The great computer science exodus (and where students are going instead)

Something strange happened on the University of California campuses this fall. For the first time since the dot-com crash, the number of computer science registrations fell. It fell system-wide 6% this year after a 3% decline through 2024, according to reporting this past week by the San Francisco Chronicle. Even as a general enrollment for a university increased by 2% nationally – according to January data from the National Student Clearinghouse Research Center – students are relying on traditional CS degrees.

The one exception is UC San Diego – the only UC campus that has one dedicated AI major this fall.

All of this may seem like a temporary problem related to news of fewer CS graduates finding work outside of college. But it is more likely that it is an indicator of the future, one that China is embracing much more enthusiastically. Like MIT Technology Review reported last JulyChinese universities have championed AI literacy, treating AI not as a threat but instead as essential infrastructure. Nearly 60% of Chinese students and teachers now use AI tools several times a day, and schools like Zhejiang University have made AI courses mandatory, while top institutions like Tsinghua have created entirely new interdisciplinary AI colleges. In China, AI fluency is no longer optional; they are table stakes.

American universities are trying to catch up. The past two years have seen the launch of dozens of AI-specific programs. MIT’s major ‘AI and decision making’ is now the second largest major on campus, the school says. As the New York Times reported in December, the University of South Florida has over-enrolled 3,000 students in one new AI and cybersecurity university during the fall semester. The University at Buffalo last summer launched and received a new department ‘AI and Society’ offering seven new specialized bachelor’s degrees more than 200 registrations before swinging open its doors.

The transition has not gone smoothly everywhere. When I spoke with UNC Chapel Hill Chancellor Lee Roberts in October, he described a spectrum: some faculty “hunched over” with AI, others with “their heads in the sand.” Roberts, a former chief financial officer who came from outside academia, pushed hard for AI integration despite faculty resistance. A week earlier, UNC had announced that this would happen merge two schools to create an AI-focused entity – a decision that was met with resistance from faculty. Roberts had also appointed a vice provost specifically for AI. “No one is going to say to students after graduation, ‘Do your best, but if you use AI you’re going to get in trouble,’” Roberts told me. “Yet we have faculty members who are now effectively saying that.”

Parents also play a role in this difficult transition. David Reynaldo, head of the admissions consultancy College Zoom, told the Chronicle that parents who once pushed their children toward CS are now reflexively steering them toward other majors that seem more resistant to AI automation, including mechanical engineering and electrical engineering.

But enrollment numbers suggest students are voting with their feet. According to one questionnaire in October, the nonprofit Computing Research Association (which includes computer science and computer engineering departments at many universities) reported that 62% of respondents said their computer programs saw a decline in undergraduate enrollment this fall. But as AI programs explode, it’s looking less like an exodus of technology and more like a migration. The University of Southern California is launching an AI degree next fall; be like that Columbia University, Pace UniversityAnd New Mexico State Universityamong many others. Students are not abandoning technology; instead, they choose programs that focus on AI.

It’s too early to tell whether this recalibration is permanent or a temporary panic. But it’s certainly a wake-up call for administrators who have struggled for years with how to handle AI in the classroom. The debate over whether to ban ChatGPT is ancient history at this point. The question now is whether American universities can act quickly enough or whether they will continue to debate what to do while students transfer to schools that already have answers.