<![CDATA[As adaptive learning systems are increasingly implemented to support personalized education, concerns have emerged regarding their tendency to rely on narrow and reductionist intelligence models. This study aims to analyze the influence of intelligence variations in adaptive learning design through a systematic literature review. Using the PRISMA 2020 protocol, fifteen empirical articles from the period 2010–2025 were critically reviewed. The results show the dominance of rule-based approaches with static profiles that rely on the linguistic-logical-mathematical dimension as the default parameter, ignoring the potential of other dimensions such as naturalist and existential. Artificial intelligence integration offers dynamic personalization potential but poses pedagogical and ethical dilemmas. Empirical evidence in Indonesia confirms the effectiveness of multiple intelligences-based adaptive learning on reading literacy and science creativity, despite constraints related to infrastructure and teacher capacity. The findings lead to three design principles: multimodal flexibility, cultural calibration of measurement instruments, and technology–pedagogical balance. This study recommends the development of hybrid prototypes and ethical standards for the use of cognitive data to realize an inclusive learning ecosystem. The implications of this review indicate that adaptive learning design should move beyond single-intelligence models by adopting culturally calibrated, multimodal, and pedagogically guided adaptation strategies. These implications provide a concrete framework for educators, designers, and developers to design more inclusive and context-responsive adaptive learning systems.]]>
