Age-Appropriate Mobile Banking Interface Design Using AI-Based Age Estimation

Abstract

The rapidly aging global population has made the accessibility of digital banking services a critical design challenge. Older adults face barriers in mobile banking applications such as small font sizes, complex navigation, and high information density. In this study, an Android mobile banking prototype has been developed that automatically determines the user’s age group through AI-based facial age estimation and accordingly presents two distinct user interface profiles. The application is built with Kotlin, Jetpack Compose, and Material Design 3, performing real-time face detection via CameraX and Google ML Kit Face Detection. The detected face region is passed to an open-source, pre-trained on-device TensorFlow Lite convolutional neural network (CNN) regression model trained on the UTKFace dataset to estimate the user’s age; since the entire computation runs locally, the facial image never leaves the device. The profile designed for users aged 55 and above incorporates literature-based design decisions including WCAG 2.2 AAA-level contrast ratios, 22 sp body text, 89 dp touch targets, and step-by-step transaction flows. For users under 55, an information-rich, modern dashboard interface compliant with Material 3 standards is provided. The prototype encompasses core banking functionalities such as balance inquiry, money transfer, bill payment, gold trading, loan application, and card management. The findings indicate that age-adaptive interfaces hold significant potential for facilitating older adults’ access to digital banking.