The purpose of this study is to design a multisensory arts integration program mediated by generative artificial intelligence and to examine its educational potential and implications. To this end, prior research on generative AI in education and arts integration was reviewed, and interdisciplinary elements were derived from the achievement standards of the 2022 revised national curriculum in art, Korean language, and music. Teaching for Understanding (TfU), developed by Harvard Project Zero, was adopted as the design framework. “Sensation” was identified as a generative topic, and an Artlog-based instructional model was developed, structured around the phases of exploration, expression, sharing, and reflection. The proposed program enables learners to reconstruct sensory experiences through multiple modes of representation, including language, visual imagery, and music, while extending creative expression and critical understanding through interaction with generative AI. This approach reconceptualizes AI as a mediating agent that fosters the expansion of thinking in arts-integrated learning and offers a practical framework for integrating digital literacy and AI ethics education.