Stop ordering machine learning algorithms by their explainability! A user-centered investigation of the performance and explainability tradeoff
COM3 Level 1
SR14, COM3 01-23
Abstract:
Machine learning algorithms enable advanced decision making in contemporary intelligent systems. Research indicates that there is a tradeoff between their model performance and explainability. Machine learning models with higher performance are often based on more complex algorithms and therefore lack explainability and vice versa. However, there is little to no empirical evidence of this tradeoff from an end user perspective. In my talk I provide empirical evidence by reporting on two user experiments. Using two distinct datasets, we measured the tradeoff for five common classes of machine learning algorithms. Second, we addressed the problem of end user perceptions of explainable artificial intelligence augmentations aimed at increasing the understanding of the decision logic of high-performing complex models. The results diverge from the widespread assumption of a tradeoff curve and indicate that the tradeoff between model performance and explainability is much less gradual in the end user's perception. This is a stark contrast to assumed inherent model interpretability.
Bio:
Christian is Full Professor of Enterprise Computing at TU Dortmund University. Before, Christian worked full-time at various institutions including the Westfälische Wilhelms-Universität in Münster, the SAP Research Center Brisbane of SAP Australia Pty Ltd, the Karlsruhe Institute of Technology, and the Julius-Maximilians-Universität Würzburg. His research focuses on intelligent systems at the intersection of business process management and artificial intelligence with frequent applications in the Industrial Internet of Things. He is on the BPM Department Editorial Board for BISE journal, as well as the editorial boards of Electronic Markets, IJMR, and JBA. He has authored over 150 scholarly publications. His work has appeared in journals such as the Journal of the Association for Information Systems, Communications of the Association for Information Systems, International Journal of Information Management, Information & Management, Business & Information Systems Engineering, Information Systems, Decision Support Systems, Future Generation Computer Systems as well as in various major international conferences including ICIS, ECIS, BPM, and HICSS and has been registered as U.S. patents.