Beyond Prediction: Managing the Repercussions of Machine Learning Applications
Aline Weber, Blossom Metevier, Yuriy Brun, Philip S. Thomas, Bruno C. da Silva
Abstract
Machine learning models are often designed to maximize a primary goal, such as accuracy. However, as these models are increasingly used to inform decisions that affect people's lives or well-being, it is often unclear what the real-world repercussions of their deployment might be-making it crucial to understand and manage such repercussions effectively. Models maximizing user engagement on social media platforms, e.g., may inadvertently contribute to the spread of misinformation and content that deepens political polarization. This issue is not limited to social media -it extends to other applications where machine learning-informed decisions can have real-world repercussions, such as education, employment, and lending. Existing methods addressing this issue require prior knowledge or estimates of analytical models describing the relationship between a classifier's predictions and their corresponding repercussions. We introduce THEIA, a novel classification algorithm capable of optimizing a primary objective, such as accuracy, while providing high-confidence guarantees about its potential repercussions. Importantly, THEIA solves the open problem of providing such guarantees based solely on existing data with observations of previous repercussions. We prove that it satisfies constraints on a model's repercussions with high confidence and that it is guaranteed to identify a solution, if one exists, given sufficient data. We empirically demonstrate, using real-life data, that THEIA can identify models that achieve high accuracy while ensuring, with high confidence, that constraints on their repercussions are satisfied.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
Lineage
Builds on5
- Being Optimistic to Be Conservative: Quickly Learning a CVaR PolicyRamtin Keramati, Christoph Dann, Alex Tamkin, Emma BrunskillAAAI 2020 · 86 citations
- Minimizing Polarization and Disagreement in Social Networks via Link RecommendationLiwang Zhu, Qi Bao, Zhongzhi ZhangNeurIPS 2021 · 68 citations
- Universal Off-Policy EvaluationYash Chandak, Scott Niekum, Bruno C. da Silva, Erik G. Learned-Miller et al.NeurIPS 2021 · 64 citations
- Estimating and Penalizing Induced Preference Shifts in Recommender SystemsMicah D. Carroll, Anca D. Dragan, Stuart Russell, Dylan Hadfield-MenellICML 2022 · 49 citations
- Preference Amplification in Recommender SystemsDimitris Kalimeris, Smriti Bhagat, Shankar Kalyanaraman, Udi WeinsbergKDD 2021 · 24 citations
Related papers
- From Predictions to Decisions: Using Lookahead RegularizationNir Rosenfeld, Sophie Hilgard, Sai Srivatsa Ravindranath, David C. ParkesNeurIPS 2020 · 27 citations
- From Plane Crashes to Algorithmic Harm: Applicability of Safety Engineering Frameworks for Responsible MLShalaleh Rismani, Renee Shelby, Andrew Smart, Edgar W. Jatho III et al.CHI 2023 · 33 citations
- Welfare-Optimal Classification with Accuracy AuctionsBana Sadi, Eden Saig, Nir RosenfeldICML 2026
- Revisiting (Un)Fairness in Recourse by Minimizing Worst-Case Social BurdenAinhize Barrainkua, Giovanni De Toni, José Antonio Lozano, Novi QuadriantoAAAI 2026
- Enforcing Constraints for Machine Learning Systems via Declarative Feature Selection: An Experimental StudyFelix Neutatz, Felix Biessmann, Ziawasch AbedjanSIGMOD 2021 · 7 citations
