The rapid digitalization of travel services presents both opportunities and challenges for older adults, yet the mechanisms underlying their proficiency in digital travel technologies remain underexplored. Drawing on a questionnaire survey conducted in China, an integrated theoretical framework (Technology perception → Digital travel proficiency → Activity participation → Life satisfaction) is developed to investigate the antecedents and consequences of digital travel proficiency among older adults. Building on the foundational finding that digital travel skills enhance life satisfaction, our study reveals several novel insights. First, digital travel proficiency does not uniformly benefit all forms of activity participation; rather, its effect is contingent on the nature of the activity, enhancing physical exercise and medical visits (an enhancement effect), yet substituting for daily shopping and social interaction (a substitution effect). Second, using random forest modeling to disaggregate technology perception constructs into their constituent measurement items and proficiency into four service-based skills, we find that the importance of perceived usefulness and perceived risk varies considerably by service type. Third, while SEM confirms that all technology perception variables significantly influence digital travel proficiency, random forest modeling further reveals that perceived social support consistently ranks as the least important predictor across all service types, calling for a shift from informal support to structured capacity-building. Our study contributes to a deeper understanding of the mechanisms underlying older adults' digital travel proficiency and its differentiated effects across service types and offers practical implications for businesses, policymakers, and researchers engaged in smart mobility systems and healthy aging.
This study investigates how virtual influencer (VI) types (human-like vs. animal-type) impact consumer responses in eco-friendly advertising. Integrating Construal Level Theory (CLT) and the Elaboration Likelihood Model (ELM), we propose that VI types influence reactions through psychological distance and message credibility, moderated by product involvement. Analysis of two online experiments revealed that in Experiment 1, animal-type VIs fostered closer psychological distance than human-like VI, thereby increasing message credibility, and the mediating effect of psychological distance was confirmed. In Experiment 2, message credibility was found to fully mediate the relationship between the VI type and brand attitude and purchase intention. Furthermore, a moderated mediating effect of product involvement was confirmed, indicating that the influence of VI type is stronger under low-involvement conditions and weaker under high-involvement conditions. These findings offer practical implications for digital marketing strategies.
With nearly 4.9 billion people worldwide engaging daily on social media, digital platforms have become central sites for shaping social norms and regulating behavior. This study examines the relationship between the social empathy comparative value of social media – defined as users' perceptions of social networking sites (SNS) as uniquely distinct from traditional media formats in their ability to foster empathy - cognitive elaboration, and informal social control, expanding on existing research in media effects and digital social behavior. Drawing upon a two-wave panel survey of Spanish citizens (T2 N = 570), we test a mediation model in which cognitive elaboration acts as a key mechanism linking social empathy's comparative value to informal social control. Findings indicate that while social media's ability to foster social empathy positively predicts cognitive elaboration, neither cognitive elaboration nor social empathy directly influences informal social control. However, mediation analysis reveals an indirect effect, suggesting that social media's role in enhancing empathy encourages deeper cognitive processing, which in turn increases the likelihood of engaging in informal social regulation. These findings contribute to ongoing discussions about how digital media shapes individuals' cognitive engagement and their willingness to regulate social behavior in online environments.
The rapid emergence of autonomous taxis has generated both enthusiasm and concern, with media discourse playing a central role in shaping public perceptions of this novel technology. While prior research on technology adoption has primarily emphasized positive functional evaluations, less attention has been paid to how media shape negative risk-based judgments. Addressing this gap, this study proposes a dual-valence evaluative framework that conceptualizes adoption intentions as the joint outcome of positive functional evaluations and negative risk perceptions. Drawing on a survey conducted in Wuhan, China (N = 1163), the study shows that media exposure is associated with adoption intentions indirectly through asymmetric evaluative pathways. Media exposure strengthens perceptions of usefulness and ease of use, which remain the dominant drivers of adoption. At the same time, media exposure attenuates affective risk perceptions, thereby weakening their inhibitory effect on adoption, while cognitive risk perceptions remain largely unaffected. These findings suggest that media influence does not operate uniformly across evaluative dimensions. To contextualize this framework, a large-scale content analysis of Chinese news coverage on autonomous taxis (N = 22,380) reveals that media discourse overwhelmingly embraces the technology, privileging optimistic narratives of technological progress and market potential while devoting limited attention to safety and risk. Together, the findings point to a media-driven evaluative asymmetry in which a predominantly positive media environment may reinforce functional beliefs and reduce affective risk responses surrounding novel technologies.
As artificial intelligence (AI) systems increasingly assume roles with social, educational, and emotional significance, understanding the psychological drivers behind individuals' readiness to delegate such roles to AI is crucial. Drawing on Self-Determination Theory (SDT), this study examines how the satisfaction of basic psychological needs (autonomy, competence, and relatedness) predicts individuals' readiness to delegate socially significant roles to AI across four domains (education, healthcare, mental health, and companionship) and 35 nations. Using data from over 35,000 participants in the 2023 Global Digital Wellbeing Survey, we applied Bayesian multilevel multivariate modelling to assess both global and culture-specific motivational associations. Results revealed that competence emerged as the most consistent predictor of AI delegation across roles, particularly in high-stakes, expertise-driven contexts (healthcare and education). Autonomy showed reliable positive associations in socially and emotionally salient domains (mental health support and companionship). Relatedness showed the weakest and most inconsistent effects but exhibited culturally contingent activation in regions such as the Confucian-Asian and Arab-Islamic clusters. Notably, motivational effects were not uniform across cultures: for example, the influence of autonomy was amplified in South Asian regions, while Nordic countries exhibited low delegation readiness despite culturally autonomy-supportive values. These findings highlight the potential importance of psychological need satisfaction and its cultural modulation in shaping global AI adoption. They also suggest that AI systems should be designed not only for functionality but also for motivational alignment. We discuss implications for ethical design, cross-cultural deployment, and the future of human-AI interaction in socially meaningful domains.
Systematic literature reviews (SLRs) are increasingly supported by large language models (LLMs), but existing evaluations rarely go beyond accuracy and overlook broader information quality dimensions. This study provides an Information Quality (IQ)-driven evaluation of LLM support for SLR tasks in communication research, focused on diagnosing failure modes and their implications for responsible use. Using 1,325 articles on cross-platform social media research, we design a seven-step evaluation pipeline that mirrors core SLR decisions and maps them onto contextual, intrinsic, and representational IQ dimensions. Two leading LLMs, GPT-4o and Claude 3.7 Sonnet, classify or extract information about article type and empirical status (empirical vs. non-empirical), data sources (e.g., media types and platform names), data collection methods, and the number and names of social media platforms. Validated human-coded annotations serve as ground truth, and mixed-method insights are reported from both quantitative performance analysis and qualitative error analysis. Both models achieve near-perfect accuracy for document type identification, yet their performance on tasks requiring boundary-setting judgments, methodological interpretation, or exact structured extraction is lower and more variable. Qualitative error analysis reveals that both models rely on surface-level cues rather than holistic reasoning; they apply narrow rules mechanically and fail to reconcile conflicting signals across different parts of a paper. The two models also exhibit task-specific error profiles rather than stable opposite decision biases: for example, GPT-4o is more conservative in digital trace detection but more expansive in platform-count extraction, while Claude 3.7 Sonnet shows stronger performance on grounded platform extraction. We conclude that current LLMs can reliably assist, but not replace, human judgment in SLRs. Effective deployment requires making model reasoning visible, designing prompts that account for the full range of methods in a field, and using models with complementary task-specific error profiles to flag ambiguous cases for human review.
Background Artificial intelligence (AI) is impacting the U.S. labor market in complex and uneven ways. While prior research has offered important insights into occupational and regional exposure to AI, existing approaches do not fully capture reskilling urgency, which is shaped not only by AI exposure but also by key labor-market factors such as skill non-transferability and job non-specialization. To address this gap, this study introduces a more comprehensive framework through the Reskilling Urgency Index (RUI). Method We compute RUI scores for 879 occupations by integrating data from O*NET, AI occupational exposure scores from prior studies, and state-level employment data from the Bureau of Labor Statistics. Findings Our findings indicate that occupations such as computer systems engineers and architects, administrative services managers, purchasing agents, web developers, financial managers, paralegals, telemarketers, and clerical or customer service workers face the highest reskilling urgency, likely due to the combination of high AI exposure, limited skill transferability, and intermediate levels of occupational specialization. State-level analysis further reveals that reskilling urgency is geographically uneven, with higher RUI observed in the District of Columbia and in states such as Massachusetts, Maryland, Illinois, Texas, Arizona, Connecticut, Utah, and Florida. Implications Overall, these results suggest that reskilling strategies and workforce policies should target not only occupations with high AI exposure but also those characterized by limited skill transferability and intermediate specialization. Moreover, such strategies should adopt a place-based approach that accounts for regional labor-market characteristics and place-specific workforce needs.
The relationship between Internet use and income illustrates fundamental processes in the reproduction of social inequality in the digital age, yet its cyclical and dynamic nature is neglected. Drawing on theories about the interplay between social and digital inequalities, especially the Resources and Appropriation Theory, this study develops a dynamic framework linking resource endowments, distinct patterns of Internet use, and feedback effects. Using data from three waves of China Family Panel Studies (2014-2018) and cross-lagged panel models, this study identifies a general positive cyclical reproduction between Internet use and income, while highlighting heterogeneity and temporal shifts across activity types. Specifically, the bidirectional cycle between online economic activities and income is the most stable, and online social activities emerge as an increasingly important type of Internet usage. In contrast, the effect of online work diminishes over time, and the feedback loop between entertainment activities and income weakens over time. The findings indicate that digital inequality is a context-dependent and cyclically reproduced process, revealing how heterogeneous Internet use patterns shape the formation of the third-level digital divide and contribute to its ongoing reproduction and evolution over time.
The persistent disparity in public services between urban and rural areas is rooted, in part, in systemic information asymmetry and the “invisibility” of rural needs. Open Government Data (OGD), as a core institutional arrangement of digital governance, is hypothesized to reshape this informational environment. However, its distributive effects on urban–rural public service provision remain empirically underexplored. This study employs a multi-period difference-in-differences (DID) model to investigate the causal impact of municipal OGD implementation on urban–rural service equalization, utilizing panel data from 280 Chinese cities spanning the period 2011–2022. The findings indicate that OGD significantly narrows urban–rural gaps in public service provision. Mechanism analysis reveals that OGD operates by promoting the equalization of urban–rural fiscal expenditures and guiding the spatial rebalancing of urban–rural populations on both the supply and demand sides. Furthermore, heterogeneity analysis uncovers a “digital compensation” effect, wherein the equalizing impact of OGD is more pronounced in small and medium-sized cities, regions with higher fiscal transparency, and areas with lower initial levels of digitalization. Overall, these results emphasize that OGD not only enhances efficiency but also reduces structural inequalities, thereby fostering inclusive development in the digital era.
The rapid adoption of generative AI in higher education has outpaced institutional policy, creating uncertainty regarding AI cheating and appropriate AI use. Building on the Unified Theory of Acceptance and Use of Technology (UTAUT) framework, this study examines the motivational predictors of AI dependence among college students and the psychological states associated with it: “noAIphobia” (anxiety when AI is unavailable) and AI use stigma. To empirically test the model, we surveyed 393 U.S. college students and analyzed the data using partial least squares structural equation modeling (PLS-SEM). Results indicate that competitive conformity—the fear of falling behind peers—is a strong predictor of AI dependence, alongside efficiency and quality expectancies. AI dependence predicts two psychological states that affect continuance intention in opposing directions: noAIphobia is positively associated with continued use as students seek to avoid AI deprivation anxiety, whereas AI use stigma is negatively associated with continued use due to fear of negative social judgment. Academic integrity concerns further amplify the dependence–stigma relationship. These findings reveal the psychological tensions students experience when using AI and underscore the need for clear institutional policies to mitigate psychological distress and promote healthy AI engagement.
The widespread adoption of Large Language Models (LLMs) has introduced a new dynamic into the field of online dating: the use of AI to author or co-author romantic communication. This study investigates the lived experiences of this phenomenon through in-depth, semi-structured interviews with two distinct groups of dating app users (N = 45): those who employ LLMs in their romantic communications and those who have received AI-assisted messages. Drawing on foundational theories of impression management and mediated authenticity, our thematic analysis reveals three major themes: (1) an authenticity paradox, detailing how users justify LLM use as a tool for expressing their "true" internal self; (2) a digital betrayal, exploring recipients' feelings of deception and the erosion of trust upon discovering AI-mediation; and (3) a personato-person leap, examining the shared anxiety of transitioning from an AI-assisted online persona to an unassisted offline encounter. Based on these findings, we develop a theory of the Cyrano Effect that explains the process and consequences of AI-mediated romantic communication. We contribute to the literature on computer-mediated communication by providing insights into how algorithmic tools are reshaping norms of authenticity, impression management, and trust in the pursuit of digital intimacy. We discuss the theoretical and practical implications for platform design and the future of technologically-mediated relationships.
Problematic social media use (PSMU) is increasingly recognized as a heterogeneous construct. The pathway model offers a robust theoretical framework for understanding this heterogeneity; however, this model has not yet been validated using diverse behavioral indicators of PSMU, such as reassurance seeking, cyberbullying, cyberstalking, sexting, and privacy disregard. Moreover, the underlying personality (i.e., impulsivity, compulsivity) and motivational dimensions (i.e., reward sensitivity, punishment sensitivity) associated with these distinct forms of PSMU remain underexplored. To address these gaps, we collected data from 351 university students (Mage = 20.55, SD = 1.53). A latent profile analysis revealed three distinct user profiles: (1) Normative Users (64%) with low engagement across all indicators; (2) Antisocial/Risky Users (14%) characterized by high levels of cyberbullying and sexting; and (3) Relationship-Reassurance Seekers (22%) marked by elevated reassurance seeking, cyberstalking, and privacy disregard. Multinomial logistic regression showed that reward sensitivity significantly predicted membership in both risk-prone profiles. Compulsivity also emerged as a significant predictor of the Relationship-Reassurance profile, while impulsivity was not a significant predictor in any comparison. These findings support the heterogeneous nature of PSMU and align with core assumptions of the pathway model. Importantly, they highlight the need for differentiated, profile-specific interventions tailored to users' motivational and psychological vulnerabilities.
This research estimates the socioeconomic effects of the fiber-to-the-home (FTTH) program conducted by Uruguayan state-owned operator ANTEL. This company, that enjoys the monopoly of FTTH fixed broadband services, conducted an aggressive program of deployments across all the country since year 2011, until reaching 100% coverage in 2023 (with respect to those services supported by copper). These massive investments were motivated by the government universalization program to modernize the country’s infrastructures, rather than to seek an economic return. Callaway and Sant’Anna (2021) differences-in-differences estimators are applied to conduct the empirical analysis. Results indicate that the policy has been successful in generating an increase in average individual incomes. However, the evidence presented here also suggests mixed effects in inequality levels, possibly because deployments started first in the capital city of each department, then favoring those already well-positioned within each region. On the other hand, the effects on income levels have been heterogeneous, being able to extract more economic gains those regions with more educated people, those with enhanced digital environment, and those located closer to the capital city.
As social companion AI become increasingly prevalent, concerns have emerged regarding users' emotional dependence on AI. Prior research has often treated such dependence as a unidimensional construct, offering limited insight into the psychological processes through which it develops. Drawing on Attachment Theory, this study proposes a moderated mediation model to explain how emotionally salient AI interactions give rise to adaptive versus maladaptive emotional dependence. Using data from two survey-based studies of Chinese users of AI companions (Study 1: n = 381; Study 2: n = 425) and PLS-SEM, the results show that mind perception of AI functions as a psychological trigger that activates post-interaction rumination. Reflection-oriented rumination mediates the relationship between mind perception and adaptive emotional dependence, whereas brooding-oriented rumination mediates the relationship between mind perception and maladaptive emotional dependence. Moreover, metacognitive awareness moderates the second-stage effects of rumination, amplifying the adaptive pathway while attenuating the maladaptive one. These findings advance research on human-AI emotional interaction by reconceptualizing mind perception as a dynamic trigger, reframing emotional dependence on AI as a cognitively differentiated phenomenon, and identifying metacognitive awareness as a key boundary condition shaping AI-related emotional outcomes.
Digital transformation has become a defining pillar of modern public administration, with governments increasingly adopting e-government initiatives to enhance service delivery and citizen engagement. However, the improvement in administrative transparency these initiatives achieve remains insufficiently understood. This study addresses this gap by developing an integrated framework to assess how e-government initiatives enhance administrative transparency, quantitatively linking digital adoption, service efficiency and information accessibility to reductions in corruption and improvements in citizen trust, thereby providing evidence-based guidance for designing accountable public management systems. Data were collected from 320 citizens using structured surveys and analyzed descriptively to summarize key perceptions of digital service usage. Pearson correlation analysis was then employed to examine the strength and direction of relationships between e-government adoption. Similarly, regression modelling was applied to assess the overall predictive influence of digital public management reform factors on transparency and citizen trust in government processes. Findings show e government initiatives strongly boost transparency and trust, with r = 0.824 and adjusted R2 = 0.76. These results highlight that information accessibility (beta = 0.454) and digital adoption (beta = 0.443) are the most influential drivers of transparency, offering actionable insights for governments seeking to design more accountable and citizen-centric digital ecosystems.
This study examines how educational attainment predicts the digital divide among middle-aged and older adults (50 + ), and how national-level factors moderate this relationship. Drawing on the Technology Acceptance Model (TAM) and Diffusion of Innovations (DOI) theory, we analyzed data from 25,360 individuals across 55 societies in the seventh wave of the World Values Survey. Results reveal a persistent global educational gradient: individuals with higher education are more likely to use the internet and social media. However, national infrastructure plays a critical moderating role. While GDP per capita and fixed broadband penetration are positively associated with internet use, only GDP per capita is positively associated with social media use. Furthermore, our findings indicate that educational disparities in social media use narrow as national fixed broadband penetration rates increase. This suggests that widely available infrastructure can compensate for individual socioeconomic disadvantages. These findings are robust across different country income groups. Our results indicate that bridging the digital divide in later life requires both educational and economic development approaches.
Google's AI Overviews (AIOs) - AI-generated summaries - mark a shift from information retrieval to AI-driven curation, raising underexplored questions about political and news exposure and algorithmic gatekeeping. This study presents the first systematic mapping of AIOs in these domains, examining their presence on desktop (RQ1a) and mobile (RQ1b) and their relationship to query characteristics (RQ2a, RQ2b). Using interface and algorithm auditing, I develop taxonomies of AIO elements, apply hierarchical cluster analysis to identify structural types, and run regression models to predict these from query characteristics. Findings show AIOs are neither uniform nor neutral: political issue queries tend to trigger comprehensive AIOs, while news politics queries elicit more restrained responses. The proposed taxonomies offer a transferable framework for cross-platform comparisons of AI-powered search and a foundation for future auditing and accountability research amid a major transformation in the search landscape.
This research focuses on the importance of content style design in YouTube channels and identifies the mechanisms through which it affects followers’ utilitarian and hedonic attitudes towards an influencer’s personal channel and their contributions to online influencer branding. This study conducted a cross-sectional, questionnaire-based survey, collecting data from 356 YouTube users. The proposed research model and hypotheses were assessed using partial least squares structural equation modeling (PLS-SEM). This study suggests that unconventional insights and content inspiration in YouTube channel style design are positively associated with followers’ utilitarian and hedonic attitudes, enhancing online influencer branding, including online consumption and contribution. In addition, content relevance only influences followers’ utilitarian attitudes, while narrative design solely affects hedonic attitudes. However, the proposed relationship between title attractiveness and followers’ attitudes towards an influencer’s personal brand is not supported. This research presents new insights into the importance of content style design in YouTube channels and sheds light on the underlying mechanisms through which content style design impacts followers’ contributions to online influencer branding.
