Aim/PurposeThis study aims to investigate how key characteristics of AI-powered personalized learning platforms influence student learning performance by examining the mediating roles of perceived value and perceived trust, as well as the moderating role of self-efficacy. The research aims to provide a clearer understanding of the psychological mechanisms that drive effective learning outcomes through technological features in the context of Vietnamese higher education.

BackgroundThe rapid advancement of artificial intelligence (AI) has transformed the digital education landscape, particularly with the emergence of AI-driven personalized learning systems. These platforms promise adaptive, learner-centered experiences by leveraging data-driven algorithms to tailor content, feedback, and support to individual needs. However, the success of such technologies is not solely dependent on their technical capabilities; it also hinges on students’ psychological responses, including how they perceive the platform’s value and trustworthiness. Despite growing implementation, a limited understanding remains of how multiple system characteristics, such as intelligence, personalization, anthropomorphism, and information quality, collectively shape these psychological factors and ultimately influence academic performance. Moreover, individual learner traits, such as self-efficacy, may determine how effectively students can translate positive perceptions into learning success. This study addresses these gaps by applying the Stimulus-Organism-Response (S-O-R) framework, integrated with the Information System Success Model (ISSM), to explore the complex interplay between technological features, psychological responses, and learning outcomes in AI-driven education. The research is conducted within the Vietnamese higher education context, offering novel insights from an emerging educational market.

MethodologyThis study employed a quantitative research design using a structured online questionnaire to collect data from university students in Vietnam who had experience with AI-powered personalized learning platforms. A non-probability convenience sampling method was employed, yielding 462 valid responses. The data were analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) with SmartPLS 3.0 to examine relationships among platform characteristics, perceived value, perceived trust, and learning performance, with self-efficacy as a moderating variable.

ContributionThis study advances the theoretical understanding of AI in education by integrating system design characteristics, including intelligence, personalization, anthropomorphism, system quality, and information quality, with psychological constructs such as perceived value and trust, under the S-O-R framework and the ISSM. It also introduces self-efficacy as a key moderating variable. The integration of these frameworks within the context of Vietnamese higher education provides new empirical evidence on how AI-enhanced platforms support learner performance in developing countries, contributing both to global and localized knowledge of technology-enhanced learning.

FindingsThe findings reveal that intelligence, personalization, information quality, and system quality of AI-powered learning platforms significantly enhance both students’ perceived value and trust, whereas anthropomorphic features only boost perceived value but do not directly influence perceived trust. Both perceived value and trust have a positive impact on student learning performance, with perceived value also strengthening perceived trust. Additionally, self-efficacy plays a moderating role, amplifying the effects of perceived value and trust on learning outcomes, suggesting that learners with higher self-efficacy benefit more from these platform features.

Recommendations for PractitionersDevelopers should prioritize enhancing system intelligence, personalization, and information quality to foster student trust and perceived value. Educators and academic institutions should focus on strengthening students’ self-efficacy through digital literacy training and personalized learning support to maximize learning outcomes. These findings provide concrete guidance for technology developers, educators, and policymakers seeking to design and implement effective AI-based learning solutions in higher education environments.

Recommendation for ResearchersResearchers should explore other psychological or contextual moderators, such as learning motivation and cognitive load, and validate the model across diverse educational environments and demographic groups to increase generalizability.

Impact on SocietyBy uncovering the mechanisms that drive effective learning in AI-supported environments, this study provides actionable guidance for creating more equitable, engaging, and high-quality digital education systems. The findings contribute to improving academic success, digital competency, and learner empowerment, thereby supporting the broader goal of technology-enabled inclusive education in developing contexts such as Vietnam.

Future ResearchFuture research could explore longitudinal effects of AI learning tools, incorporate behavioral data, and examine the interplay between affective responses and cognitive evaluations in AI-driven learning dynamics.