A uses and gratifications approach to examining users’ continuance intention towards smart mobile learning

Smart Mobile Learning (SML), integrating artificial intelligence (AI) and mobile learning, offers personalized learning experiences anytime, anywhere. Applications like Liulishuo utilize AI technologies such as natural language processing (NLP), speech recognition, machine learning, and recommendation systems to enhance learning. While SML's popularity is growing, retaining users is crucial. This study addresses a research gap by examining the specific constructs representing gratification towards SML, particularly the intelligent dimension, using the uses and gratifications (U&G) framework. The U&G theory, focusing on users' active media selection based on needs and expectations, is well-suited for investigating post-acceptance behavior and user continuance intention. Unlike technology acceptance models that often neglect emotional aspects, U&G incorporates emotional factors associated with personal and social realms, providing a more comprehensive understanding of sustained technology use. This research aims to identify the various factors composing SML-related gratification and their hierarchy in predicting continued usage; propose new gratifications for the intelligent dimension of SML; and empirically uncover the importance of various gratifications impacting continuance intention towards SML using PLS-SEM.

Existing uses and gratifications (U&G) research on mobile learning lacks specific constructs for categorical gratification levels, and lacks research on the intelligent dimension of SML. Table 1 presents a typology of uses and gratifications across various media, showing a range of gratifications identified in previous research, including utilitarian, hedonic, social, and content-related gratifications. However, these studies don't specifically address the unique aspects of SML's AI-driven features. The current study builds upon this existing research by focusing on the specific gratification constructs within the context of SML, particularly addressing the role of intelligence and proposing a more refined understanding of the gratifications associated with AI-powered learning.

This quantitative study uses survey data from 495 Liulishuo users in China. The measurement scales for constructs like perceived intelligence, convenience, perceived enjoyment, concentration, status, achievement, education, and continuance intention were adapted from previous literature and tailored to the SML context, using a seven-point Likert scale. A rigorous translation process ensured linguistic and conceptual equivalence of the survey items. A pilot test was conducted before the main survey. Conditional random sampling was used to target Liulishuo users, resulting in a sample predominantly comprising young college students and working professionals. Data were analyzed using partial least squares structural equation modeling (PLS-SEM) via SmartPLS 3.0. PLS-SEM was chosen due to its suitability for complex models with numerous variables and its robustness compared to covariance-based structural equation modeling (CB-SEM). The measurement model's reliability and validity were assessed using composite reliability (CR), Cronbach's alpha, and average variance extracted (AVE). Discriminant validity was evaluated using Fornell and Larcker's criteria. The structural model's goodness of fit was assessed using the GoF measure.

The study found that all hypotheses were supported except for H4 (concentration's effect on continuance intention). Descriptive statistics showed that the sample (51.9% female, 95.8% with college degrees or higher, 98% aged 19-45) aligned with Liulishuo's target audience. The measurement model demonstrated good reliability and validity (CR > 0.800, Cronbach's alpha > 0.700, AVE > 0.500) and discriminant validity. The structural model exhibited a good fit (GoF = 0.64, explaining 63.1% of the variance in continuance intention). Intelligence (β = 0.171, p < 0.01) was the most significant predictor of continuance intention, followed by education (β = 0.150, p < 0.05), status (β = 0.147, p < 0.05), achievement (β = 0.131, p < 0.05), convenience (β = 0.125, p < 0.05), and perceived enjoyment (β = 0.120, p < 0.05). Concentration did not significantly influence continuance intention.

The findings support the importance of technology gratification (intelligence and convenience) and content gratification (education) in predicting users' continuance intention toward SML. The significant role of intelligence aligns with previous research on AI-based systems and confirms the value users place on intelligent learning guidance and personalized feedback. The importance of education emphasizes the need for high-quality learning materials. The significant effects of social (status), utilitarian (achievement), and hedonic (perceived enjoyment) gratifications highlight the multifaceted nature of user engagement. The lack of a significant effect of concentration might be due to the specific characteristics of Liulishuo or the nature of mobile learning itself. The study's findings provide insights into the relative importance of various user gratifications, which can inform the design and development of future SML applications. The results showcase the applicability of U&G theory in understanding post-acceptance behavior in the context of SML.

This study significantly contributes to the understanding of users' continuance intention toward SML by identifying specific gratifications and their relative importance. It extends U&G theory to the SML domain and introduces "intelligence" as a novel factor of technology gratification. The findings have practical implications for SML developers, suggesting a focus on enhancing AI features, providing high-quality educational content, and incorporating social and motivational elements to increase user engagement. Future research could investigate the generalizability of these findings across different cultural contexts and SML platforms, explore the moderating roles of individual differences, and examine the long-term impact of these gratifications on learning outcomes.

The study's limitations include its geographical scope (China), which may introduce cultural bias. The sample, while representative of Liulishuo's user base, might not be fully generalizable to all SML users globally. Future studies could address these limitations by using samples from diverse cultural backgrounds and comparing results across different SML platforms. Further research could also investigate the potential influence of economic factors on user gratifications and continuance intention.