On the 4th of May, IEAI researchers Auxane Boch, Ellen Hohma, and Maria Pokholkova hosted the workshop “System of AI Accountability in Financial Services” as part of their current project “Towards an Accountability Framework for AI Systems” in collaboration with Fujitsu Global.
The main objective of this event was to foster meaningful discussions and knowledge exchange among experts in the field of AI ethics, with a specific focus on quantifying ethical considerations in AI applications. The participants comprised a diverse group of specialists from the financial industry, including insurance, fintech, and financial services. Their backgrounds spanned various domains, such as data science, AI ethics, AI governance, and strategy.
Over the course of four immersive hours, the workshop attendees collectively worked towards formulating and achieving a consensus on five crucial scalable characteristics that define the ethicality of an example use case within the realm of AI Credit Scoring. The intention was to develop a framework that enables the evaluation of the ethical aspects of AI applications in a scalable and standardized manner.
The identified characteristics aimed to reflect the extent to which AI products adhere to the ethical principles of AI, focusing specifically on the principle of 'explainability & transparency' in this instance. By establishing a clear set of characteristics, the workshop participants aimed to provide a robust foundation for assessing the ethical implications of AI Credit Scoring and similar use cases.
To facilitate a quantitative comparison of the ethicality of the use case application, the workshop participants devised a numeric scale. This scale would enable the scoring of the defined characteristics, thereby enabling a more objective and measurable evaluation of the ethical considerations associated with AI Credit Scoring.
The workshop was a resounding success, with lively discussions, insightful perspectives, and fruitful collaborations taking place throughout the event. The engagement and expertise of the participants contributed to the development of a framework that promises to enhance the accountability of AI systems in the financial services sector.
The outcomes of this workshop are expected to have a significant impact on the ongoing research and development of AI accountability frameworks. The scalable characteristics and the numeric scale generated during the event will serve as valuable tools for organizations seeking to assess the ethical implications of their AI applications, particularly in the domain of AI Credit Scoring.
- The workshop resulted in the identification of five characteristics representing the degree of adherence to the ethical principle of ‘transparency & explainability’ in credit scoring systems.
- A scale was developed to assess the current implementation of these characteristics and thus quantify the ethicality of the investigated credit scoring application.
- Our preliminary results pave the way for further steps in quantifying AI systems' adherence to ethical principles.
Overall, the workshop on the "System of AI Accountability in Financial Services" proved to be a vital step forward in addressing the ethical challenges posed by AI systems. By bringing together experts from various fields and establishing a consensus on scalable characteristics, this event has paved the way for greater transparency and accountability in the financial industry's adoption of AI.
This workshop on the "System of AI Accountability in Financial Services" was a vital step forward in addressing the ethical challenges posed by AI systems. By bringing together experts from various fields and establishing a consensus on scalable characteristics, this event paved the way for greater transparency and accountability in the financial industry's adoption of AI.