Citation for the Ranyard Medal 2024
A Social Evaluation of the Perceived Goodness of Explainability in Machine Learning
Journal of Business Analytics, Volume 5 (1), 29–50
https://www.tandfonline.com/10.1080/2573234X.2021.1952913
This year’s Ranyard Medal is presented to Jonas Wanner, Lukas-Valentin Herm, Kai Heinrich and Christian Janiesch, for their Paper entitled “A social evaluation of the perceived goodness of explainability in machine learning”
Machine learning and Artificial Intelligence are now seen as the go-to solution to almost every problem. But there remains a key question as to whether potential users trust the advice offered by such methods, accepting its direction or modifying it to suit their understanding of the problem. Many of these methods offer solutions – recommendations or forecasts – that are black box in essence. This paper explores the concept of explainable machine learning, which they break into two dimensions: global, focused on the algorithm’s features, and local explainability concerned with the specifics of its performance. The authors have applied several machine learning algorithms to various contrasting problems and evaluated the users’ experience of these models through a survey. Among others, they found that the perceived explainability depends on the familiarity with the problem type and interaction of the modellers with the users. The findings of the research we see as very helpful for modellers in creating models that are more useful to practitioners, thereby achieving a more significant impact with machine learning methods better integrated with user needs.