Download PDFOpen PDF in browserExplainable by Design: A design framework to support the design of explainable user interfacesEasyChair Preprint 159927 pages•Date: August 7, 2025AbstractThe technological advancement of artificial intelligence (AI) and large language models (LLMs) are rapidly changing what systems can do and how people interact with them. Usable and Explainable AI (XAI) is identified as a key human-centred AI (HCAI) challenge to address. Although existing research offers high-level principles and guidelines to design for AI, there is limited support in how to translate these into interface decisions, specifically for the design of explainable user interfaces (XUI). This research aims to address the gap through the development of a practical framework for user experience (UX) designers. Through the RtD methodology, the research follows a qualitative approach across four phases. In Phase 1 a scoping review and thematic analysis identified interface-level XUI guidelines. In Phase 2 the guidelines were validated and operationalised into practitioner-facing reflective design questions through two expert reviews. In Phase 3, a participatory workshop with UX designers classified UI patterns across explanation dimensions to support explainability. Finally the outputs across all phases were synthesised into a practical and flexible framework for UX designers. A set of 5 learning cards introduce the theoretical foundations, 14 XUI guidelines accompanied by reflective questions support theory in practice, and a UI pattern decision tree to guide the selection of design patterns. The result is a design artifact that bridges academic theory and design practice for the design of XUIs in AI products and systems. Keyphrases: Explainable AI (XAI, Explainable user interfaces (XUI), human-centered design, interaction design, user experience
|