USC leads new phase of national AI optimization institute
USC researchers will help lead the next phase of a national artificial intelligence institute developing tools to improve decision-making across power grids, manufacturing, and global supply chains.
Researchers at the University of Southern California are set to take on a significantly larger role in a national effort to apply artificial intelligence to complex logistical problems. The U.S. National Science Foundation has awarded $20 million to renew the Artificial Intelligence Institute for Advances in Optimization, known as AI4OPT, through 2031. This renewed funding allows the institute to deepen its research into how AI can navigate uncertainty and make rapid decisions in systems where time and resources are scarce.
The initiative is led by Georgia Tech in partnership with USC and the University of California, Berkeley. The core goal is to develop intelligent systems that can respond quickly to changing conditions. These tools are designed to help utilities balance increasingly complex power grids, allow manufacturers to adjust production lines efficiently, and enable supply chain operators to adapt to shortages and disruptions more effectively than traditional methods.
Increased funding expands USC research team
USC will receive $5.6 million for this new phase, which is more than double the funding allocated to the university during the institute’s first cycle. This financial increase supports a significant expansion of the university’s involvement, growing the research team from two faculty members to five. The interdisciplinary group includes experts from the USC Viterbi School of Engineering, the USC Mark and Mary Stevens School of Computing and Artificial Intelligence, and the USC Marshall School of Business.
Bistra Dilkina, a co-principal investigator and associate professor of computer science at USC, noted that this renewal allows the team to move basic science toward real-world applications. The expanded team aims to fuse AI with optimization techniques, a combination that can achieve advances neither field could deliver on its own. This approach is intended to address the growing complexity of critical infrastructure, where decision-makers often need to respond faster than traditional algorithms allow.
Bridging the gap between theory and practice
The institute’s work focuses on the science of finding the best course of action when limited by time, resources, or other factors. As power grids and supply chains become more unpredictable, the need for efficient decision-making tools has grown. AI4OPT researchers are working to teach machines to solve these complex optimization problems more efficiently, thereby reducing the risk of failure in critical sectors. This transition from foundational research to large-scale application is a central focus of the five-year grant.
Pascal Van Hentenryck, the lead principal investigator at Georgia Tech, emphasized that combining AI and optimization is changing how society solves challenges in energy, manufacturing, and logistics. By leveraging the strengths of both disciplines, the institute aims to create systems that are not only more accurate but also more adaptable to the dynamic nature of modern infrastructure. This collaborative effort seeks to ensure that technological advancements have a tangible impact on societal stability.
Expanding educational reach through new programs
Beyond research, the institute is committed to expanding educational initiatives across various levels. AI4OPT’s programs span K-12 education, graduate studies, and faculty training. As part of the renewed institute, USC will bring the Seth Bonder summer camp in computational and data science for engineering to high school students in Los Angeles. This effort aims to introduce young learners to the fundamentals of AI and optimization in an engaging, hands-on environment.
The university has also leveraged the institute’s presence to enhance its graduate education offerings. This includes the launch of the first PhD specialization and certificate in AI plus optimization, supported by a National Science Foundation Research Traineeship grant. These educational pathways are designed to build skills in AI, optimization, and real-world applications for students at different career stages, ensuring a steady pipeline of talent for the field.
The collaboration between Georgia Tech, USC, and UC Berkeley represents a significant investment in the intersection of artificial intelligence and operational efficiency. By expanding the team and increasing funding, the institute is positioned to address some of the most pressing logistical challenges of the modern era. The work highlights the growing importance of integrating advanced computational methods into critical infrastructure to maintain reliability and efficiency.
