What do the physics of beatboxing have to do with artificial intelligence? Well, at the University of Southern California it's all part of a blueprint for keeping the technology human-centered.Shri Narayanan, a professor of engineering, psychology, and linguistics, runs USC’s Signal Analysis and Interpretation Lab. At a recent visit, he showed “Marketplace Tech” real-time MRI videos of beatboxers mapped by machine learning algorithms. “Can you believe that the human vocal instrument can provide this, make this sound?” he said. “So, what you’re seeing here is a cross section of a person’s head. You can see that there’s the tongue, the lips, the airway trachea, and this is all machine learning.”Those models aren't just for tracking musical performers. Narayanan's core research is in behavioral signal processing using AI to analyze physical patterns in humans to tackle profound clinical challenges — from mapping neurodevelopment in autism, to identifying early biomarkers for depression or suicidal ideation.He described it as, “words kind of tell something about my intent, my emotions, the context, who I am. Machines try to do that, behavioral signal processing essentially looks at that. But more than that, it's interested in what happens if there are changes in these systems to communicate.”Shri Narayanan, a USC professor of engineering, linguistics, and psychology , at his Signal Analysis and Interpretation Laboratory.Jesús Alvarado/MarketplaceThis isn’t the buzzy AI of chatbots or coding agents that have been all the rage, but research like this is at the core of a $200 million bet USC is making about what a university can still offer in the age of AI.The new Mark and Mary Stevens School of Computing and Artificial Intelligence is pushing to embed the technology across the campus. USC will launch a new AI major this fall, as well as minors, designed for non-STEM students to learn to apply AI to disciplines like history and art.Gaurav Sukhatme, incoming director of the new school, said higher education has to evolve constantly to stay ahead of the technology.“We have a very project-driven curriculum, which allows us to refresh the projects that the students do as part of their degree programs, and we keep changing those projects and updating them,” Sukhatme explained. “So, as new topics are born and created, new elective courses spring up much faster than in a more traditional curriculum.”But innovation in this AI era has largely been driven by industry players with deep enough pockets to fund the large-scale computing power today’s models run on. So what can academia offer? “One of the things [the] university is really good at is posing questions not about what's happening today, but about what might happen a decade from now, five years from now, 20 years from now, which may not always be tied to having access to the latest compute, but it is having the freedom to think about entirely new ways of doing things,” he said.As institutions like USC rush to bend every discipline towards AI, Sukhatme said the ethics of these systems must remain core to the curriculum.Narayanan, whose students have used SAIL research to launch mental health startups, said keeping this advanced tech safe means keeping humans in the loop from the early design phase.“For me, humans are at the center of everything, because this is for humans, by humans,” Narayanan said. “While we know the benefit of saying, ‘okay, this can help support people, provide care for people,’ what is the cost of doing that, especially human cost?”He pointed to privacy being a concern his students are taught when working with large language models in their research. “For example, revealing that you are not a native speaker of a language. There are biases that you are of a certain identity. These things matter,” he said.Narayanan stressed that AI has to have safeguards against these types of biases when working in the space of mental health and its accompanying research. “This is not something just computer scientists can do. That's why I seek out not only scientists, computer scientists, or neuroscientists, clinicians, but also people who think about the social implications of this — even philosophers,” he said. “Some of these questions have been asked by people, humans, for a long time, maybe in a different form, in a different technology era. But I think these paths have been at least touched upon maybe in a new context.”As the technology races ahead, Narayanan said the university has become more critical than ever: It’s a human-scaled lab for studying these big questions.