As generative AI becomes embedded in everyday life, public experience with AI is increasingly shaped by consumer-facing tools such as chatbots and AI copilots. Under what conditions do people accept AI-driven labor displacement? Anchored in TAM2/TAM3, this study develops a bounded model of public acceptance in a high-externality context, where evaluations extend beyond private utility to include the legitimacy of substituting AI for human workers and its perceived societal consequences. Using survey data from 7,204 respondents across 19 countries, we test a moderated mediation model in which GenAI experience is associated with acceptance primarily through trust in AI and perceived social benefit. National AI development modestly strengthened, and collectivism modestly weakened, the translation of GenAI experience into these beliefs, consistent with the view that macro context conditions the belief-construction stage, though these moderation effects were small. The study offers a contextually bounded refinement of TAM2/TAM3 for socially consequential domains where technological deployment redistributes risks and benefits beyond the individual user.
Commercial mental health chatbots increasingly incorporate large language model (LLM) conversational capabilities, but their implications for user experience and evaluation remain unclear. Treating ChatGPT’s public release (November 30, 2022) as an ecosystem-level shock that increased LLM visibility, we analyzed 75,953 Apple App Store and Google Play reviews (2018–2025) from six chatbots. Four apps (Elomia, Youper, Yana, and Sintelly) comprised a documented LLM-integration pathway, whereas two apps (Woebot and Wysa before its generative rollout) served as a non-generative comparison pathway within pre-specified windows. The weekly panel was analyzed using a step-plus-ramp design (52-week ramp), comparative interrupted time series models with Newey–West standard errors, and difference-in-differences models with unit and platform-by-week fixed effects and Driscoll–Kraay standard errors. Macro-level search and news indicators showed sustained increases in attention after the release. In the review data, the difference-in-differences model showed an immediate relative decline in the Latent Valence Index for apps on the LLM-integration pathway, while both model families indicated gradual recovery. Segmented post-period modeling located a delayed turning point approximately 80 weeks post-release. Text analyses indicated a concurrent shift in evaluative framing: discussion of session flow and access frictions declined, whereas anthropomorphic appeal and companionship increased. Expectation-gap language narrowed, social-orientation language increased, and emotional-burden language rose modestly. Together, these patterns are consistent with evaluative re-anchoring rather than a return to prior baselines, underscoring the importance of longitudinal post-market monitoring and governance attentive to relational and technical safety.
Drawing on data from Chinese A-share listed companies spanning 2017 to 2023, we treat the institution of National New Generation Artificial Intelligence Innovation and Development Pilot Zones as a quasi-natural experiment. Using a multi-period difference-in-differences approach, we investigate how AI innovation policy can influence corporate digital transformation. The empirical results demonstrate that these policies significantly promote corporate digital transformation, a finding that remains robust after accounting for endogeneity, propensity score matching, and concurrent policies’ confounding effects. Mechanism analysis reveals that AI innovation policies drive this transformation by augmenting human capital, lowering transition costs, and stimulating R&D investment. Heterogeneity analysis further indicates that the promotional effect is more pronounced among non-state-owned corporations and technology-intensive corporations. Moreover, the impact is significantly stronger in regions characterised by high government prioritisation of the digital economy and robust intellectual property protection (IPP), compared to those with low government focus or weak legal frameworks. Finally, economic consequence analysis reveals that institution-driven digital transformation ultimately enhances corporate total factor productivity. By analysing the connection between AI pilot zones and corporate digital transformation, we highlight the importance of institutional support in driving industrial transformation. We also provide an empirical road map for emerging economies to achieve high-quality economy expansion through advances in technology.
This study examines the exports of both small and medium enterprises (SMEs) within the context of their adaptation to artificial intelligence (AI) technologies and robotics by employing a highly original, analytical framework. Theoretically, the study is grounded in the dynamic capabilities approach, which highlights how SMEs enhance efficiency and adapt to uncertain and disruptive environments. Empirically, the analysis draws on a sample of 15,134 SMEs operating in EU countries. Binary choice models are employed to explore the effects of AI and robotics on SME exports across the EU and the results indicate that the adoption of AI and robotics by EU-based SMEs positively influences their exports. To address any potential export self-selection biases, Heckman selection models were employed in the robustness checks, incorporating propensity score matching and the interaction between AI and robotics. The main findings remain robust. Furthermore, the study identifies the heterogeneous effects of AI and robotics on SME exports. AI adoption significantly increases exports across all destinations, with the strongest export growth observed in Latin America and the Caribbean, while robotics adoption exhibits the largest export impact in China. These results underscore the effectiveness of AI and robotics for SMEs within the EU context. Building on these findings, this study empirically contributes to the literature by examining AI and robotics through the lens of the dynamic capabilities approach, particularly for highly fragile firms, such as SMEs. This dimension is largely overlooked in both the AI literature and the relatively scarce research that is focused on robotics in this context.
The use of artificial intelligence (AI) is rapidly expanding across various professional domains, including academia. Yet, limited empirical evidence exists on which academics are adopting AI, how it influences their research productivity, and what implications it holds for their skillsets. Contributing to ongoing debates on the impact of AI on scientific work, this study draws on original data from a recent large-scale survey of Italian academics to explore patterns of AI use, perceived effects on research output, and skill development. Our findings reveal that 39% of respondents report using AI, with significant variation depending on the type of research activity and stage of the research process. AI adoption is more prevalent among younger scholars and those engaged in applied or interdisciplinary research. Overall, AI use is positively associated with higher research productivity and with more favourable expectations about the impact of AI on researchers’ skills. However, the analysis also uncovers nuanced attitudes towards AI: trust in these technologies is not universal, and pockets of scepticism persist across segments of the academic population.
Global concerns about possible adverse effects of social media on adolescents’ emotional well-being are prevalent, but empirical evidence remains inconsistent. This study examines how social media use is associated with adolescents’ self-reported happiness and loneliness by drawing on data from the 2019 Teenage Media Use Survey conducted by the Korea Press Foundation. A random forest model is applied to an analytic sample of 1,573 social media users to estimate the relative predictive importance of personal emotional characteristics and media-use factors. The findings show that the relationship between social media use and adolescent self-reported emotional well-being is multifaceted, involving social media use (SMU) characteristics, usual emotional states, and emotions experienced during SMU. Our analysis indicates that the most important predictors were perceived SNS communication quality and SNS-related feelings of loneliness or happiness. Notably, perceived SNS communication quality exhibited a less straightforward pattern: although SNS-related happiness and loneliness were closely aligned with corresponding emotional classifications, higher perceived communication quality was associated with a greater likelihood of low-happiness and high-loneliness classifications. The relative predictive importance and direction of these predictors also differed across adolescents’ usual emotional state groups. Overall, the findings suggest that SNS experiences are not uniformly associated with adolescent well-being, but may carry different meanings depending on their usual emotional states. The findings emphasize the need to consider emotional context in youth digital well-being research and social media policy.
Change-oriented organisational citizenship behaviour (OCB) is crucial for public sector adaptability. However, studies examining the impact of hindrance techno-stressors induced by digital transformation on such a behaviour are limited. Integrating the Conservation of Resources (COR) theory and Social Information Processing (SIP) theory, this study investigates how these stressors (techno-invasion, techno-overload and techno-complexity) jointly influence change-oriented OCB. Data were collected from 502 Chinese frontline civil servants through a survey and analysed using hierarchical regression analysis and instantaneous indirect-effect analysis. Results reveal that techno-invasion and techno-overload exhibit U-shaped relationships with change-oriented OCB. Furthermore, techno-invasion positively predicts techno-overload, thereby influencing change-oriented OCB. Notably, techno-complexity does not moderate the association between techno-invasion and techno-overload, thus failing to condition the indirect effect on change-oriented OCB. This study advances change-oriented OCB research by challenging the linear assumption that hindrance techno-stressors are inherently detrimental. By introducing a multi-stressor interplay perspective, it demonstrates that techno-stressors sequentially cascade rather than operate parallelly. Moreover, it extends the COR theory by integrating the SIP theory, evincing how contextual cues legitimise risky resource acquisition strategies under severe resource loss, and offering practical insights for public managers.
The rapid diffusion of digital technologies has reshaped public life by transforming patterns of interaction, co-presence, and social behaviour in urban environments. This study does not address these transformations in isolation; instead, it examines how spatial design professionals evaluate the behavioural consequences of digital technologies in relation to their conceptualisations of the functions of public space. Professional interpretation forms the core analytical focus of the study. More specifically, the analysis focuses on the relationship between experts' disciplinary orientations toward public space and their professional assessments of key behavioural dimensions in digitally mediated public life, including social isolation, media and digital effects, homogenisation/placelessness, social anxiety, and digital addiction. To this end, a quantitative modelling approach based on expert assessments was adopted. Parametric statistical methods, including Pearson correlation, linear regression, and multiple linear regression analyses, were used to examine the relationship between professionals’ evaluations of public space functions and their professional assessments of ICT-related behavioural impacts. The findings indicate that the importance experts assign to public space functions is a statistically significant but modest predictor of how they professionally assess the behavioural consequences of digital technologies in urban life. These associations appear across several dimensions, including social isolation, placelessness, digital addiction, social anxiety, and assessed media and digital effects. However, while orientations toward public space functions are associated with these assessments, the purposes of ICT use show limited predictive influence, with digital addiction emerging as the only dimension associated with specific technology-use purposes. Overall, professionals' assessments of digitalisation appear to be more closely related to how they understand the functions of public space than to their own purposes for using ICT. Digitalisation can therefore be understood as a socio-technical transformation whose implications for urban space and public life are interpreted through professional values and disciplinary understandings.
The rapid proliferation of digital technologies in academic environments has transformed how students' access, manage, and store academic information. While these technologies enhance learning and knowledge acquisition, they also encourage the excessive accumulation of digital academic materials. Academic Digital Information Hoarding (ADIH) has emerged as a salient behaviour among Gen Z students, driven by cognitive and psychological dynamics. Grounded in attachment theory, this study examines how cognitive and affective factors interact within competitive higher education contexts to shape tendencies toward ADIH. Focusing on Saudi higher education, the study investigates the effects of Information Overload (IO), Fear of Missing Out (FoMO), Information Literacy (IL), and Epistemic Egoism (EE) on ADIH, with Perceived Academic Competition (PAC) modelled as a mediator. Data were collected from 504 undergraduate and postgraduate students across 28 Saudi universities during the 2024/2025 academic year through an online survey and analysed using PLS-SEM with SmartPLS software. Results indicate that EE (β = 0.108), IO (β = 0.171), FoMO (β = 0.561), and PAC (β = 0.249) significantly predict ADIH, whereas IL showed no significant direct effect (β = 0.014). PAC significantly mediates the relationships between EE, IO, FoMO, and ADIH, suggesting that competitive academic environments may intensify security-seeking responses to academic uncertainty and reinforce digital retention as a compensatory strategy. The findings extend current research on academic digital hoarding by showing how cognitive and affective factors operate within competitive academic settings. They also suggest that universities should address digital retention alongside information literacy and academic competition.
This study provides a multilevel synthesis of knowledge hiding (KH) and workplace digital technologies (DTs) literature. While these streams have been examined separately, research at the intersection of KH and workplace DTs remains fragmented. Prior research shows that KH not only occurs at the individual level, but also at the team and organizational levels. Across these same levels, DTs reshape how work is coordinated, how communication unfolds, and how autonomy, interdependence, and control are structured. Together, these developments make a multilevel understanding of KH essential for understanding knowledge dynamics in digital workplaces. To address this need, we conducted a systematic literature review and examined 39 empirical studies. The review identifies key themes, theoretical perspectives, and research gaps across employee, team, and organizational levels. Our findings reveal a strong focus on individual-level antecedents of KH, with limited attention to the team and organizational level. Although DTs are typically implemented to facilitate knowledge sharing, our synthesis shows that they can unintentionally exacerbate KH by creating relational and cognitive strains that unfold across multiple levels. Building on these insights, we propose a multilevel framework to guide future research. This study advances understanding of knowledge dynamics in digital workplaces by highlighting both the enabling and constraining effects of DTs on knowledge exchange. In doing so, it contributes to broader debates on the organizational consequences of digitalization and provides actionable insights for managing knowledge behaviors across organizational levels.
This study examines how overeducation, digital literacy, and demographic heterogeneity jointly shape internal migration in China. Using nationally representative CFPS data and a multi-method empirical strategy that includes logit and probit models, propensity score matching, and double machine learning, the analysis reveals that overeducation is positively related to the likelihood of migration, and that this effect intensifies under higher digital literacy. Age and gender moderate this relationship, with older adults and women exhibiting greater sensitivity to digital literacy when overeducated. Mediation analyses show that income and job satisfaction partially transmit the effect of the overeducation-digital literacy interaction on migration. These findings demonstrate that digital literacy has become an important stratifying mechanism shaping labor market reallocation and mobility. The policy implications highlight the need for digital inclusion strategies, tertiary education reform, and the creation of new employment opportunities to foster more equitable migration outcomes.
Although prior research has identified artificial intelligence (AI) as a key driver of supply chain innovation (SCI), further investigation is required into whether and how AI influences SCI, and the contingencies that may affect such an impact. Integrating socio-technical systems theory (STST) and contingency theory (CT), we develop a conceptual model that links AI to SCI through the mediating role of supply chain ambidexterity (SCA) and incorporates the moderating effects of supply chain complexity (SCC). Drawing on survey data from 245 Chinese manufacturing firms, we first ensure measurement quality through tests of common method bias, non-response bias, reliability, and validity. We then proceed to test our research hypotheses using structural equation modeling and hierarchical regression analysis. The results show that AI enhances SCI both directly and indirectly through the mediating role of the two dimensions of SCA, namely SC exploration and exploitation. Additionally, the findings indicate that SCC has a positive moderating effect on the relationships between AI and SCI and between AI and SC exploration/exploitation. Besides deepening our theoretical understanding of the drivers of SCI, and specifically of the processes of AI-driven SCI, the study offers practical insights for advancing innovation in the SC.
AI-induced job insecurity has become a salient work stressor amid rapid AI adoption in the workplace. While existing research has predominantly focused on its quantitative dimension, which reflects employees' concerns about potential job loss by AI, AI-induced qualitative job insecurity, which refers to employees' worries about potential deterioration of important job features due to AI, remains underexplored. Drawing on the cognitive appraisal theory of stress, this study investigates how AI-induced qualitative job insecurity influences knowledge workers' career commitment via their threat and challenge appraisals. We further examine the moderating roles of social support and learning adaptability. Data from 395 knowledge workers in the language service industry, a sector at the forefront of AI disruption, are primarily analyzed using partial least squares structural equation modeling (PLS-SEM). The results show that AI-induced qualitative job insecurity indirectly reduces career commitment by simultaneously activating threat appraisal and suppressing challenge appraisal, tentatively positioning it as a threat stressor. Moreover, social support attenuates the activating effect of AI-induced qualitative job insecurity on threat appraisal, whereas learning adaptability weakens its suppressive impact on challenge appraisal. Our research investigates an understudied construct of AI-induced qualitative job insecurity and expands relevant scholarly scopes to career-level outcomes, offering implications for human-centered AI adoption in the workplace and sustainable professional development.
