College enrollments are down and tuition revenue has dropped, pushing many universities into the red even as they raise prices. Syracuse University Chancellor J. Michael Haynie said the moment is 'a reminder that the way we have always done things is no longer adequate or enough.'
The strain is already showing on campus. The Hechinger Report noted that efforts to keep colleges afloat can push up meal plan costs while cutting library hours and academic advising staff. That means students are paying more for an experience that is delivering less.
AI has made the problem sharper. Universities are now enrolling students who have spent years using generative AI, and that is changing how they learn. One set of findings cited in the piece says AI adoption raises homework scores by 18% but lowers monthly exam scores by 20% within six months. Another finding says regular LLM usage has been found to reduce neural, linguistic and behavioral performance levels.
A MIT study also described AI as 'upending foundational elements of the MIT education experience,' 'undermining student mastery and confidence,' 'eroding the social contract' between instructors and students, and making it harder to assess student progress. The piece says other institutions are facing similar questions and that no one has it fully figured out.
The academic warning signs are not limited to campuses. The OECD’s Programme for International Student Assessment found that the overall performance of 15-year-olds in OECD countries has declined, with the lowest average levels ever observed in science, reading and mathematics.
That leaves colleges with a hard choice. The old system rewards polished outputs, but AI can produce outputs faster than any human. The market is already telling universities that the real premium is on critical thinking, problem-solving, adaptability, and learning agility. One study cited in the piece found that nearly three quarters of hiring managers, 73%, prized critical thinking/problem-solving, while 65% looked for adaptability/learning agility.
The answer, the piece argues, is to stop pretending students will simply avoid AI and to return to methods that test whether they can explain their reasoning. Cambridge and Oxford have long required regular essays followed by tutor discussions, while French classes préparatoires use weekly oral sessions called colles to test problem-solving and depth of knowledge.
That is the part of higher education that still makes economic sense. Capital rewards clear rules, not credential inflation. If universities want to justify the price they charge, they have to prove they are building human judgment, not outsourcing it to the software layer. The taxpayer and the family footing the bill deserve more than a costly system that grades output while ignoring whether students can actually think.
