The Metaverse's Impact on AI News Anchors: Unveiling User Engagement Factors

By Dr Silpaja Chandrasekar, PhDReviewed by Susha Cheriyedath, M.Sc.Aug 24 2023

In a paper published in the journal Systems, researchers examined the growing influence of the Metaverse on the meta-human industry and human-artificial intelligence (AI) interactions. This study used an expectation confirmation theory-based model to examine the replacement of traditional newscasters with AI news anchors in China, identifying factors such as perceived intelligence, satisfaction, and trust that influenced users' continued intention to watch news from AI anchors. The findings provided valuable insights for commercializing AI news anchors, explaining over 80% of the variance in user intent.

Background

The concept of the Metaverse has spurred remarkable advancements in the realm of meta-humans, permeating a range of industries, including education, retail, healthcare, and gaming. Of particular note is the emergence of virtual anchors, who are gaining substantial traction as key conduits for news delivery and interactive experiences. This trend has found fertile ground in China, where projections indicate that the market value of virtual anchors is poised to soar to an impressive 2.5 billion Chinese Yuan Renminbi (CNY) by 2024.

Related work

Previous studies have examined the investigation into users’ enduring engagement with virtual anchors within the Expectation Confirmation Model (ECM) framework. This established model has previously been employed to delve into factors such as perceived intelligence, confirmation, satisfaction, and continuance intention. By leveraging insights derived from earlier research and integrating distinctive attributes like anthropomorphism specific to AI anchors, the present study explores the intricate mechanisms underlying users’ consistent interaction with AI news anchors.

Proposed method

Based on the ECM, the framework proposed in this study includes factors such as Continuance Intention (CI), Satisfaction (SAT), Confirmation of Expectation (CE), Trust (TRU), Perceived Anthropomorphism (ANT), Perceived Intelligence (PI), Perceived Novelty (PN), Perceived Attractiveness (PA), and Information Quality (IQ). CI, rooted in users’ prior experiences, denotes the sustained intent to use technology over the long term. CI is adopted to comprehend users’ acceptance of the ongoing commercialization of AI news anchors. CE, resulting from the juxtaposition of pre-usage expectations with post-usage performance, contributes to users’ confirmation or disconfirmation experience. Subsequently, SAT reflects users’ emotional response to their prior consumption experience.

Trust, pivotal for human-AI interaction, encompasses cognitive and emotional dimensions, facilitating successful AI anchor adoption. PA captures human-like attributes in non-human entities, influencing positive feelings and engagement. PI reflects users’ evaluation of the competence and performance of AI. In contrast, PN corresponds to users’ quest for novel experiences, affecting satisfaction. PA also evaluates the physical appeal of AI anchors, influencing attitudes. IQ assesses content credibility and accuracy, tied to user trust. This study endeavors to unravel the intricate interplay of these factors in shaping users’ continuous engagement with AI news anchors.

To ensure participant understanding, an introduction and a 3-minute 30-second video showcasing AI anchors "Xiao C" and "Shen Xiaoya" were presented. These anchors were selected for their influence, current content, and moderate anthropomorphism. Via an online survey conducted from March to April 2023, 598 valid responses were collected. The participants, evenly split by gender, were mainly educated individuals aged 18-50 from diverse regions, including China, Europe, and Australia. Most had limited exposure to AI news anchors, with 29.9% having watched videos and 58.9% having some knowledge about them. Variables were measured on a six-point scale for clarity, following Rahman's approach.

Experimental results

The measurement model, evaluated using the PLS algorithm, demonstrated strong indicator reliability and satisfactory convergent validity. The structural model, assessed through bootstrapping, supported 11 out of 14 hypotheses. Notably, SAT positively influenced CI, TRU positively predicted both CI and SAT, Perceived Anthropomorphism ANT positively affected SAT, PI positively predicted TRU, CI, and CE, and PA positively impacted SAT. IQ positively influenced trust. The model exhibited substantial predictive and explanatory capacities, with R2 and Q2 values above zero for endogenous variables.

Mediation and moderation analyses revealed strong mediating effects for ANT and CE on CI, SAT on CI, and IQ on TRU through partial mediation of TRU and SAT. Gender moderated the ANT-CI relationship, and previous consumption moderated the PN-SAT relationship. Age, education, attention to AI news anchors, and knowledge about AI news anchors did not significantly moderate.

The study investigated users’ intention to keep watching AI news anchors, finding around 64.0% to 70.1% with positive intention; SAT, PI, and TRU were the key drivers. Factors like ANT, attractiveness, confirmation, and IQ indirectly affected intention through satisfaction. TRU also mediated between PI and intention. Gender and prior exposure influenced certain relationships. PN did not directly predict satisfaction, more as a tool than novelty. For those who watched AI news anchor videos, the novelty was related to SAT due to curiosity.

Conclusion

To conclude, this paper presents a comprehensive theoretical model that employs constructs such as ANT, PI, PA, IQ, PN, and TRU to elucidate users' Cognitive Interest (CI) regarding AI news anchors. The analysis reveals that CI is influenced directly by factors like SAT, TRU, and PI. Additionally, PA and CE impact CI by mediating through the avenue of satisfaction. Moreover, gender and prior experience are found to moderate relationships, such as ANT & CI and PN & SAT.

The practical implications drawn from the study emphasize the importance of focusing on the utility of AI anchors, enhancing algorithm transparency, minimizing anthropomorphism, and leveraging users’ strengths. However, it’s important to note the limitations of the research, including its cross-sectional approach and limited exposure to AI anchors. In the future, employing longitudinal studies and observational data could provide insights into the sustained attention towards AI news anchors over the long term.