AI & Trust Calibration for Financial Literacy : Design Science Research for Trust in Human-AI Interaction

Abstract

Would you trust your savings to the algorithms of AI financial advisor instead of the hands of Human professional? Artificial Intelligence (AI) is changing the lives of individuals and organizations alike, and the hype spells the fourth industrial revolution. With AI repeatedly exceeding even the most exaggerated expectations and the industry relentlessly pushing new innovations of AI technology, largely considered "black-box", for common user and domain expert alike, knowing when to truly trust the AI answer as a fact, is surprisingly difficult. In high-stakes domains, like Finance, where informed decisions are paramount, trustworthy AI plays a significant role. This thesis embarks on a quest to find out how to create a trustworthy AI personal assistant for improving financial literacy and to help navigate in digital financial environments. Research is simultaneous to thesis for MEng in same topic by the same author, both using Design Science Research (DSR) methodology. Design cycles of DSR cover topics in AI knowledge veracity, personalization and memory. As the DSR artifact, an AI Agent chatbot application is created, experimented with and found convincingly capable simulating a human advisor. Findings highlight cross-discipline co-operation in prompt engineering and prompt management research, but many others appear during the sprawling storytelling of research.