Understanding the Dynamics in Creating Domain-Specific AI Design Guidelines: A Case Study of a Leading Digital Finance Company in South Korea

Abstract

Artificial intelligence (AI) significantly impacts the user experience (UX) across a wide range of products and services. However, many practitioners find it challenging to design the UX for AIenhanced products. Although there are human-AI interaction design guidelines available, they are often too generic to fully address domain-specific challenges in fields such as finance and healthcare. Additionally, they sometimes overlook the constraints imposed by different working environments. In this research, we conducted a case study at one of the leading financial corporations in South Korea to develop their own AI product interaction guidelines. We interviewed 13 practitioners including bank UX practitioners, design agency UX designers, and academic UX researchers who were involved in this initiative, to gather insights into their experiences with crafting these guidelines. Our findings revealed a mixed reception towards existing AI design resources among practitioners, underscoring the necessity to incorporate elements specific to banking services, such as the company's customer experience (CX) principles, overarching agent personas, multifaceted channels, and diverse interaction modalities. We also observed the importance of understanding the dynamics among various stakeholders within a company.We discuss the implications and design considerations for creating domain-specific resources specialized in AI interactions.