Designing Trust in AI-Assisted Fashion Education

Artificial intelligence is transforming fashion design education by enabling users to produce visually convincing garments, campaign images, and videos with unprecedented speed. However, as the visual quality of AI-assisted outputs becomes increasingly independent of the depth of human-led design reasoning, final outcomes become less reliable as evidence of student learning, authorship, and disciplinary judgment. This article examines this problem through a reflective practice-based case. An educator-led fashion project was documented from initial asset development to AI generation, redesign and market-ready visualization. The analysis reveals several critical tensions, including the limited perceptibility of the designer–system ratio, the limited visibility of the connection between visual research and print development, the mixed sources of the uploaded assets, and the risk that repeated generation and visual selection may replace deliberate design reasoning. The article introduces the concept of the “output–evidence gap” to describe the disconnection between the apparent quality of AI-assisted outcomes and the evidence they provide of authorship, learning, and critical judgment. It subsequently proposes a Designing Trust Framework structured around framing, sourcing, translating, curating, generating, diagnosing, intervening, and accounting. The article argues that trust in AI-assisted design education cannot be based on visual polish or platform-declared human control alone. Instead, it should be built through trackable processes, domain-specific judgment, meaningful human intervention, and deliberate pedagogical mediation.