GAM Coach: Towards Interactive and User-centered Algorithmic Recourse

crown jewel figure
With a novel interactive interface and an adaptation of integer linear programming, GAM Coach empowers people who are impacted by machine learning-based decision-making systems to iteratively generate algorithmic recourse plans that reflect their preferences. Take loan application as an example. (A) The Coach Menu helps a rejected loan applicant browse diverse recourse plans that would lead to loan approval. After the user selects a plan, (B) the Feature Panel visualizes all feature information with progressive disclosure, enabling users to explore how hypothetical inputs affect the model’s decision and specify recourse preferences—such as (B1) the difficulty of changing a feature and (B2) its acceptable range of values—guiding GAM Coach to generate actionable plans. (C) The Bookmarks window allows users to compare bookmarked plans and save a verifiable receipt.
Demo Video
Machine learning (ML) recourse techniques are increasingly used in high-stakes domains, providing end users with actions to alter ML predictions, but they assume ML developers understand what input variables can be changed. However, a recourse plan's actionability is subjective and unlikely to match developers' expectations completely. We present GAM Coach, a novel open-source system that adapts integer linear programming to generate customizable counterfactual explanations for Generalized Additive Models (GAMs), and leverages interactive visualizations to enable end users to iteratively generate recourse plans meeting their needs. A quantitative user study with 41 participants shows our tool is usable and useful, and users prefer personalized recourse plans over generic plans. Through a log analysis, we explore how users discover satisfactory recourse plans, and provide empirical evidence that transparency can lead to more opportunities for everyday users to discover counterintuitive patterns in ML models. GAM Coach is available at:
GAM Coach: Towards Interactive and User-centered Algorithmic Recourse
  title = {{{GAM Coach}}: {{Towards Interactive}} and {{User-centered Algorithmic Recourse}}},
  booktitle = {{{CHI Conference}} on {{Human Factors}} in {{Computing Systems}}},
  author = {Wang, Zijie J. and Vaughan, Jennifer Wortman and Caruana, Rich and Chau, Duen Horng},
  year = {2023},
  doi = {10.1145/3544548.3580816}