Uncertainty Quantification
When AI can explain to us that it's unsure, it adds a critical layer of transparency for its safe deployment and use. We’re developing ways to foster and streamline the common practices of quantifying, evaluating, improving, and communicating uncertainty in the AI application development lifecycle.
Our work
IBM’s Uncertainty Quantification 360 toolkit boosts trust in AI
ReleasePrasanna Sattigeri and Vera Liao7 minute readAI boosts the discovery of metamaterials vital for next-gen gadgets
ResearchYoussef Mroueh, Karthikeyan Shanmugam, and Payel Das10 minute read
Publications
Uncertainty Analysis of Molecular Quantum Properties Prediction Using Chemical Foundation Models
- Siya Kunde
- Emilio Ashton Vital Brazil
- et al.
- 2025
- ACS Spring 2025
User centered approach of applicability domain analysis for improving AI-assisted decision-making
- Siya Kunde
- Emilio Ashton Vital Brazil
- 2025
- ACS Spring 2025
Uncertainty Characterization of Foundation Models for Reliable Applications in Materials and Chemistry
- Emilio Ashton Vital Brazil
- Priscilla Barreira Avegliano
- et al.
- 2025
- ACS Spring 2025
Uncertainty Analysis in Predicting Molecular Properties Using Chemical Foundation Models
- Siya Kunde
- Emilio Ashton Vital Brazil
- et al.
- 2025
- AAAI 2025
Sequential uncertainty quantification with contextual tensors for social targeting
- 2024
- KAIS
Graph-based Uncertainty Metrics for Long-form Language Model Generations
- Mingjian Jiang
- Yangjun Yangjun
- et al.
- 2024
- NeurIPS 2024