Web-Based and Mobile AI Tools in Southeast Asian Higher Education: A Systematic Review and Thematic Synthesis

Web-based and mobile artificial intelligence tools have spread rapidly in Southeast Asian universities, yet evidence on what they achieved in real learning remained fragmented. This systematic review synthesised empirical studies that implemented or evaluated such tools in degree-level higher education in Southeast Asia between 2021 and 2026. Following the PRISMA 2020 guideline, 642 Scopus records were screened, 79 full texts were assessed and 59 studies were included. Study quality was appraised with the Mixed Methods Appraisal Tool, measured outcomes were summarised by effect direction, and findings were integrated through three-stage thematic synthesis. Most studies came from Indonesia, Vietnam and Thailand, concerned language learning, used general-purpose chatbots and relied on non-randomised designs. Of 45 studies that measured learning, 36 reported improvements, mainly in uncontrolled designs or against routine teaching, whereas four of six comparisons with equivalent human support found no difference. Five analytical themes emerged: polished products with uncertain learning, pedagogy over presence, an affective double edge, unequal readiness and comparator-dependent conclusions. The evidence supported these tools as comparable and scalable complements within teacher-mediated, bounded designs rather than as superior replacements. The review proposed an orchestration framework and a research agenda that prioritised active comparators and independent, delayed measures of learning.