The university must not become a supply chain for AI

Published 11 June 2026 · tech

Is AI going to be the answer to everything? That seems to be the proposition of many commencement speakers at US universities this graduation season

Is AI going to be the answer to everything? That seems to be the proposition of many commencement speakers at US universities this graduation season. Graduating students, however, have not always welcomed the message. At ceremony after ceremony, they have responded with boos and jeers. Their reaction is not hard to understand. Students are leaving university at a time when AI is being not only as a tool they must learn to use, but as a force that may transform the labour market they are about to enter. Yet the challenge goes beyond jobs. Universities are also being encouraged to remake themselves around AI, adopting it as a solution to budget pressures, administrative burdens and the demands of employers. This is where the real danger lies. In the “age of AI”, universities risk becoming victims of their own uncritical embrace of the technology, especially at a time of deep financial strain. Industry stakeholders have strongly encouraged them to move in this direction. A recent paper by Cisco, the US networking and technology giant, claimed that “forward-thinking institutions view AI as a solution to their resource constraints”, adding that “AI can automate routine tasks, improve student services and help universities operate more efficiently”. It also insisted that universities must embrace their “role as supply chains for AI-related skills”, explaining that “students entering the workforce expect AI integration, and employers increasingly demand AI literacy”. This is a revealing way to talk about higher education. Universities are being told to see AI not only as a tool, but as an organising principle: their students imagined as future workers in need of AI literacy, their staff encouraged to streamline their labour, their institutions remade to be more efficient, more automated and more closely aligned with the labour market. Several have accepted this logic. The University of Minnesota, Dartmouth College and Syracuse University have all signed deals with AI companies.

In 2025, California State University (CSU) reached a $17m deal with OpenAI to provide the company’s “education-focused” chatbot to its more than half a million students and faculty. Surveys show that many CSU faculty and students are not convinced by “AI’s dazzling promises”. Yet that scepticism did not prevent the agreement from being treated as a landmark. For OpenAI, signing up the largest public university system in the United States was proof of concept that AI could be embedded across higher education at scale. For CSU, it was a “huge branding opportunity”, since no other university in the world had adopted AI at this scale. The financial logic is harder to follow. Despite facing roughly $144m in budget cuts, CSU last month renewed the deal on costlier terms, committing to $13m a year over three years, about $39m in total, deepening its bet on AI at the very moment it was cutting elsewhere. What happens when universities begin to treat more of their work as something to be automated, outsourced or made cheaper through AI? We saw a small but telling example at the graduation ceremony at Glendale Community College (GCC) in Arizona. The college’s leadership used an AI system to read the names of graduating students as they received their diplomas. The system was unable to match the correct names to the students walking across the stage, and the name on the jumbotron did not match the student receiving the diploma. GCC President Tiffany Hernandez was booed by graduating students and their families when she explained what was happening. “Yep, yep. So that is a lesson learned for us,” she said. One graduating student told media outlets that Hernandez’s apology “didn’t feel sincere and it kinda felt like they didn’t care”, adding: “I would have liked a little more thought to have gone into it rather than pushing something as simple as reading some names off to an AI device.” The problem becomes more serious still when AI moves from administration into teaching and assessment.