UT to Lead the Next Gen of AI

NSF AI Institute for Foundations of Machine Learning (IFML) secures renewed funding to support breakthrough research in generative AI.

Karen Davidson

NSF IFML

The National Science Foundation (NSF) has reaffirmed its confidence in UT Austin by awarding the NSF AI Institute for Foundations of Machine Learning (IFML) a second round of funding, making it one of only five institutes nationwide to receive renewed investment. The new support is part of a broader $100 million initiative to secure American leadership in artificial intelligence.

This continued investment will supercharge foundational AI research at UT Austin, positioning the university—and the state of Texas—at the forefront of the next wave of transformative technologies. From ensuring accurate and fair imaging in generative AI to enabling breakthrough discoveries in drug development, IFML is laying the groundwork for high-impact applications across industries.
 

Accurate and Reliable AI  
 

IFML researchers focus on what lies beneath the surface of today’s most powerful AI systems: the algorithms, architectures, and theoretical frameworks that determine model accuracy, reliability, and scalability. Their work has already shaped how machine learning is applied in critical areas such as healthcare imaging, where IFML-developed algorithms enhance the clarity and speed of MRI scans, and in generative AI, where diffusion models are used to reduce noise in visual data.

Support from NSF will enable IFML to tackle some of the most pressing challenges in the field, including establishing best practices for training and fine-tuning massive models, ensuring deep networks are robust and interpretable, and developing methods for domain adaptation in fields like protein engineering and AI in healthcare.

Crucially, the funding will also fuel the institute's workforce development efforts, supporting new postdoctoral fellows and graduate students and building on the university’s successful Master of Science in Artificial Intelligence program.
 

A Commitment to Open Science and Cross-Sector Impact  
 

A core tenet of the institute is a commitment to open-source tools and frameworks. “Machine learning is the engine powering AI across every industry,” said IFML Director Adam Klivans, professor of computer science at UT Austin. “But too often, it’s locked behind proprietary walls. IFML is committed to open-source development, ensuring that the breakthroughs we make are accessible and impactful across sectors—from tech to healthcare to academia.”

This ethos of accessibility has led to real-world adoption. Notably, IFML has contributed to the development of tools like OpenCLIP and DataComp, which have helped redefine how generative AI systems process visual and textual information. In its next chapter, the Institute will extend this work into new domains such as clinical imaging and protein structure analysis, where high-stakes decisions depend on trustworthy AI systems.
 

A National Network of Expertise  
 

Led by UT Austin, IFML is a deeply collaborative effort that brings together top researchers from institutions including the University of Washington, Stanford University, Santa Fe Institute, Caltech, UC Berkeley, UCLA, Wichita State University, Boston College, and the University of Nevada–Reno.

With this renewed funding and expanded research agenda, IFML is poised to accelerate the development of next-generation AI—ensuring UT Austin’s place as a global leader in foundational machine learning and delivering long-term value for science, industry, and society.