CalvinBackup/www-project-ai-testing-guide
1
0
Fork 0
mirror of https://github.com/OWASP/www-project-ai-testing-guide.git synced 2026-06-01 03:51:42 +02:00
Code Issues Packages Projects Releases Wiki Activity

Update AITG-APP-14_Testing_for_Explainability_and_Interpretability.md

Browse Source
This commit is contained in:
Matteo Meucci
2025-11-13 20:46:34 +01:00
committed by GitHub
parent ff2f879937
commit fcd5e0da08
1 changed files with 4 additions and 10 deletions
Download Patch File Download Diff File Expand all files Collapse all files
Document/content/tests/AITG-APP-14_Testing_for_Explainability_and_Interpretability.md
+4 -10
View File
@@ -1,4 +1,4 @@
## AITG-APP-14 - Testing for Explainability and Interpretability
# AITG-APP-14 - Testing for Explainability and Interpretability
### Summary
This test focuses on evaluating vulnerabilities related to insufficient explainability and interpretability in AI-generated outputs. Lack of explainability can undermine trust, complicate validation and auditing processes, and lead to misinformed or unjustifiable decision-making. An AI system that cannot explain its reasoning is a "black box," making it impossible to verify its decisions, identify biases, or hold it accountable.
@@ -36,15 +36,9 @@ AI-generated outputs must:
- **Develop Explanation Templates**: For recurring decision types, use templates to structure the model's output, ensuring that all key factors and the final reasoning are presented clearly and consistently.
### Suggested Tools
- **SHAP (SHapley Additive exPlanations)**
- A powerful framework for interpreting predictions and understanding the contribution of each feature to model outputs.
- Tool Link: [SHAP GitHub Repository](https://github.com/slundberg/shap)
- **LIME (Local Interpretable Model-agnostic Explanations)**
- Enables local explanations of model predictions, providing insights into individual predictions.
- Tool Link: [LIME GitHub Repository](https://github.com/marcotcr/lime)
- **InterpretML**
- Open-source Python package that incorporates various explainability techniques.
- Tool Link: [InterpretML on GitHub] (https://github.com/interpretml/interpret)
- **SHAP (SHapley Additive exPlanations)** - A powerful framework for interpreting predictions and understanding the contribution of each feature to model outputs - [SHAP GitHub Repository](https://github.com/slundberg/shap)
- **LIME (Local Interpretable Model-agnostic Explanations)** - Enables local explanations of model predictions, providing insights into individual predictions - [LIME GitHub Repository](https://github.com/marcotcr/lime)
- **InterpretML** - Open-source Python package that incorporates various explainability techniques - [InterpretML on GitHub](https://github.com/interpretml/interpret)
### References
- Lundberg, Scott M., and Su-In Lee. "A Unified Approach to Interpreting Model Predictions." Advances in Neural Information Processing Systems (NeurIPS), 2017. [Link](https://papers.nips.cc/paper/2017/hash/8a20a8621978632d76c43dfd28b67767-Abstract.html)