I Introduction

The widespread use of data technology systems (e.g., machine learning, neural networks and adaptive algorithms) in the banking industry has resulted in an increase in automated decision-making processes to expedite compliance with regulatory requirements.1 Specifically, automated practices such as sandbox programs and application programming interfaces (APIs) support regulatory oversight over fintech products through sophisticated software which requires advanced IT infrastructure.2 However, automated mechanisms must be designed, customised, set up, maintained and overseen: for example, natural language processing and cognitive computing need the intervention of human capital in order to avoid disruption for customers.3 Financial firms have become hardwired into digital platforms, which raises concerns about the potential risks of increased competition and disintermediation: the increase in data breaches and cyber-crimes brings a degree of uncertainty in digital banking.4 Market users rely on artificial intelligence (AI) to assess business products and formulate investment decisions, but they have limited understanding of the negative effects of undesired outcomes on their choices.5