Purpose This study examines how users' everyday experiences with digital wallet apps, as a major category of FinTech applications, shape their well-being rather than focusing only on continued use. Using real user reviews from Saudi Arabia, it identifies four key experiential factors: complaint handling effectiveness, perceived system reliability, information ambiguity, and user autonomy. Drawing on Self-Determination Theory (SDT), the study explains how these factors shape user well-being. Design/methodology/approach We applied a sequential mixed-methods design guided by Computational User Feedback Analytics (CUFA) to Arabic and English reviews collected from the Google Play Store and Apple App Store. CUFA integrates hierarchical clustering, Latent Dirichlet Allocation( topic modeling, sentiment analysis, multi-model classification, human validation, and regularized regression. We interpreted the review-derived constructs through SDT and tested them using Ridge and Lasso regression. Findings Our findings indicate that complaint handling effectiveness, perceived system reliability, and user autonomy positively shape user well-being. In contrast, information ambiguity negatively affects user well-being. Originality/value Unlike prior studies that rely predominantly on structured surveys and predefined adoption models, this study applies CUFA to derive, validate, and test theoretically meaningful constructs using large-scale, unsolicited user feedback. It also extends SDT into the fintech context by showing how specific digital wallet experiences support or frustrate users' psychological needs.
Purpose Cognitive bias in large language models (LLMs) undermines their reliability due to the practical limitations of existing mitigation methods. This study introduces a divergence-reflection-convergence (DRC) framework, which leverages a computationally efficient and inference-time strategy designed to identify and mitigate cognitive biases without costly retraining. Design/methodology/approach This study proposes a novel three-stage DRC framework utilizing a competitive reasoning process where the model first generates divergent viewpoints, subsequently reflects on them to extract a decision rule, and finally converges on a robust output. Furthermore, we validate the DRC framework on five popular LLMs across four common dimensions of cognitive bias, quantifying performance with evaluation metrics including asymmetric kappa (AK) and total variation distance (TVD). Findings The experimental results demonstrate that LLMs refined by the DRC framework provide more consistent and reliable outputs across all scenarios. Notably, this effect is more pronounced for mitigating internal biases than external biases, suggesting the DRC framework is highly effective at correcting LLMs' flawed internal heuristics. In terms of specific evaluation metrics, the DRC framework successfully transforms all negative or near-zero AK scores into positive values and markedly reduces TVD scores, both of which signify a substantial bias reduction. Originality/value This study develops a DRC framework as a novel inference-time mitigation scheme for LLM bias that addresses the practical limitations of retraining-based approaches. Its competitive reasoning process offers a scalable solution for deploying responsible AI. Concurrently, we propose a more nuanced evaluation metric, the AK, tailored for quantifying LLM bias in subjective contexts.
Purpose Quantitative work on gambling-like mechanics (GLMs) in video games has largely treated them as a uniform category, overlooking differences in reward structure and similarity to traditional gambling. This study examines predictors of expenditure across three disaggregated GLM categories – cosmetic loot boxes (CLB), game-affecting loot boxes (GALB) and other in-game gambling-like content (OGLC). Design/methodology/approach Data were collected via an online survey of adult gamers (N = 587). Three model sets used the same gaming engagement and problem behavior indicators as predictors, each comprising a two-part approach: binary logistic regression on the full sample to predict purchase participation, and OLS linear regression among purchasers to predict expenditure amounts. Findings Financial engagement – expenditure on non-random free-to-play content and battle passes – emerged as the dominant predictor class; gaming frequency and gaming disorder were null throughout. Problem gambling severity predicted participation selectively – most strongly for OGLC – but predicted expenditure amounts in no category. Predictor patterns dissociated between participation and amount decisions in ways aggregated or uniform analytic approaches would have obscured. Practical implications Increased focus should be given to paid game mechanics that closely mimic the visual and esthetic presentation of traditional gambling formats. Engagement systems such as battle passes should be carefully assessed for their association with increased expenditure on GLMs. Attention should be paid to the risks related to purchase entry into GLMs, rather than to expenditure amounts alone. Originality/value This study is among the first to apply a disaggregated GLM taxonomy empirically and to model participation and expenditure amount as distinct outcomes.
Purpose E-commerce platforms increasingly deploy generative AI (GAI) to summarize large volumes of customer reviews. This study examines sentiment distribution bias in platform-provided GAI review summaries, defined as distortion in the relative representation of positive, neutral and negative sentiment compared with their proportional presence in the associated review corpus. Design/methodology/approach Grounded in information theory, we characterize four sentiment distribution bias patterns and develop an input-output framework linking source review characteristics and summary design features to sentiment bias. We analyzed data from an online travel platform covering 847 Tokyo hotels with GAI review summaries linked to 111,115 customer reviews to assess sentiment bias and conducted a complementary experiment to examine downstream consumer responses. Findings The platform's GAI review summaries exhibited predominant positive bias: positive sentiment was overrepresented, whereas neutral and negative sentiments were underrepresented relative to the review corpora. Lower source review volume, higher rating entropy and lower positive imbalance were associated with greater sentiment bias, whereas longer summaries, greater attribute granularity and lower attribute sentiment diversity were associated with lower bias. An experimental study showed that, relative to a low-bias summary, a GAI review summary with greater positive bias improved consumers' attitude toward the hotel and increased booking intention, despite being perceived as less helpful. Originality/value This research documents sentiment distribution bias in platform-provided GAI review summaries, identifies input and output conditions under which such bias varies and examines its consequences for consumer responses. The findings extend the online review and GAI literatures while informing platform governance and responsible GAI deployment.
Purpose This study aims to resolve a key paradox in influencer marketing: why brands are increasingly designing virtual influencers with aesthetic imperfections (e.g. freckles), despite conventional wisdom suggesting that perfection is key. It investigates when and why these imperfections enhance consumer engagement, specifically by comparing their effects on virtual versus human influencers. Design/methodology/approach The research employs two main experiments and a supplementary robustness study. Participants were exposed to images of either a virtual or a human influencer, both manipulated to have aesthetic imperfections. The data were analyzed using ANOVA and a moderated mediation model to test the proposed theoretical framework. Findings The two main studies show that aesthetically imperfect virtual influencers elicit higher consumer engagement than aesthetically imperfect human influencers, partly through perceived scarcity. A supplementary study further rules out perceived authenticity and anthropomorphism as alternative mediators and shows that the effect extends to purchase intention, especially under symbolic product appeal framing. Practical implications Brands can enhance virtual influencer engagement by using minor aesthetic imperfections as scarcity cues, especially for consumers seeking uniqueness and symbolic products. Originality/value This study compares consumer responses to aesthetic imperfections in virtual and human influencers. It helps reconcile mixed findings in the literature and advances theory by identifying perceived scarcity as a psychological mechanism that enhances virtual influencer appeal. The study also extends scarcity theory to the digital context and offers practical guidance for designing “perfectly imperfect” virtual personas that engage targeted consumer segments.
