Algorithmic Fairness in Finance
As AI is gaining a new spot in Finance due to unforeseen possibilities offered by Deep Learning in building complex decisional models, some questions arise regarding the fairness of algorithms that have been trained in order to maximize a given utility function that prioritizes quantitative figures such as returns, risks and costs, somehow irrespective of what they represent in reality. The outcome of algorithms is generally limited to the narrow decisional scope they are applied to, but a wide adoption of AI algorithms should question us about the effect of summing local narrow decisions at systemic level.
Luigi Troiano is professor of Artificial Intelligence, Data Science and Machine Learning at University of Sannio, Department of Engineering, Italy. He is chairman of the Italian section of ISO IEC/JTC1/SC 42 for Artificial Intelligence and Big Data, and University Ambassador of NVIDIA Deep Learning Institute (DLI). His research is devoted to mathematical modeling and algorithm development with applications to Finance and other industries. His expertise is designing, experimenting and validating algorithms, along their implementation in software systems for industrial environments, including some large international companies. He is coordinator of Computational and Intelligent Systems Engineering Laboratory (CISELab) at University of Sannio, aimed at developing research in Big Data and Deep Learning.