Public universities face a demographic cliff of fewer college-age students that threatens enrollment, budgets, and program survival, according to a new report from the Manhattan Institute published in 2026. The report argues that university boards must navigate this crisis while simultaneously managing artificial intelligence's disruptive impact on teaching, research, and student life. The solution proposed: adopt Stanford scholar Niall Ferguson's "cloister and starship" model, where students master traditional learning in AI-free environments before engaging with advanced technology.

The report documents troubling patterns in how college students currently use AI and spend their time. An August 2025 survey of 1,047 American college students found that 85% had used generative AI for schoolwork in the prior academic year, with a quarter admitting their main use was having AI complete assignments without student input. Yet 97% said they wanted their institutions to protect academic integrity from AI, and only 27% of AI users thought it helped them "improve their thinking." Beyond academics, 72% of current teenagers have tried using AI for companionship, with more than half as regular users. The typical American college student now works only 15 hours per week on academics, seldom reads more than 10 pages for a class, and rarely writes papers longer than 10 pages. An MIT study titled "Your Brain on ChatGPT" found that chatbot use reduces students' brain activity and weakens neural connectivity.

The report's author, Brian J. A. Boyd, warns that rushing to integrate campus-wide AI products like OpenAI's ChatGPT Edu risks worsening isolation and disengagement trends. He points to a San Francisco State University information session where an OpenAI employee described becoming "friends" with the chatbot and said students could decide "what kind of relationship they want" with it—including a forthcoming "adult mode." According to Boyd, the fundamental issue is that "learning results from what the student does and thinks", and AI used poorly allows students to coast without work rather than amplify their existing skills. He compares the risk to nineteenth-century German forest management, where experts used new science to plant fast-growing monocrops that produced abundant lumber initially but led to "Waldsterben"—forest death—in the second generation due to depleted ecosystems.

Boyd's recommendations center on strengthening core curricula while preparing students for an AI-driven economy. He calls for implementing the National Association of Scholars' model General Education Act, which requires a 42-credit core taught seminar-style to no more than 18 students with all digital technology banned. These "cloister" courses would focus on reading physical books, handwriting essays, and discussing timeless questions about what it means to be human. Only after mastering traditional thinking should students enter the "starship" through interdisciplinary programs in "Civic Entrepreneurship & Tool AI" that combine apprenticeship-style job preparation with humanistic values. The report urges boards to create venture funds supporting graduates who launch in-state AI startups addressing local needs, building state-controlled "fleets" rather than leaving citizens as passive passengers on starships controlled by distant tech giants. For research, Boyd recommends treating AI as "co-intelligence" that complements human expertise rather than replaces it, with mandatory retention of AI chat transcripts for replication studies and peer-review reforms to prevent journals from being overwhelmed by "confident-sounding garbage." The stakes couldn't be higher: Anthropic's CEO predicts 50% of entry-level white-collar jobs will vanish by 2030, the year today's freshmen graduate, raising fundamental questions about whether citizens can meaningfully participate in self-government when AI systems outperform humans at most economically valuable work.