USC

USC awarded $5.6 million from the National Science Foundation

The award would go towards AI optimization on campus.

[One-sentence description of what this media is: "A photo of a vaccine site on USC campus" or "Gif of dancing banana". Important for accessibility/people who use screen readers.]
The USC Viterbi building on Sept. 14, 2022. (Photo by Tomoki Chien)

USC was awarded $5.6 million for AI research, one of three schools to receive grants from the National Science Foundation through the renewal of the AI Research Institute for Advances in Optimization. USC will create the world’s first Ph.D program for AI and optimization using the funding.

The $20 million in total grants went to USC, the University of California, Berkeley and Georgia Tech which is the lead school in the project.

After five years of first-phase research on the combination of artificial intelligence and optimization, the grant is meant to scale AI for millisecond-level decision-making that is energy-efficient.

Through the award, the university’s AI Research Institute for Advances in Optimization (AI4OPT) team is working to implement AI research into daily functions and help USC gain traction as a leader in AI, according to an article published on Viterbi’s website announcing the award.

USC’s AI4OPT team consists of four Viterbi professors who are research leaders in AI and the intersection of industrial systems. Bistra Dilkina, an Associate Professor of Computer Science at Viterbi and co-principal investigator, or secondary researcher for AI4OPT. The team will create technologies that allow decision-making systems to adapt to new, changing conditions efficiently and effectively.

The schools collaborating are Viterbi, Stevens School of Computing and AI and Marshall School of Business, with team members varying in disciplines from uncertainty modeling to drone development.

In this next phase of AI research through the National Science Foundation, USC will also bring the Seth Bonder summer camp to Los Angeles, according to Viterbi’s announcement. The camp, which is offered at Georgia Tech, is intended for high school students who are enthusiastic about engineering to engage in practical applications in machine learning and human sciences.

AI4OPT’s second phase of research will prioritize four integrated areas, or “thrusts.” Two will aim to advance the science of AI and optimization, and the other two will combine those advances to energy systems and manufacturing, according to an article on AI4OPT’s website.

For Jinyi Ye, a Ph.D computer science student, this award has changed how student researchers and professors alike interact with AI.

“We (student researchers) are at the frontier of a lot of new technologies like AI agents,” Ye said. “Professors are using these agents to write funding proposals, and students are using these agents to write papers. But on the downside is that we are delegating our agency to the agent.”

NSF has also developed the world’s first Ph.D certification in AI and Optimization, the Operations Research and Artificial Intelligence (ORAI) specialization, with funding from the National Science Foundation Research Traineeship grant, according to an article published by Viterbi. In this program, students will learn about the processes used by the AI4OPT team, specifically how to critically research the intersection of AI and optimization, according to the ORAI program website.

Additionally, the students will be trained in responsibility, ethics and the use of incomplete, biased or noisy data in AI. As written on the program website, this Ph.D certificate, one at the forefront of AI research worldwide, is open to all students, regardless of country of origin and citizenship, though scholarship opportunities are only eligible for U.S. residents.

The award comes almost one year after the university announced that ChatGPT Edu would be available for university use in an AI Summit that took place in December of last year. As of January 2026, all students, faculty and staff have access to OpenAI’s program and models like GPT-5.

Camille Culbertson contributed research to this story.

Edited by: Leo Li & Jadon Frank