On the Explainability of Financial Robo-Advice Systems

Significant investment and development have been made in integrating artificial intelligence (AI) into finance applications. However, the opacity of AI systems raises concerns about essential characteristics needed in sensitive finance applications, such as transparency and accountability. Our study addresses these concerns by investigating a process for analysing explanations generated by AI-based robo-advice systems to comply with the explanation requirements of key EU regulations, including the Markets in Financial Instruments Directive (MiFID) II. We adopt a comprehensive methodology that involves analysing these regulations to identify the specific questions that must be answered by an explanation generated by a robo-advice system to meet legal explainability requirements. Our findings provide a nuanced understanding of which Explainable AI technology may be needed to answer those questions by AI-based robo-advice systems. We demonstrate this through practical case studies on financial advice given by robo-advisers based on AI language-generating models. These case studies highlight our research’s practical utility for the finance industry, which might seek to exploit these new technologies to generate financial advice that meets legal explainability requirements. This study fills a crucial gap in aligning Explainable AI applications in finance with stringent provisions of EU regulations. It provides a practical framework for developers and researchers to ensure their AI innovations advance technology and adhere to legal and ethical standards.