Fuzzy Kano Validation of Generative AI-Curated Features for Simponia Redesign

Abstract

Background: Generative AI (Gen AI) has accelerated UX design processes, yet no systematic protocol exists to validate AI-curated features against actual user preferences before prototyping. Objective: This study applies the Fuzzy Kano Model as a validation checkpoint for Gen AI-curated features in the redesign of Simponia, a web-based portfolio and community management platform at Universitas Muhammadiyah Malang. Methods: An exploratory mixed-methods case design was used. Claude Opus generated 36 feature candidates from a Structured Requirements Document and ISO/IEC 25010 criteria. Fuzzy Kano validation involved 87 Informatics students; fuzzy membership and the Last of Maxima classification method were used to determine Kano categories, while weighted scoring was retained only for post-classification prioritization. Results: No features were classified as Must-be or One-dimensional (maximum Must-be membership degree = 0.073). Eleven features (30.6%) were classified as Attractive, and 25 (69.4%) as Indifferent. Triangulation indicated that 58.3% of features were in Quadrant 2, suggesting a misalignment between AI-based technical evaluation and actual user preferences. Sixteen features (44.4%) were flagged as ambiguous (Δμ < 0.10). Conclusion: The Fuzzy Kano Model demonstrates potential as a validation checkpoint by exposing systematic divergence between AI curation and user preferences, with a Quadrant 2 (Q2) rate of 58.3%, indicating the importance of user validation following autonomous AI-assisted feature curation in resource-constrained educational contexts.