Literature Review: Designing for Explainability in Financial Credit Assessment: XAI Interaction Strategies for Non-expert Users

Abstract

This review focuses on XAI interaction strategies for non-expert users in financial credit assessment. As AI models gain traction in finance, especially in credit scoring, explainability becomes crucial for trust, fairness, and user understanding. Regulatory, ethical, and practical needs drive XAI development. LIME and SHAP are key techniques for explaining complex models.

However, XAI's success in financial credit assessment relies on user - centered design. Principles like contextualization and interactivity are important for effective explanation experiences. User - centric evaluation methods are essential to gauge XAI's impact on users.

There are still challenges, such as balancing explanation complexity and user comprehension, and addressing ethical issues. Future research should focus on user - adaptive interfaces, causal explanations, and integrating XAI with financial education. Overall, XAI design in financial credit assessment is a human - centered problem. Optimizing user experience and following research directions can revolutionize financial credit assessment and build a more reliable financial future.