Does Form Realism Make AI Greenfluencers More Persuasive? Examining the Roles of Perceived Behavioral Realism, Transportation, and Parasocial Interaction

Research Purpose
AI influencers, also known as virtual influencers (VIs), are computer-generated personalities designed to imitate, interact with, and influence humans on social media (Kim et al., 2024a). These entities have garnered significant attention from both academia and industry professionals. Beyond their role in commercial endorsements, previous studies have illustrated the potential of AI influencers as “green endorsers” capable of delivering compelling messages to enhance pro-environmental engagement (Gerrath et al., 2024).

Miao et al. (2022) introduced a theory of avatar marketing that underscores the importance of virtual avatars’ form realism (i.e., human-like appearances) and behavioral realism (e.g., human-like behaviors) in determining marketing effectiveness. More importantly, the theory posits that it is the congruence between form realism and behavioral realism that is pivotal for achieving persuasive outcomes. Nonetheless, existing research has predominantly concentrated on the influence of AI influencers' form realism (i.e., anthropomorphic appearances, visual human-likeness) in shaping audience attitudes and behaviors (Arsenyan & Mirowskwa, 2021; Kim et al., 2024a; 2024b). This focus has largely overlooked the communicative elements of AI influencers on social media (such as their narrative capabilities), and how these elements interact with form realism to influence perceived behavioral realism. To address this research gap, we propose the first research question:

RQ1: Do AI influencers’ form realism and narrative capability interact to affect perceived behavioral realism?

Narrative persuasion literature suggests that transportation into a narrative and immersive involvement with characters in stories enable audiences to engage in parasocial interactions and establish emotional bonds with these characters. As a result, audiences tend to be less skeptical towards the persuasive messages conveyed (Boukes & LaMarre, 2021; Green & Brock, 2000; Murphy et al. 2013). As such, further research is needed to unravel mechanisms that account for AI influencers’ narrative persuasiveness in pro-environmental campaigns. This leads to our second research question:

RQ2: How does the behavioral realism induced by AI influencers’ form realism and narrative capability impact pro-environmental intention?

Methodology
To address these research gaps, this study utilized a 2 (form realism: low vs. high) × 2 (narrative capability: yes vs. no) between-subjects factorial design to investigate the influence of form realism and narrative capability on the intention to adopt pro-environmental lifestyles. Additionally, the study explored the roles of perceived behavioral realism, transportation, and parasocial interaction as serial mediators.

A priori power analyses using G*Power indicated that a minimum sample size of 158 participants was necessary to attain at least 80% power for this research design. A convenience sample of 161 social media users was recruited to participate in the online experiment. After excluding 5 entries that failed the attention checks, the final sample consists 156 participants, of whom 60.3% were female with ages ranging from 18 to 58 (M = 30.09, SD = 7.06).

Participants who provided informed consent completed a pre-test questionnaire that includes control variables (e.g., perceived familiarity with AI influencer, etc.). Following this, participants were first directed to an introduction page: a mock-up magazine article about a fictitious AI influencer (which was determined with a pilot test). Then, they were randomly assigned to one of four conditions, where they viewed a mock-up RedNote post featuring the same fictitious AI influencer’s endorsement of pro-environmental behaviors. After 25s, participants completed a post-test questionnaire, including the manipulation checks, dependent measures, and demographics. All multi-item measures demonstrated good reliability (Cronbach’s alpha > .78). Post-hoc checks confirmed that the manipulations for form realism and narrative capability were successful.

Findings and Contributions
To answer RQ1, a two-way ANCOVA was conducted. The results indicate that form realism has a main effect on behavioral realism, F(1, 148) = 21.11, p < .001 , η_p² = .13. Participants exposed to AI influencers with high form realism reported higher levels of perceived behavioral realism (M = 5.38, SE = .12) compared to those exposed to AI influencers with low form realism (M = 4.63, SE = .12). Similarly, the narrative capability demonstrated by the AI influencer is associated with higher levels of perceived behavioral realism. However, the interaction effect of form realism and narrative capability on perceived behavioral realism was non-significant, F(1, 148) = .00, p = .93, η_p² = .00.

To answer RQ2, a serial mediation analysis was conducted using PROCESS Model 6 with 10,000 bootstrap samples. The results indicate that a highly realistic AI influencer could exhibit higher perceived behavioral realism. This increased perceived behavioral realism led to greater transportation and parasocial experiences, resulting in higher intentions to adopt a pro-environmental lifestyle (B = 0.05, SE = 0.03; 95% CIs [ 0.01, 0.12]). Similarly, the narrative capability of the AI influencer also had a significant effect on pro-environmental intentions, which was serially mediated by perceived behavioral realism, transportation, and parasocial interaction (B = 0.03, SE = 0.02; 95% CIs [ 0.00, 0.07]).

The research sought to broaden academic knowledge regarding the efficacy of AI influencers as advocates for environmental causes. It assessed the roles of AI avatars’ form realism and narrative capability in facilitating consumers’ pro-environmental behavioral intentions through perceived behavioral realism. While prior research emphasizes the importance of VIs' anthropomorphism, this study, based on Miao et al. (2022)’s avatar marketing theory, goes beyond human-like appearances to examine the relationship between form realism and behavioral realism in determining persuasiveness. Additionally, this study contributes to the narrative persuasion by highlighting transportation and parasocial interaction with AI influencers as crucial precursors to behavioral intention. These observations extend and underscore the importance of narrative persuasion literature from human to non-human communication by identifying transportation and parasocial interaction as crucial mechanisms that enhance the persuasiveness of AI influencers' endorsements.