Explainable AI
To trust AI systems, explanations can go a long way. We’re creating tools to help debug AI, where systems can explain what they’re doing. This includes training highly optimized, directly interpretable models, as well as explanations of black-box models and visualizations of neural network information flows.
Our work
Saška Mojsilović wants to channel AI for good. She may also make you rethink sour cabbage
NewsDebugging foundation models for bias
ResearchNew AI research could help find the best treatment for bowel disease
ResearchProject Debater argues for key point analysis of surveys, reviews, and social media
ReleaseTapping into the inner rhythm of living organisms with AI and ML
ResearchAI, you have a lot of explaining to do
Release
Publications
- AUTOLYCUS: Exploiting Explainable AI (XAI) for Model Extraction Attacks against Decision Tree Models
- 2023
- NDSS 2023
- 2023
- AAAI 2023
- 2023
- AAAI 2023
- 2022
- Big Data 2022
- 2022
- ICLR 2022
- 2022
- PLoS ONE
Tools + code
AI Explainability 360
This open source toolkit contains eight algorithms that help you comprehend how machine-learning models predict labels throughout the AI application lifecycle. It’s designed to translate algorithmic research into the real-world use cases in a range of files, such as finance, human capital management, healthcare, and education.
View project →ECQA Dataset
The open-source Explanations for CommonsenseQA (ECQA) dataset is a resource to teach AI systems how to reason about the correct and incorrect answers to everyday common-sensical questions.
View project →