journal of Interactive Advertising, 23, 293–306. https://doi.org/10.1080/15252019.2023.2236102 Campbell, C., Plangger, K., Sands, S., & Kietzmann, J. (2022). Preparing for an era of deepfakes and AI‐generated ads: A framework for understanding responses to manipulated advertising. Journal of Advertising, 51(1), 22–38. https://doi.org/10.1080/00913367.2021.1909515 Chen, L., Yan, Y., & Smith, A. N. (2023). What drives digital engagement with sponsored videos? An investigation of video influencers' authenticity management strategies. Journal of the Academy of Marketing Science, 51(1), 198–221. https://doi.org/10.1007/ s11747-022-00887-2 Claeys, P., Charry, K., & Tessitore, T. (2024). To be real or not to be real? The effect of genuine (vs. nongenuine) depictions of social media influencers on followers' well‐being and brand purchase intention. Psychology & Marketing, 41(1), 203–222. https://doi.org/10.1002/ MAR.21920 Dwivedi, Y. K., Kshetri, N., Hughes, L., Rana, N. P., Baabdullah, A. M., Kar, A. K., Koohang, A., Ribeiro‐Navarrete, S., Belei, N., Balakrishnan, J., Basu, S., Behl, A., Davies, G. H., Dutot, V., Dwivedi, R., Evans, L., Felix, R., Foster‐Fletcher, R., Giannakis, M., … Yan, M. (2023). Exploring the darkverse: a multi‐perspective analysis of the negative societal impacts of the metaverse. Information Systems Frontiers, 25(5), 2071–2114. https://doi.org/10.1007/ s10796-023-10400-x Emplifi. (2023). The state of influencer marketing. https://go.emplifi.io/ state-of-influencer-marketing-2023.html Ferrari, F., & McKelvey, F. (2023). Hyperproduction: A social theory of deep generative models. Distinktion: Journal of Social Theory, 24(2), 338–360. https://doi.org/10.1080/1600910X.2022.2137546 Franke, C., Groeppel‐Klein, A., & Müller, K. (2023). Consumers' responses to virtual influencers as advertising endorsers: Novel and effective or uncanny and deceiving. Journal of Advertising, 52(4), 523–539. https://doi.org/10.1080/00913367.2022.2154721 Gerrath, M. H. E. E., Olya, H., Shah, Z., & Li, H. (2024). Virtual influencers and pro‐environmental causes: The roles of message warmth and trust in experts. Journal of Business Research, 175, 114520. https:// doi.org/10.1016/j.jbusres.2024.114520 Gutuleac, R., Baima, G., Rizzo, C., & Bresciani, S. (2024). Will virtual in fluencers overcome the uncanny valley? The moderating role of social cues. Psychology & Marketing, 41(7), 1419–1431. https://doi. org/10.1002/MAR.21989 Hackley, C. E. (2020). Qualitative research in marketing and management: doing interpretive research projects. Routledge. https://www. routledge.com/Qualitative-Research-in-Marketing-and Management-Doing-Interpretive-Research-Projects/Hackley/p/ book/9781138332218 Ham, J., Li, S., Shah, P., & Eastin, M. S. (2023). The “mixed” reality of virtual brand endorsers: Understanding the effect of brand engagement and social cues on technological perceptions and advertising effec tiveness. Journal of Interactive Advertising, 23(2), 98–113. https:// doi.org/10.1080/15252019.2023.2185557 Han, M. C. (2021). The impact of anthropomorphism on consumers’ purchase decision in chatbot commerce. Journal of Internet Commerce, 20(1), 46–65. https://doi.org/10.1080/15332861.2020. 1863022 Hennig‐Thurau, T., Aliman, D. N., Herting, A. M., Cziehso, G. P., Linder, M., & Kübler, R. V. (2023). Social interactions in the metaverse: Framework, initial evidence, and research roadmap. Journal of the Academy of Marketing Science, 51(4), 889–913. https://doi.org/10. 1007/s11747-022-00908-0 Hovland, C., Janis, I., & Kelley, H. (1953). Communication and persuasion. Yale University Press. https://psycnet.apa.org/record/1953- 15071-000 Huaman‐Ramirez, R., Lunardo, R., & Vasquez‐Parraga, A. (2022). How brand self‐disclosure helps brands create intimacy with customers: The role of information valence and anthropomorphism. Psychology & Marketing, 39(2), 460–477. https://doi.org/10.1002/MAR.21609 Igarashi, R., Bhoumik, K., & Thompson, J. (2024). Investigating the effec tiveness of virtual influencers in prosocial marketing. Psychology & Marketing, 41(9), 2121–2135. https://doi.org/10.1002/mar.22031 Kietzmann, J., Mills, A. J., & Plangger, K. (2020). Deepfakes: Perspectives on the future “reality” of advertising and branding. International Journal of Advertising, 40(0), 473–485. https://doi.org/10.1080/ 02650487.2020.1834211 Kim, D. Y., & Kim, H.‐Y. (2021). Influencer advertising on social media: The multiple inference model on influencer‐product congruence and sponsorship disclosure. Journal of Business Research, 130, 405–415. https://doi.org/10.1016/j.jbusres.2020.02.020 Kim, H., & Park, M. (2023). Virtual influencers' attractiveness effect on purchase intention: A moderated mediation model of the Product– Endorser fit with the brand. Computers in Human Behavior, 143, 107703. https://doi.org/10.1016/j.chb.2023.107703 Kim, I., Ki, C.‐W., Lee, H., & Kim, Y.‐K. (2024). Virtual influencer marketing: Evaluating the influence of virtual influencers' form realism and behavioral realism on consumer ambivalence and marketing per formance. Journal of Business Research, 176, 114611. https://doi. org/10.1016/j.jbusres.2024.114611 Koles, B., Audrezet, A., Moulard, J. G., Ameen, N., & McKenna, B. (2024). The authentic virtual influencer: Authenticity manifestations in the metaverse. Journal of Business Research, 170, 114325. https://doi. org/10.1016/j.jbusres.2023.114325 Lee, Y.‐H., & Yuan, C. W. (2023). I'm not a puppet, I'm a real boy! Gender presentations by virtual influencers and how they are received. Computers in Human Behavior, 149, 107927. https://doi.org/10. 1016/j.chb.2023.107927 Leung, F. F., Gu, F. F., & Palmatier, R. W. (2022). Online influencer mar keting. Journal of the Academy of Marketing Science, 50(2), 226–251. https://doi.org/10.1007/S11747-021-00829-4/TABLES/4 Li, H., Lei, Y., Zhou, Q., & Yuan, H. (2023). Can you sense without being human? Comparing virtual and human influencers endorsement effectiveness. Journal of Retailing and Consumer Services, 75, 103456. https://doi.org/10.1016/j.jretconser.2023.103456 Li, M. (2022). Influence for social good: Exploring the roles of influencer identity and comment section in Instagram‐based LGBTQ‐centric corporate social responsibility advertising. International Journal of Advertising, 41(3), 462–499. https://doi.org/10.1080/02650487. 2021.1884399 Lim, R. E., & Lee, S. Y. (2023). “You are a virtual influencer!”: Under standing the impact of origin disclosure and emotional narratives on parasocial relationships and virtual influencer credibility. Computers in Human Behavior, 148, 107897. https://doi.org/10.1016/j.chb. 2023.107897 Lou, C., Kiew, S. T. J., Chen, T., Lee, T. Y. M., Ong, J. E. C., & Phua, Z. (2023). Authentically fake? How consumers respond to the influence of virtual influencers. Journal of Advertising, 52(4), 540–557. https:// doi.org/10.1080/00913367.2022.2149641 Lv, L., Huang, M., Guan, D., & Yang, K. (2023). Minor flaws are better: The positive effect of aesthetic imperfection about avatar endorsers on brand authenticity. Journal of Business Research, 166, 114125. https://doi.org/10.1016/j.jbusres.2023.114125 Marriott, H. R., & Pitardi, V. (2024). One is the loneliest number… two can be as bad as one. The influence of AI Friendship Apps on users' well‐ being and addiction. Psychology & Marketing, 41(1), 86–101. https:// doi.org/10.1002/mar.21899 Meng, L., Bie, Y., Yang, M., & Wang, Y. (2024). Watching it motivates me to become stronger: virtual influencers’ impact on consumer self‐ improvement product preferences. Journal of Business Research, 178, 114654. https://doi.org/10.1016/j.jbusres.2024.114654 Meng, L., Bie, Y., Yang, M., & Wang, Y. (2025). The effect of human versus virtual influencers: The roles of destination types and self‐ referencing processes. Tourism Management, 106, 104978. https:// doi.org/10.1016/J.TOURMAN.2024.104978 Miao, F., Kozlenkova, I. V., Wang, H., Xie, T., & Palmatier, R. W. (2022). An emerging theory of avatar marketing. Journal of Marketing, 86(1), 67–90. https://doi.org/10.1177/0022242921996646 Miles, M. B., & Huberman, A. M. (1994). Qualitative data analysis: An expanded sourcebook (2nd ed.). Sage Publications, Inc. https:// psycnet.apa.org/record/1995-97407-000 Mirowska, A., & Arsenyan, J. (2023). Sweet escape: The role of empathy in social media engagement with human versus virtual influencers. International Journal of Human‐Computer Studies, 174, 103008. https://doi.org/10.1016/j.ijhcs.2023.103008 Mouritzen, S. L. T., Penttinen, V., & Pedersen, S. (2023). Virtual influencer marketing: The good, the bad and the unreal. European Journal of Marketing, 58, 410–440. https://doi.org/10.1108/EJM 12-2022-0915 Moustakas, E., Lamba, N., Mahmoud, D., & Ranganathan, C. (2020). Blurring lines between fiction and reality: Perspectives of experts on marketing effectiveness of virtual influencers. 2020 International Conference on Cyber Security and Protection of Digital Services (Cyber Security), 1–6. https://doi.org/10.1109/CyberSecurity49315.2020. Muniz, F., Stewart, K., & Magalhães, L. (2023). Are they humans or are they robots? The effect of virtual influencer disclosure on brand trust. Journal of Consumer Behaviour, 23, 1234–1250. https://doi. org/10.1002/cb.2271 Munnukka, J., Talvitie‐Lamberg, K., & Maity, D. (2022). Anthropomorph ism and social presence in human–virtual service assistant interac tions: The role of dialog length and attitudes. Computers in Human Behavior, 135, 107343. https://doi.org/10.1016/J.CHB.2022. Niederman, F., & Baker, E. W. (2023). Ethics and AI issues: Old container with new wine? Information Systems Frontiers, 25(1), 9–28. https:// doi.org/10.1007/S10796-022-10305-1/TABLES/1 Nowak, K. L., & Rauh, C. (2005). The influence of the avatar on online perceptions of anthropomorphism, androgyny, credibility, homo phily, and attraction. Journal of Computer‐Mediated Communication, 11(1), 153–178. https://doi.org/10.1111/j.1083-6101.2006. tb00308.x Ozdemir, O., Kolfal, B., Messinger, P. R., & Rizvi, S. (2023). Human or virtual: How influencer type shapes brand attitudes. Computers in Human Behavior, 145, 107771. https://doi.org/10.1016/j.chb.2023. 107771 Quach, S., Cheah, I., & Thaichon, P. (2024). The power of flattery: En hancing prosocial behavior through virtual influencers. Psychology & Marketing, 41(7), 1629–1648. https://doi.org/10.1002/MAR.22001 Radomskaya, V., & Pearce, P. L. (2021). Adding character: The role of destination mascots in tourism development. Tourism Management, 84, 104248. https://doi.org/10.1016/j.tourman.2020.104248 Sands, S., Campbell, C. L., Plangger, K., & Ferraro, C. (2022). Unreal influence: Leveraging AI in influencer marketing. European Journal of Marketing, 56(6), 1721–1747. https://doi.org/10.1108/EJM-12-2019-0949 Sands, S., Ferraro, C., Demsar, V., & Chandler, G. (2022). False idols: Unpacking the opportunities and challenges of falsity in the context of virtual influencers. Business Horizons, 65(6), 777–788. https://doi. org/10.1016/j.bushor.2022.08.002 Shen, R.‐P., Liu, D., Wei, X., & Zhang, M. (2022). Your posts betray you: Detecting influencer‐generated sponsored posts by finding the right clues. Information & Management, 59(8), 103719. https://doi.org/10. 1016/j.im.2022.103719 Statista. (2022). Engagement rates among IG influencers worldwide 2021. https://www.statista.com/statistics/992887/growth-engagement rate-influencers-followers/ Strauss, A., & Corbin, J. (1998). Basics of qualitative research: Techniques and procedures for developing grounded theory. (2nd ed.). Sage Publications, Inc. https://psycnet.apa.org/record/1999-02001-000 Taylor, E. (2023). The pros and cons of virtual influencers. https://emplifi.io/ resources/blog/virtual-influencers-pros-and-cons Thomas, V. L., & Fowler, K. (2021). Close encounters of the AI kind: Use of AI influencers as brand endorsers. Journal of Advertising, 50(1), 11–25. https://doi.org/10.1080/00913367.2020.1810595 Utley, E. (2024). From niche to necessary: The rise of AI‐powered virtual avatars in Chinese E‐Commerce, Alizila News from Alibaba. https:// www.alizila.com/ai-virtual-avatars-chinese-ecommerce-retail-2024/ Wan, C., Lee, D., & Ng, P. (2024). The role of anthropomorphism and racial homophily of virtual influencers in encouraging low‐ versus high‐ cost pro‐environmental behaviors. Psychology & Marketing, 41(8), 1833–1853. https://doi.org/10.1002/mar.22013 Wang, Q., Miao, F., Tayi, G. K., & Xie, E. (2019). What makes online content viral? The contingent effects of hub users versus non–hub users on social media platforms. Journal of the Academy of Marketing Science, 47(6), 1005–1026. https://doi.org/10.1007/S11747- 019-00678-2/TABLES/9 Wu, Y., Nambisan, S., Xiao, J., & Xie, K. (2022). Consumer resource integration and service innovation in social commerce: The role of social media influencers. Journal of the Academy of Marketing Science, 50(3), 429–459. https://doi.org/10.1007/S11747-022-00837-Y/TABLES/6 Xie‐Carson, L., Benckendorff, P., & Hughes, K. (2023). Not so different after all? A netnographic exploration of user engagement with non‐ human influencers on social media. Journal of Business Research, 167, 114149. https://doi.org/10.1016/j.jbusres.2023.114149 Xie‐Carson, L., Magor, T., Benckendorff, P., & Hughes, K. (2023). All hype or the real deal? Investigating user engagement with virtual influ encers in tourism. Tourism Management, 99, 104779. https://doi. org/10.1016/j.tourman.2023.104779 Yan, J., Xia, S., Jiang, A., & Lin, Z. (2024). The effect of different types of virtual influencers on consumers' emotional attachment. Journal of Business Research, 177, 114646. https://doi.org/10.1016/j.jbusres. 2024.114646 Ye, G., Hudders, L., De Jans, S., & De Veirman, M. (2021). The value of influencer marketing for business: A bibliometric analysis and man agerial implications. Journal of Advertising, 50(2), 160–178. https:// doi.org/10.1080/00913367.2020.1857888 Yu, J., Dickinger, A., So, K. K. F., & Egger, R. (2024). Artificial intelligence‐ generated virtual influencer: Examining the effects of emotional display on user engagement. Journal of Retailing and Consumer Services, 76, 103560. https://doi.org/10.1016/j.jretconser.2023.103560 Zelenskaya, M., & Rundle‐Thiele, S. (2022). Wildlife watcher kylie: Co‐ designing a virtual ambassador for the koala watch programme. Studies in Australasian Cinema, 16(2–3), 93–108. https://doi.org/10. 1080/17503175.2022.2148413 Zhang, X., Shi, Y., Li, T., Guan, Y., & Cui, X. (2023). How do virtual AI streamers influence viewers' livestream shopping behavior? The effects of persuasive factors and the mediating role of arousal. Information Systems Frontiers, 1, 1–32. https://doi.org/10.1007/ s10796-023-10425-2 Zhao, T., Ran, Y., Wu, B., Lynette Wang, V., Zhou, L., & Lu Wang, C. (2024). Virtual versus human: Unraveling consumer reactions to service failures through influencer types. Journal of Business Research, 178, 114657. https://doi.org/10.1016/j.jbusres.2024.114657 Zhou, S., Blazquez, M., McCormick, H., & Barnes, L. (2021). How social media influencers' narrative strategies benefit cultivating influencer marketing: Tackling issues of cultural barriers, commercialised con tent, and sponsorship disclosure. Journal of Business Research, 134, 122–142. https://doi.org/10.1016/j.jbusres.2021.05.011 SUPPORTING INFORMATION Additional supporting information can be found online in the Sup porting Information section at the end of this article. How to cite this article: Sorosrungruang, T., Ameen, N., & Hackley, C. (2024). How real is real enough? Unveiling the diverse power of generative AI‐enabled virtual influencers and the dynamics of human responses. Psychology & Marketing, 41, 3124–3143. https://doi.org/10.1002/mar.22105

Showing the abstract — retrieve the full paper via the Exa API.