Using predictive multiplicity to measure individual performance within the AI Act
University of Tübingen and Tübingen AI Center(图宾根大学和图宾根人工智能中心);University of Tübingen and CZS Institute for Artificial Intelligence and Law(图宾根大学和图宾根人工智能与法律研究所);University of Tübingen, Tübingen AI Center and CZS Institute for Artificial Intelligence and Law(图宾根大学、图宾根人工智能中心和图宾根人工智能与法律研究所)
Karolin Frohnapfel, Mara Seyfert, Sebastian Bordt, Ulrike von Luxburg, Kristof Meding
备注:
ACM Conference on Fairness, Accountability, and Transparency, 2026. Updated version
When building AI systems for decision support, one often encounters the phenomenon of predictive multiplicity: a single best model does not exist; instead, one can construct many models with similar overall accuracy that differ in their predictions for individual cases. Especially when decisions have a direct impact on humans, this can be highly unsatisfactory. For a person subject to high disagreement between models, one could as well have chosen a different model of similar overall accuracy that would have decided the person's case differently. We argue that this arbitrariness conflicts with the EU AI Act, which requires providers of high-risk AI systems to report performance not only at the dataset level but also for specific persons. The goal of this paper is to put predictive multiplicity in context with the EU AI Act's provisions on accuracy and to subsequently derive concrete suggestions on how to evaluate and report predictive multiplicity in practice. Specifically: (1) We introduce the AI Act's accuracy provisions and argue that incorporating information about predictive multiplicity could serve compliance with specific provisions for providers. (2) Based on this legally rigorous analysis, we suggest individual conflict ratios and $δ$-ambiguity as tools to quantify the disagreement between models on individual cases and to help detect individuals subject to conflicting predictions. (3) Based on computational insights, we derive easy-to-implement rules on how model providers could evaluate predictive multiplicity in practice. (4) Ultimately, we suggest that information about predictive multiplicity should be made available to deployers under the AI Act, enabling them to judge whether system outputs for specific individuals are reliable enough for their use case.
Investigating the Impact of Balancing, Filtering, and Complexity on Predictive Multiplicity: A Data-Centric Perspective
Eskisehir Technical University, Department of Statistics(埃斯克谢希大学统计系);Warsaw University of Technology, Faculty of Mathematics and Information Science(华沙技术大学数学与信息科学学院);University of Warsaw, Faculty of Mathematics, Informatics and Mechanics(华沙大学数学、信息学与力学学院)
On Arbitrary Predictions from Equally Valid Models
Hasso Plattner Institute, University of Potsdam, Germany(波茨坦大学霍索普纳研究所,德国);Chair for AI in Healthcare and Medicine, Technical University of Munich, Germany(慕尼黑技术大学医学与健康人工智能教研室,德国);BioMedIA, Imperial College London, United Kingdom(伦敦帝国理工学院BioMedIA,英国);Munich Center for Machine Learning, Germany(慕尼黑机器学习中心,德国);Universitätsklinikum Hamburg-Eppendorf, Germany(汉堡-埃彭多夫大学医院,德国)
An Experimental Study on the Rashomon Effect of Balancing Methods in Imbalanced Classification
Eskisehir Technical University, Department of Statistics, Turkiye(埃斯基谢普技术大学统计系,土耳其);Warsaw University of Technology, Faculty of Mathematics and Information Science(华沙技术大学数学与信息科学系);University of Warsaw, Faculty of Mathematics, Informatics and Mechanics(华沙大学数学、信息学与力学系)
Horizon-Constrained Rashomon Sets for Chaotic Forecasting
Dept. of Electrical Engineering, California State University Long Beach, USA;WorkOnward Inc., USA;Dept. of Computer Engineering \& Computer Science, California State University Long Beach, USA
Cross-model Fairness: Empirical Study of Fairness and Ethics Under Model Multiplicity
ARC Centre of Excellence for Automated Decision-Making and Society(自动化决策与社会卓越研究中心);RMIT University(皇家墨尔本理工大学);ETH Zurich(苏黎世联邦理工学院);Institute of Computer Science, University of Tartu(塔尔图大学计算机科学研究所);UNSW Sydney(新南威尔士大学悉尼分校)