A structural investigation of consumer interactions with artificial intelligence curation systems reveals that distinct generational cohorts rely on divergent cognitive mechanisms when assessing algorithmic fashion platforms. The research, published in Fashion and Textiles by Dayun Jeong and Young Sam Kim, examined how users process automated recommendations that assemble coordinated outfits and personalized discovery paths. By integrating the Task–Technology Fit framework with the Technology Acceptance Model, the authors demonstrated that while Generation Z consumers evaluate artificial intelligence primarily through feature alignment and functional utility, Generation X users depend heavily on operational simplicity to judge practical value. Despite these pronounced variations during early belief formation, structural modeling established that the downstream psychological pathways directing final behavioral intention operate in a statistically uniform manner across age cohorts once satisfaction and usefulness are achieved.
Artificial intelligence curation technology has emerged as a distinct development within digital apparel retail, moving beyond conventional recommender engines that merely calculate statistical similarities between individual items. As detailed in the investigation, modern curation platforms integrate multidimensional behavioral signals, such as consumer click patterns and dwell times, with detailed garment attributes like fabric composition and silhouette structure. Advanced systems combine visual catalog data and conversational interfaces to generate dynamically sequenced wardrobes, capsule collections, and styled outfit combinations. These multimodal architectures perform domain-specific visual reasoning and assess stylistic compatibility across separate garments to resolve complex consumer dressing dilemmas.
To rigorously examine consumer responses to these sophisticated algorithmic environments, the authors constructed a comprehensive theoretical architecture linking Task–Technology Fit with the Technology Acceptance Model. While the classic acceptance paradigm posits that perceived usefulness and perceived ease of use govern behavioral intention, it leaves the formative origins of those cognitive beliefs largely unexplained. Integrating the fit framework resolves this theoretical gap by establishing how alignment between technological capabilities and specific user tasks serves as a direct precursor to cognitive utility appraisals. In fashion contexts, where users evaluate complex aesthetic combinations across visual interfaces, system misalignment risks impairing operational fluency. The hybrid conceptual framework addresses whether functional alignment stimulates perceived usefulness and ease of use, and whether those beliefs subsequently influence consumer satisfaction and continuous behavioral intention across varying generational groups.
The study formulated nine foundational hypotheses to verify these structural associations and explore cohort-level moderation across Generation Z, Millennials, and Generation X. Hypotheses 1 and 2 predicted that high task–technology fit would exert positive effects on perceived usefulness and perceived ease of use. Hypothesis 3 proposed that ease of use positively influences perceived usefulness, reflecting the cognitive principle that intuitive interfaces allow consumers to realize practical utility more readily. Subsequent hypotheses stated that perceived usefulness positively influences consumer satisfaction and behavioral intention, while perceived ease of use independently predicts satisfaction and intentional persistence. Hypothesis 8 proposed that overall satisfaction functions as a primary positive driver of behavioral intention. Finally, Hypothesis 9 posited that generational membership acts as a structural moderator, producing significant differences in path coefficients across demographic cohorts.
Empirical testing of this conceptual model was conducted using a quantitative survey administered to digital fashion consumers in South Korea. The sampling design focused on the South Korean market because of its advanced digital commerce infrastructure, mature mobile shopping culture, and rapid adoption of artificial intelligence interfaces. Data collection was managed through a specialized online consumer panel agency, implementing strict screening procedures to ensure that all participants possessed verified prior experience using digital fashion platforms offering automated recommendations. The finalized analytic sample comprised 300 eligible respondents, representing a diverse demographic distribution. Female respondents accounted for 74.7 percent of the sample and male participants represented 25.3 percent. When classified across age brackets, 28.6 percent were in their fifties, 25.0 percent in their thirties, 23.7 percent in their forties, and 22.7 percent in their twenties, yielding a cohort distribution of 45.7 percent Millennials, 36.0 percent Generation X, and 18.3 percent Generation Z.
Measurement instrumentation adapted established items from foundational literature, utilizing seven-point Likert scales ranging from strongly disagree to strongly agree. Survey indicators capturing task–technology fit were adapted from Goodhue and Thompson, while items assessing perceived usefulness, perceived ease of use, satisfaction, and behavioral intention were operationalized from Davis and extended retail adoption scholarship. To verify measurement integrity, the authors conducted preliminary descriptive and distributional assessments, observing normal univariate distributions across all five latent variables, with skewness ranging between negative 0.425 and negative 0.263 and kurtosis spanning negative 0.012 to positive 0.372. Confirmatory factor analysis demonstrated that all standardized factor loadings exceeded 0.50 with statistical significance. Construct reliability exceeded the 0.70 benchmark for every construct, and average variance extracted values surpassed the 0.50 threshold, establishing robust convergent validity and indicator reliability.
Evaluation of the pooled structural equation model demonstrated exceptional fit to the empirical data across multiple contemporary indices. Key goodness-of-fit statistics reached high levels, including a comparative fit index of 0.981, a Tucker–Lewis index of 0.977, a goodness-of-fit index of 0.918, an adjusted goodness-of-fit index of 0.893, and a normed fit index of 0.948. Residual evaluations confirmed minimal approximation discrepancy, with a root mean square residual of 0.036 and a root mean square error of approximation of 0.043. The correlation matrix documented significant positive associations among all latent variables at the p < .01 level, led by a strong correlation of 0.803 between consumer satisfaction and behavioral intention.
Direct path analysis supported seven of the eight primary structural hypotheses within the baseline model. Task–technology fit exerted a substantial positive influence on perceived usefulness with a path coefficient of 0.656 (p < .01) and on perceived ease of use at 0.584 (p < .01), validating Hypotheses 1 and 2. Perceived ease of use positively influenced perceived usefulness at 0.337 (p < .01), supporting Hypothesis 3. Perceived usefulness demonstrated strong structural effects on consumer satisfaction (beta = 0.586, p < .01) and behavioral intention (beta = 0.204, p < .05), upholding Hypotheses 4 and 5. Perceived ease of use significantly predicted satisfaction with a coefficient of 0.341 (p < .01), confirming Hypothesis 6. Crucially, the direct structural path linking perceived ease of use to behavioral intention failed to achieve statistical significance (beta = 0.072, p = 0.254), leading to the rejection of Hypothesis 7. In contrast, overall satisfaction exhibited the strongest predictive influence on continuous behavioral intention, generating a coefficient of 0.684 (p < .01) in support of Hypothesis 8.
Before evaluating generational moderation through multi-group structural modeling, the researchers conducted hierarchical measurement invariance tests to establish cross-cohort equivalence. The configural invariance baseline model exhibited satisfactory fit with a chi-square of 984.315, a comparative fit index of 0.933, a Tucker–Lewis index of 0.922, and an error of approximation of 0.047. Constraining factor loadings across Generation Z, Millennials, and Generation X in the metric invariance model produced a non-significant chi-square difference of 25.725 (p = 0.845), proving that latent constructs were measured identically across cohorts. Although subsequent scalar constraints yielded a significant chi-square delta of 66.182 (p < .001), goodness-of-fit deterioration remained negligible, confirming partial scalar invariance and justifying formal pairwise comparisons using constrained-model chi-square difference tests with one degree of freedom.
The multi-group structural comparisons revealed pronounced cohort asymmetries across early belief and evaluation pathways, providing partial acceptance for Hypothesis 9. Pairwise analysis indicated that the path from task–technology fit to perceived usefulness was significantly weaker for Generation X than for both Generation Z (p = 0.003) and Millennials (p = 0.019). Conversely, the influence of perceived ease of use on perceived usefulness proved significantly stronger in Generation X than in Generation Z (p = 0.002). In affective evaluation, perceived usefulness drove satisfaction far more powerfully among Generation Z than among Millennials (p = 0.041) and Generation X (p = 0.012). Conversely, the structural link between perceived ease of use and satisfaction was significantly stronger in Millennials (p = 0.038) and Generation X (p = 0.007) than in Generation Z. Regarding direct intention paths, perceived ease of use significantly influenced behavioral intention solely among Millennials, showing statistically significant pairwise elevation over Generation Z (p = 0.022) and Generation X (p = 0.019).
A critical theoretical finding from the multi-group analysis is the complete convergence of downstream behavioral intention mechanisms across all three generational cohorts. While upstream cognitive appraisals of usefulness and satisfaction are shaped along distinct cohort-dependent pathways, the structural links connecting perceived usefulness to behavioral intention and consumer satisfaction to behavioral intention exhibited no statistically significant pairwise differences between Generation Z, Millennials, and Generation X. These empirical outcomes suggest that once fundamental beliefs regarding system utility and emotional satisfaction are established, users across all age brackets convert those evaluations into future platform engagement at comparable rates. The finding cautions against uncritical overgeneralization of demographic stereotypes, demonstrating that generational distinctions reside within initial experiential appraisals rather than terminal adoption decisions.
The authors outline several practical and methodological implications alongside key research constraints. For retail platform architects, the empirical findings support a two-tier operational strategy: user interface teams should deploy cohort-tailored onboarding and experiential features to stimulate upstream belief formation, while standardizing downstream value presentation and satisfaction recovery protocols. Limitations of the investigation include its localized sample of 300 South Korean consumers, the reliance on cross-sectional self-reported survey data, and potential selection biases associated with voluntary panel recruitment. Future studies must incorporate longitudinal tracking, expand cross-cultural demographic sampling, and examine emerging governance considerations such as algorithmic transparency, data ethics, and consumer trust to advance theoretical comprehension of artificial intelligence adoption across evolving digital retail environments.
Subject of Research: Generational differences and technology acceptance mechanisms in artificial intelligence fashion curation platforms
Article Title: Generational differences in AI adoption among fashion curation platform users
Article References: Jeong, D., & Kim, Y. S. (2025). Generational differences in AI adoption among fashion curation platform users. Fashion and Textiles, 12(1), Article 41. https://doi.org/10.1186/s40691-025-00448-5
Image Credits: AI Generated
DOI: 10.1186/s40691-025-00448-5
Keywords: Artificial Intelligence, Fashion Curation, Technology Acceptance Model, Task-Technology Fit, Consumer Behavior, Generational Cohorts, Structural Equation Modeling, Digital Retail, Generational, differences, adoption, fashion
News Source: Denise Maddox. (November 10, 2026). Study Identifies Generational Differences in AI Fashion Curation Platform Adoption. Scienmag.
