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
Teaching AI models to improve themselves
ResearchPeter HessIBM and RPI researchers demystify in-context learning in large language models
NewsPeter HessThe latest AI safety method is a throwback to our maritime past
ResearchKim MartineauFind and fix IT glitches before they crash the system
NewsKim MartineauWhat is retrieval-augmented generation?
ExplainerKim MartineauDid an AI write that? If so, which one? Introducing the new field of AI forensics
ExplainerKim Martineau- See more of our work on Explainable AI
Publications
Comprehensive Layer-Wise Analysis of SSL Models for Audio Deepfake Detection
- Yassine Elkheir
- Younes Samih
- et al.
- 2025
- NAACL 2025
Workshop on Neuro-Symbolic Software Engineering
- Christian Medeiros Adriano
- Sona Ghahremani
- et al.
- 2025
- ICSE 2025
New Frontiers of Human-centered Explainable AI (HCXAI): Participatory Civic AI, Benchmarking LLMs and Hallucinations for XAI, and Responsible AI Audits
- Upol Ehsan
- Elizabeth Watkins
- et al.
- 2025
- CHI 2025
Explain Yourself, Briefly! Self-Explaining Neural Networks with Concise Sufficient Reasons
- Shahaf Bassan
- Ron Eliav
- et al.
- 2025
- ICLR 2025
Rationalization Models for Text-to-SQL
- Gaetano Rossiello
- Nhan Pham
- et al.
- 2025
- ICLR 2025
Discovering Group Structures via Unitary Representation Learning
- Ben Huh
- 2025
- ICLR 2025