The Michigan Institute for Data and AI in Society (MIDAS) hosted the University of Michigan Uncertainty Quantification (UQ) Incubator from May 31 to June 3, 2026, bringing together researchers from a range of disciplines, including uncertainty quantification (UQ), artificial intelligence, science and engineering, to explore how AI systems can become more reliable, interpretable and trustworthy.
Supported by Schmidt Sciences, the workshop connected UQ experts who develop methods for measuring uncertainty with scientists and engineers applying AI to complex challenges across research domains.
As AI becomes increasingly integrated into scientific discovery, researchers need ways to understand not only what AI models predict, but also how confident they are in those predictions. The UQ Incubator focused on advancing uncertainty-aware approaches that can help researchers build AI systems that are more transparent, dependable and effective in real-world applications.
Held at the University of Michigan, the three-day event featured keynote presentations, tutorials, peer-learning activities, networking sessions and collaborative research prototyping. Participants formed interdisciplinary teams to develop research ideas, generate preliminary results and identify opportunities for continued collaboration.
Seven teams received $2,500 follow-up research awards to continue advancing projects launched during the workshop.
The awardee teams were:
- AI-enabled and sustainable recovery of rare earth elements from e-waste
Team AtomUQ: Chia-Hao Lee, Seth Temple, Anqi Qiu, Nanta Sophonrat - Uncertainty-aware prediction of alcohol use disorder from large-scale incomplete datasets
Team Blue Tiger: Chenlan Wang, Nathaly Villacís, Dimah Dera - An assumption-level view of uncertainty quantification in machine-learning pipelines for groundwater prediction
Team AquaProbe: Mahzad Khoshlessan, Morli Li, Omolola Ogbolumani, Joseph Osumeje, Jennifer Pazour - Uncertainty prediction of rogue wave formation
Team Ship Savers: Adwait Sharma, Ponkrshnan Thiagarajan, Gary Wu, Yue Cynthia Wu - Uncertainty quantification in orthopedic surgery
Team Joint Efforts: Hsuan Chou, Ruipu Li, Zihan Li, Kuan-Fu Chen - Predicting geomagnetic storms: Regression modeling of disturbance storm time index with conformal prediction
Daniel Adamiak, Elser Lopez, Verrah A. Otiende, McKenna J. D. Breddan - Uncertainty-aware light level optimization for low-light cell imaging in disease screening
Linyun He, Fan Wu
By bringing together researchers across disciplines, the UQ Incubator helped build a community focused on developing the tools, methods and partnerships needed to make AI a more trusted resource for scientific discovery and engineering innovation.
MIDAS thanks Schmidt Sciences for its generous support, along with the keynote speakers, tutorial leaders, organizers and participants who contributed to the success of the workshop.