Survey of EU ethical guidelines for commercial AI: case studies in financial services

A macro perspective examining the general nature of AI implementations and how enforcement should be structured under the new frontier of AI technologies is severely needed. The paper critically analyzes real and potential ethical impacts of AI-enabled systems as well as the standard process regulators, researchers, and firms use to assess the risks of these technologies. Three real-world cases are detailed as each type of AI implementation highlights varying degrees of ethical impacts. The paper surveys whether the current September 2020 European Parliament standard titled “European framework on ethical aspects of artificial intelligence, robotics and related technologies” will handle all cases effectively. Future regulators enforcing principles must be able to discern how each AI implementation should be handled as it is evident that there is a spectrum of ethical applicability on the use-case level. The cases examined are in using AI to automate the mortgage application process, find matching attributes for trade reconciliation tasks, and optimizing order entries within trading algorithms. The focus of the cases provided is in the financial industry as firms in this sector, on average, spend the highest percentage of their total revenue on information technology projects. Though the EU framework is thorough in recommending an ex-ante approach, providing risk classifications, and detailing effective measures to mitigate harm from AI-enabled systems, it falls short in identifying certain macroeconomic harms caused by AI-enabled systems and tracking the applications responsible within complex, interlinked systems.