These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
RecourseBench
RecourseBench is an open, modular software framework for benchmarking algorithmic recourse methods that generate counterfactual explanations telling individuals what changes would be needed to reverse an unfavourable automated decision. Developed by researchers at the University of Waterloo, the framework decomposes the evaluation pipeline into five decoupled layers (Data, Processing, Model, Recourse Method, and Evaluation), allowing new datasets, models, and methods to be integrated without altering existing components.
It features a four-tier reproducibility classification system that automatically tests whether each integrated method's implementation reproduces the results reported in its source publication, using a defined statistical tolerance and transparent per-method documentation of reproduction barriers (e.g. missing artifacts, undocumented hyperparameters). This turns reproducibility from an implicit assumption into a measurable, auditable property of the benchmark.
Results are available through an interactive web interface that allows practitioners to configure comparisons by dataset, model architecture, method, and metric profile. It provides practical infrastructure for AI developers, auditors, and researchers seeking to select, evaluate, or govern recourse-generating systems.
Auto-discovered on 2026-08-26 by OECD Catalogue Automation
About the tool
You can click on the links to see the associated tools
Tool type(s):
Objective(s):
Target sector(s):
Country/Territory of origin:
Lifecycle stage(s):
Type of approach:
Maturity:
Target users:
Stakeholder group:
Benefits:
Geographical scope:
Required skills:
Risk management stage(s):
Programming languages:
Use Cases
Would you like to submit a use case for this tool?
If you have used this tool, we would love to know more about your experience.
Add use case
Partnership on AI
