This big university system is embracing AI. Students and faculty aren’t all on board
Pingnan Lu for NPR
Leaders of the California State University system, the CSU, want it to become the nation’s first artificial intelligence-powered institution of its kind.
It entered into a $17 million no-bid contract with OpenAI last year to provide students, faculty and staff with a new resource: ChatGPT Edu — a version of the popular generative AI chatbot intended for use by educational institutions. The system recently renewed that contract for another $13 million a year for the next three years.
“No other university system in the U.S. or internationally is doing anything like this, not at this scale,” said Mildred García, the CSU’s chancellor, during a February 2025 press conference announcing the partnership.
But in a recent survey, majorities of its students and faculty said they were skeptical of the benefits of AI for education, and they worry about AI’s impacts on job security, creativity and the environment.
Colleges across the country – from Syracuse Universityto Dartmouth College to the University of Minnesota – have inked similar deals with AI companies, but as the largest public four-year system in the U.S., the CSU’s partnership stands out.
As higher ed scrambles to figure out the benefits and harms of AI, the CSU offers an early look at what happens when an administration commits to a technology that its own community isn’t convinced will improve education.
What the university has to gain
In December 2024, university leaders flagged a potential partnership with OpenAI as “a huge branding opp[ortunity],” according to an internal CSU planning document obtained by NPR.
Ed Clark, chief information officer for the CSU’s office of the chancellor, told NPR in an email that “the planning document demonstrates the extent to which the CSU thoughtfully approached selecting a vendor that could support our commitment to innovation, accessibility and academic excellence.”
Clark said the system chose to partner with OpenAI because they offered “the most cost-effective option that could make it even possible to bring AI tools to more than a half a million students, faculty and staff.”
Learning in the age of AI
College students, professors are making their own AI rules. They don’t always agree
A separate document obtained by NPR, dated 2025, shows that the CSU expected questions around its partnership with OpenAI. The document, titled “Potential follow-up questions on ChatGPT Initiative,” advises officials to explain the no-bid contract by saying the deal is “essential for the success of the CSU’s AI strategy.”
The document goes on to say, “After conducting extensive research and evaluating various AI tools and vendors, it was determined that OpenAI is uniquely positioned to meet our needs.”
AI won’t be used to teach classes, and Clark, the CSU spokesperson, told NPR in an email that the technology should supplement learning, not replace it. Both the CSU and OpenAI frame AI adoption as a necessity to prepare students for careers steeped in this technology.
“As they prepare for the workforce, AI literacy is becoming part of career readiness… so the CSU’s role is to help students understand how AI is changing their disciplines and how to use it ethically and responsibly,” Clark said.
Leah Belsky, vice president of education at OpenAI, told NPR they share a responsibility to “help students use these tools well… to harness their full potential and succeed in the AI-driven future of work.”
But Martha Kenney, a professor and science and technology scholar at San Francisco State University, part of the CSU, says some faculty and students reject the idea that AI in higher ed is an inevitability, and that their perspective deserves consideration.
“I think refusing this technology needs to be a position that’s on the table,” Kenney says. She says rejecting this technology on campuses is justified, given generative AI’s environmental impact and the use of copyrighted work to train models. She also questions the educational value of technology like ChatGPT Edu: She says offering a chatbot that allows students to take shortcuts on assignments is “cheating our students out of an education.”
Kenney co-authored a petition that called on the CSU not to renew its contract for ChatGPT Edu.
But Clark says the “online petition does not reflect overall sentiment from within our community.” He says the CSU’s survey shows strong support for AI given that majorities of students and faculty say it has had a positive impact on their learning and work.
Clark also says the CSU chose to renew its agreement with OpenAI after its generative AI advisory committee, which is composed of students, faculty and staff, “unanimously recommended renewing the contract.”
How students, staff and faculty feel about AI
The CSU serves about 470,000 students and, according to the system, it awards nearly half of all bachelor’s degrees in California. Its student body is diverse: Roughly half are Hispanic, more than a quarter of undergraduates are the first in their family to attend college and many students work while they attend school.
Last fall, the system invited students, staff and faculty across all 22 campuses to take a survey on their views around AI. More than 94,000 people responded, and the results show widespread use of generative AI tools, but also significant ambivalence about the technology.
The survey did not ask whether students, faculty and staff agreed with the system’s decision to spend millions on a contract with OpenAI.
Among the survey’s topline findings:
- More than half of students and around 6-in-10 faculty and staff report using AI regularly for coursework and tasks related to their jobs.
- Roughly 65% of students and 59% of faculty said they were skeptical AI was benefitting education overall.
- 80% of students said they wouldn’t be comfortable turning in AI-generated work as their own.
- About 64% of students said AI has “positively affected” their learning, while about 35% said AI “negatively affected” their learning.
- About 56% of faculty reported AI had positively affected their teaching, research and administrative experience. But in a separate survey question, 52% reported a negative effect.
Roughly 84% of students said they used ChatGPT. About a quarter of them said they used the version provided by CSU and the vast majority said they used the free version.
Large majorities of students and faculty also worry about AI’s impact on creativity (83% of students, 82% of faculty), job security (82% of students, 78% of faculty) and the environment (80% of students, 84% of faculty).
The survey had some limitations, says David Goldberg, an associate professor at San Diego State University, part of the CSU, and one of the survey authors.
“The findings are based on the people who did respond. We don’t know the opinions of the people who didn’t,” Goldberg explains. Still, he says the responses are a “pretty good representation across different fields of study and across different demographics.”
Goldberg says the survey illustrates a tremendous amount of nuance in opinion across all groups.
“Even within one student, you can be using the tool a lot, see real advantages, and at the same time see these negatives,” he says.
What students stand to gain – and lose – from AI on campus
Sejal Daterao is one of those students with complicated feelings.
Learning in the age of AI
Teachers are using software to see if students used AI. What happens when it’s wrong?
Daterao, 30, says she enrolled in the information systems master’s program at California State University, Long Beach, to learn how to use AI more efficiently.
As a student, she says she uses ChatGPT Edu and other AI tools to conduct research, summarize text and video lectures, and create quizzes targeted to the subjects she’s studying.
And she says she’s grateful the CSU provides access to ChatGPT Edu — which includes features not available on the free version of ChatGPT. As a grad student, she says it would be hard for her to foot the bill for a premium subscription.
“Helping students use such technologies firsthand is really a good thing, honestly,” Daterao says.
But she doesn’t describe herself as pro-AI.
She’s frustrated by the occasional false information AI chatbots generate and by tech companies’ use of creative work to train AI models without providing credit and compensation to artists.
“It has a lot of bad sides, and a lot of good sides,” Daterao says. “If you are smart, if you are being ethical, you can use the good sides in a really amazing way.”
But another student, H, doesn’t see many redeeming qualities to AI. She’s in her fourth-year studying computer science at San José State University, also part of the CSU, and she asked that NPR refer to her by only her first initial because she’s actively applying to tech jobs and doesn’t want her opinions on AI to impact her employment prospects.
H says she was annoyed when she noticed that her classmates were using AI to write assignments for them.
“It was pissing me off, which is why I completely avoided using it at first,” she says.
Eventually, H says she began using AI chatbots for “menial tasks” like writing emails, and then to help her on coding assignments.
But she noticed that when she used AI to code, “I found that I was using it more as a crutch instead of actually helping. So that was one of the telltale signs that I should stop using it.’
She says her resistance to AI has only deepened as she’s learned about the environmental impacts of data centers.
H understands that the CSU is under pressure to adapt to an emerging technology. But she says she’s “a little disappointed that they accepted it with open arms immediately.”
And H worries that pushing AI use in coursework will prevent students from learning the foundational skills they need to be successful.
“It’s something I’ve struggled with,” she says. “Trying to use it to learn basics kind of led to just not learning basics, but using it to avoid putting in effort.”
Educators “can’t ignore the technology”
Zach Justus, a communications professor and director of faculty development at California State University, Chico, part of the CSU, has spent the last few years encouraging faculty to adapt their teaching to the AI age — which means experimenting with the technology to figure out what it can and can’t do.
He says he’s excited about the innovative ways some faculty members are leveraging and allowing students to use AI. But he says adaptation, in certain circumstances, also includes redesigning coursework to prevent AI-use.
“The most important thing that we tell faculty is that they cannot ignore the technology,” Justus says. “If we ignore it, we are not doing our jobs.”
He says he understands the critiques of the university system’s contract with OpenAI – including the argument that the system shouldn’t spend millions on an AI chatbot when it’s facing budget cuts. But he says it’s also a problem if only some students can afford the premium versions of this software.
Without the system providing these tools to students, Justus says, “You’re just systematically advantaging students with more financial resources, and that’s crappy.”
English professor Jennifer Trainor isn’t ignoring AI, but she’s not a fan of it either.
Trainor, who teaches at San Francisco State University, says her approach is to teach students about AI and the ethical questions it raises. She says she safeguards the learning process from AI by requiring students to brainstorm and draft by hand during class time. And she allows students to use AI to edit their writing, but she requires them to reflect critically on the changes it made.
“I am really trying to get them to do their own writing and thinking,” Trainor says. “And I’m also giving them chances to see what happens when they use tools to improve their writing and thinking.”
Some students refuse to engage with AI altogether, Trainor says. She describes it as a “groundswelling of resistance” on the campus.
“They’re ethically opposed to the environmental impacts and the bias and the erasure of their jobs and voices and creativity. [They] don’t like it,” Trainor says.
But it’s clear that, for now, it’s not going anywhere.
This reporting was supported by a grant from the Tarbell Center for AI Journalism and the Omidyar Network’s Reporters in Residence program.
Edited by: Nicole Cohen
Considering ed Clark, chief information officer for the CSU’s office of the chancellor, told NPR in an email that “the planning document demonstrates the extent, it raises some real questions about what happens next.
Reading that a separate document obtained by NPR, dated 2025, shows that the CSU expected questions around its partnership with OpenAI — hard to argue with the logic there.
13 million. The real figures are likely much higher.
Mildred Garc has been pushing this agenda for a while now.
The fact that ed Clark, chief information officer for the CSU’s office of the chancellor, told NPR in an email that “the planning document demonstrates the extent really puts things into perspective.
Hard to look at 13 million and not question the official narrative.
On one hand a separate document obtained by NPR, dated 2025, shows that the CSU expected questions around its partnership with OpenAI. But at the same time leaders of the California State University system, the CSU, want it to become the nation’s first artificial intelligence-powered institution of its kind.
Basically the system chose to partner with OpenAI because they offered “the most cost-effective option that could make it even possible to bring AI. What matters is whether anything changes because of it.
What stands out is ed Clark, chief information officer for the CSU’s office of the chancellor, told NPR in an email that “the planning document demonstrates the extent. That is the part worth paying attention to.
94,000 people — and that is probably just the official count.
94,000 people is hard to ignore, no matter which side you are on.
Syracuse Universityto is in a tough spot here, curious how they navigate it.
On one hand in December 2024, university leaders flagged a potential partnership with OpenAI as “a huge branding opp[ortunity],” according to an internal CSU planning document obtained by NPR. But at the same time ed Clark, chief information officer for the CSU’s office of the chancellor, told NPR in an email that “the planning document demonstrates the extent.
The fact that the system chose to partner with OpenAI because they offered “the most cost-effective option that could make it even possible to bring AI really puts things into perspective.
The detail about in December 2024, university leaders flagged a potential partnership with OpenAI as “a huge branding opp[ortunity],” according to an internal CSU planning document obtained by NPR is something people should sit with.