Earticle

In platform work environments, algorithmic dispatch systems function as a central mechanism for both efficiency and control. However, prior research has primarily focused on the technical features of algorithms or workers' static perceptions, lacking empirical analysis of how platform workers' responses evolve over time. Therefore, this study aimed to investigate how delivery riders' perceptions and reactions to algorithmic dispatch evolve. A mixed-methods approach was employed in this study: grounded theory analysis of interviews revealed three core themes, including AI dispatch and operational systems, delivery income and work intensity, and rider skills and work practices. These qualitative insights provided a deeper understanding of riders' experience-based strategies and the factors influencing their algorithmic dependence. OLS regression using three waves of survey data identified key factors influencing algorithmic dependence. Results show that algorithmic dependence significantly decreases as operational load increases and over time. In particular, the study found that riders with higher operational load demonstrate a preference for autonomous decision-making over algorithmic control. Additionally, over the three survey waves, riders' reliance on algorithmic dispatch diminished as they became more familiar with the system. These findings highlight that acceptance of algorithmic control is not static but dynamic and adaptive. This study deepens the understanding of algorithm-user interaction and offers meaningful implications for platform algorithm design and labor policy, suggesting ways to enhance user engagement and mitigate resistance.

Electronic word-of-mouth (WOM) plays a crucial role in movie-going decisions, yet spoiler content in reviews remains controversial. Although spoilers may potentially harm enjoyment, they may also alleviate uncertainty, thereby enhancing the viewing experience. Despite such tension, there has been a lack of research on which types of spoilers have either detrimental or informative effects on decision-making. To address this gap, we propose a framework to classify spoiler reviews as either detrimental or informative using natural language processing techniques. Based on the classification results, we conducted econometric analyses to assess their distinct impacts on box office revenue by using data from 322 movies. Our overall findings show that informative spoilers increases box office revenue whereas detrimental spoilers have no significant effect. However, moderating analyses reveal important context effects. Detrimental spoilers reduce revenue for narrative-driven or long-duration films and during weekends but show null effects in other contexts. Similarly, informative spoilers do not always increase demand, with their positive effects diminishing under certain conditions. This study offers valuable insights for review platforms and studios by highlighting the need for nuanced spoiler management to enhance audience engagement and revenue.

Generative AI is increasingly shaping human experiences across creative, social, and professional domains. While its digital capabilities offer new opportunities for enhancing user engagement, personalization, and productivity, its dual impact on users’ psychological needs remains underexplored. Grounded in self-determination theory, this study investigates how generative AI both satisfies and frustrates users’ psychological needs and examines the broader implications of this dynamic. Employing a netnographic methodology, the research draws on over 1.5-year of naturalistic online AI Community observation, resulting in a rich dataset comprising 2,062 pages of data. This research uncovers paradoxical relationships between generative AI features and users’ needs – autonomy, competence, and relatedness – highlighting the simultaneous fulfillment and frustration they can provoke. These findings offer critical insights for the user-centered design of generative AI systems, and have implications for developers, business professionals, and policymakers aiming to balance innovation with psychological well-being.

Unlike company created OBCs, Member created online brand communities (OBCs), are characterised by stronger social activities and more diverse topics that may not be directly linked to the brands. This raises the question of whether active participation in member initiated OBCs leads to stronger brand loyalty. This paper applies the commitment-trust theory to test if users participation in member-initiated OBCs increases the loyalty of brand within community members. The commitment-trust theory indicates that correspondence bonds and reliability are key factors of significant relational variables in relationship marketing. We identify information quality and social usefulness as antecedents of correspondence bonds and reliability and brand loyalty as outcomes of the relationships. The model is verified through data obtained from 530 users of IT related online brand communities in South Korea. The observations of this work indicates the increase of brand bonds through sense of belongings and information quality that lead to improved trust and commitment on the OBCs. In particular, sense of belonging is an important factor for member initiated OBCs due to the higher level of independence from the brand companies. Also,we test the Direct, Indirect, and Total Effects of Trust on Brand bonds via OBC Commitment. The results highlight partial mediation between Trust and Brand Loyalty through OBC Commitment. The bootstrapping analysis also supports the robustness of the indirect effect, providing further evidence of the mediation process and relaiablity plays a critical role in influencing Brand Bonds.

Recent developments in user-centered artificial intelligence (AI) services, especially large language model (LLM) systems, have significantly reshaped human-computer interactions by enabling more natural and conversational engagements. These advancements have facilitated the automation of corporate marketing functions such as customer support, content creation, and personalized communication. Consequently, businesses are increasingly adopting AI-driven conversational agents to improve service quality, while consumers have come to expect real-time, relevant, and high-quality responses. This study proposes a dual-path model to explore how ChatGPT’s human-like characteristics and performance-related attributes affect user satisfaction with AI services. By examining both human-like and performance factors, this research aims to offer a comprehensive understanding of AI service adoption. The findings provide valuable insights for the academic community and offer practical implications for both understanding users of ChatGPT and businesses seeking to enhance their AI-based customer engagement strategies.

Netnography is the practice of studying cultural attributes in human beings, or the process of analyzing and exploring the cultural logic or perspectives behind a particular group or community with a common interest. This research aims to clearly declare what kind of practice the ChatGPT lovers do in the ChatGPT community for their common goals. This netnographic study examined text data taken from posts by ChatGPT enthusiasts in the ChatGPT Super Topic Community on Sina Weibo. ChatGPT enthusiasts can share ChatGPT operation methods and their feelings to better facilitate the use of the ChatGPT system and enable the value cocreation process among users (i.e., the value practice process). ChatGPT community on Sina Weibo were the research objects, and the information in online posts was crawled and summarized through text mining. Text content was encoded by the method of initial coding and selective coding using NVivo 20 software. Finally, nine value practice types were identified based on Latent Dirichlet Allocation (LDA) verification in Python in this community for ChatGPT enthusiasts. This study serves as a preliminary exploration of the theoretical framework of value practice in the ChatGPT community and provides inspiration for subsequent researchers to explore ChatGPT-related issues.

Accurate classification of public policy is essential for effective policy analysis, design, comparison, and formulation across countries. However, manual classification by policy experts can lead to inconsistencies and human errors, highlighting the need for a more reliable and efficient approach. This study proposes a deep learning-based model to support policy classification using artificial intelligence. Leveraging Korean policy datasets, comprising administrative data (1988–2018), legislative data (1987–2018), and media data (1988–2020), previously curated by experts, we developed an AI model for automated policy classification based on the KoBERT language model. Designed as a supplementary tool for policy experts, this model enhances classification consistency, reduces decision-making time, and improves overall productivity. Moreover, the model enables the classification, comparison, and evaluation of diverse policies at both local and national levels, offering valuable support for strategic public policy development. The proposed model achieved a Top-1 accuracy of 62.4% and a Top-3 accuracy of 71.6%, outperforming traditional baselines and demonstrating its practical potential for real-world policy analysis.

Due to the rapid growth of digital finance, it is important to understand the factors influencing user adoption. The current study examined how personality traits (PT) drive users towards FinTech adoption (FIA). This study employs the Big Five Personality Model to explore the antecedents of FIA. A partial modeling structural equation modeling (PLS-SEM) and fuzzy-set qualitative comparative analysis (fsQCA) methodology was used for finding out the relationship between PT and FIA. The outcomes from the analysis indicate that all the proposed variables directly and positively influence FIA. These findings have implications for FinTech service providers and academicians for their understanding of the key characteristics driving this sector. This study significantly enhances the existing literature by examining the role of PT, User Innovativeness (UI), and Financial Literacy (FL) as key drivers of FIA, offering unique and innovative insights into these determinants. Furthermore, the study provides a vigorous methodological contribution by integrating linear PLS and nonlinear fsQCA techniques.

This study examines how varying degrees of fidelity in key metaverse elements―avatars, environmental space, and branded products―affect consumer responses within metaverse-based consumption contexts. A survey was distributed to metaverse users from general participants in South Korea. We conducted three experimental studies to test the role of avatar, environmental space and branded products. By distinguishing between the categories of metaverse implementation, we found that real world based-features lead to a more positive consumer response in each setting. This study categorizes the metaverse spaces consumers currently engage with into two primary types: reality-based and virtual-based environments. By distinguishing between these two environments, this study seeks to understand which aspects of metaverse features lead to a more positive consumer response in each setting.

Recently, many companies are developing platform business models by applying IoT, big data, and AI technologies as a part of their digital transformation (DT) strategies. Using qualitative research method, the purpose of this study is to investigate how manufacturing companies use IT vendor’s capabilities to expand, integrate, and reorganize their capabilities in pursuing the new platform business model. Case studies were conducted on manufacturing companies that are pursuing big data-based platform business as their DT strategies. It was founded that the roles of IT vendors include joint development of the technology required for the platform business, formation of business ecosystem for platform services, and co-selling/co-marketing of the services to the target customers. Since the cases are limited to manufacturing companies, there may be limitations in the generalization of research results. Thus, more studies across various industries are suggested for the future study.

This research investigates the economic impact of remote work policies following the COVID-19 pandemic, and whether remote work created or destroyed value. It delineates the shift from traditional office work to remote work, highlighting economic challenges presented by decreased collaboration, diminished corporate culture, and potential losses in productivity. The research employed an event study methodology to examine 33 companies implementing remote work policies post-COVID-19. Of these, 22 showed statistically significant results, with 12 increasing firm value and 10 experiencing a decline. Companies that succeeded in enhancing value shared key factors, such as robust digital infrastructure, flexible work policies, strong leadership, and continuous employee support. These strategies improved collaboration and productivity while reducing operational costs. Conversely, unsuccessful remote work programs struggled with diminished collaboration, disengaged employees, and operational inefficiencies. Common problems included rapid transition without sufficient infrastructure, cybersecurity risks, and investor skepticism. Research findings showed that while some companies successfully created value through robust digital infrastructure, flexibility, and strong leadership, others saw a decline in value due to rapid transitions, loss of innovation, and poor management of remote work logistics. These failures highlight the need for well-planned remote work strategies.