Modern online digital art repositories have become an important asset in the toolbox of higher education institutions active in the fields of art and culture. These repositories which include works from faculty, students and alumni may act both as a way for people to showcase their work and as a means for the departments themselves to gain recognition. In the era of generative artificial intelligence (GenAI) the user experience (UX) offered by these endeavours may be optimized to better address the specific interests of each different visitor.

This study explores how GenAI methodologies may be utilized to enhance UX through the case of the AVARTS Portfolio, the digital art repository of the Department of Audio and Visual Arts of the Ionian University. Focus is placed in methods that allow better similarity-based recommendations and a more insightful search functionality, which will help visitors easily discover artworks better suited to their interests both based on their browsing history and on provided search queries. In order to accommodate the practical realities of institutional digital art repositories, emphasis is placed in both performance and low resource requirements.

The presented methodology discusses the generation of semantic embeddings based on the textual descriptions, classification and annotation of works of digital art. These dense vector representations allow the system’s database to perform similarity checks with minimal input from the GenAI agent. The cases of artworks with high similarity may then not only be presented to the user but also used as context in a process of Retrieval Augmented Generation (RAG) to compose a detailed proposal of new content tailored to the visitors’ explicit and implicit interests. This type of personalization in content suggestion aims at increasing user engagement by offering an overall more adaptable and interactive user experience.

Through detailing the implementation process, this study addresses the technological and resource management challenges of incorporating GenAI capabilities in higher education art repositories in a manner that is both practical and impactful. Moreover, considerations are raised concerning the ethical use of GenAI in digital curation and the effects it may have on the entire Web art exhibition ecosystem. A series of best practices and implementation suggestions are presented, and conclusions are drawn regarding this initial applied approach to GenAI powered UX personalization. Through harnessing the power of artificial intelligence, academic digital art repositories may be able to better adapt to the ever-evolving landscape of the Web and become high quality art and cultural content providers to the benefit of both the academic community and the institutions themselves.

Keywords: User Experience, Generative Artificial Intelligence, Web Development, Digital Art, Higher Education.