The foundation of higher education is undergoing a seismic shift as universities integrate artificial intelligence into academia and, furthermore, into the core of the student experience. While administrators champion these tools as essential for “AI fluency,” pushback from neurological researchers and faculty suggests that the cost of adoption may be higher than any tuition increase.
In a landmark move in 2025, the California State University (CSU) system became the nation’s first “AI-powered” university system via a $17 million partnership with OpenAI to provide ChatGPT Edu to all students. However, the announcement arrived alongside a proposed $375 million budget cut to the CSU system. These cuts have already led Sonoma State University to eliminate 23 academic programs, including physics and economics. San Francisco State has similarly suspended graduate programs devoted to Anthropology and Women and Gender Studies.
CSU Chancellor Dr. Mildred García shared her thoughts in an interview with the California State University News on the adoption of artificial intelligence in the CSU system.
“The comprehensive strategy will elevate our students’ educational experience across all fields of study, empower our faculty’s teaching and research, and help provide the highly educated workforce that will drive California’s future AI-driven economy,” said García.
However, the concern among educators is that consistent AI usage is increasingly impacting students’ physiological health. A June 2025 MIT study titled “Your Brain on ChatGPT” discovered a 47% drop in neural connectivity across brain regions associated with memory and critical thinking in students who used AI to draft essays. Researchers found that 83% of heavy AI users could not recall key points of their own “written” work, compared to only 10% of those who wrote unaided.
The study further noted that even when these students returned to “brain-only” writing, their neural connectivity remained weak, and their writing sounded more robotic and monotonous. This data supports a University of Pennsylvania study, which found that Turkish high school students using ChatGPT performed 47% better on math homework but 17% worse on the actual exams.
“The real danger of technology lies in humans adapting their minds to the machine logic,” wrote Joseph Weizenbaum in his 1976 book Computer Power and Human Reason. Weizenbaum is the creator of early chatbot ELIZA and a prominent critic of AI.
Paul Marty, Associate Vice Provost for Academic Innovation at Florida State University, believes the role of the university must change in response to these tools.
“Higher education is not about learning things – it’s about learning how to apply the things you learn, how to define problems and develop solutions to those problems,” Marty said. “That’s how AI is reshaping higher education – by shifting the focus from learning things to lifelong learning, building human-human connections and developing an innovation mindset in our students.”
As students become more dependent on AI, universities are struggling with enforcement. Vanderbilt University recently disabled Turnitin’s AI detector entirely, citing high false-positive rates that unfairly target ESL students. Conversely, the Ohio State University has stopped treating AI as an academic integrity violation, instead requiring “AI Fluency” courses for all undergraduates to meet a job market where AI-related listings have grown 600% since 2010.
Marty noted that academia has survived disruptive technologies before, from the invention of writing to the calculator.
“What history has shown us is that academics are very good at adopting disruptive technologies over time and changing the way they teach,” Marty said. “We have the opportunity here to reshape what we consider important in higher education, so that we’re not just checking boxes for compliance but creating meaningful and engaged learning experiences for our students.”
However, the “AI-first” mentality has led to friction between faculty and students. The New York Times reported that a senior at Northeastern University discovered her professor used ChatGPT to generate lecture slides. The slides were filled with garbled text and errors, even though the syllabus banned students from using the tool.
While critics argue against “uncritical adoption,” researchers at Stanford and Cambridge argue the tool is invaluable to their work. Professor Chelsea Finn at Stanford is currently using AI to create adaptable robotics, such as the “Mobile ALOHA,” which can cook meals for people with disabilities. Other Stanford researchers like Jeff Maers and Aditi Sheshadri are leveraging AI to solve real-world problems in sustainable mining and climate modeling.
To ensure AI does not replace human training, Marty argues that students must focus on what makes them unique.
“What will set humans apart from AI is how humans mediate those answers for other people,” Marty said. “What you know will be far less important than how you know it and whether you can share that knowledge with other people with empathy, inspiration, and critical thinking skills.”
The challenge for the 2025 academic year remains finding the balance between using AI to accelerate research and preventing the cognitive “deskilling” of the next generation.
