Reporting on the connections between education and work

ASU Makes a Big Play with OpenAI for Students. Can Others Keep Up?

The mega-university joins the University of Michigan system as the only ones to ink major AI partnerships thus far, raising questions about what kinds of colleges—and which students—are going to shape the emerging tech.

Arizona State University announced its new partnership with OpenAI just a week ago, but there’s already an arm’s-length list of questions about the implications for higher education and workforce development.

Just a few: How exactly will the partnership better prepare students for the jobs of the future? Will such collaborations force institutions to boost AI literacy among their faculty and staff members, who have to date received little training on generative AI?

And, chief among them for advocates of economic mobility, can community colleges and other open access institutions afford to give students the same access to generative AI tools enjoyed by those attending selective and better-financed institutions? And if not, will government or private funders fill the gap?

Partnerships like ASU and OpenAI’s may defray the cost of providing students with access to AI tools, but they are expensive nonetheless—so expensive that it may be impossible for broad-access institutions to scale.

ASU, with 114K students on-campus and online, has built itself around the goal of educating a broad swath of Arizonans and more. And 33% of its undergraduates are eligible to receive federal Pell grants for low-income students. But the university nevertheless has selective admissions criteria and resources, including a top 100 endowment, that are unheard of at most institutions serving large numbers of low-income students.

The only other university, thus far, to announce a similar AI partnership is the University of Michigan system.

The stakes aren’t just about which students get immediate access to AI tools, says Philipp Schmidt at Axim Collaborative, but what kinds of students the tech and teaching approaches are designed for in the first place. In announcing the partnership, Brad Lightcap, OpenAI’s Chief Operating Officer, said the company plans to learn from ASU as it plans a broader expansion into higher education.  

Because of the current high costs of AI, open access institutions might miss their chance to shape the pedagogy around it, says Schmidt, vice president of technology innovation at Axim, a nonprofit founded by Harvard University and the Massachusetts Institute of Technology to address equity gaps in higher education.

“Institutions that have a commitment to serving a broader range of students are particularly important in this conversation,” he says. “It’s not something that should just involve the elite institutions. It should really involve the institutions that are serving the majority of students in the U.S.”

A Few Knowns, Lots of Unknowns

The details: The partnership with ASU is a first for OpenAI, which released its revolutionary generative AI tool ChatGPT in November 2022. ASU will use the partnership to create personalized tutors, develop AI avatars that function as “creative buddies” for students, and scale courses in prompt engineering, according to CNBC.

In February, ASU also will begin accepting applications from faculty and staff for using ChatGPT-4 to enhance student success, foster creative research, or streamline administrative tasks. ASU employees and students working on approved projects then will receive free access to the tool, which costs users $20 a month. The Chronicle of Higher Education reports that ASU hasn’t capped the number of projects it will approve.

Last year, the University of Michigan system and Microsoft formed a similar partnership that provides a custom AI platform for all students on four campuses.

The why: A July 2023 report by McKinsey Global Institute puts a fine point on ASU and Michigan’s motivation for forging such relationships: generative AI is speeding the automation of tasks currently performed by human coders, customer service workers, product designers, and dozens of other job categories. The report estimates that 30% of hours worked today could be automated by 2030, with generative AI contributing to a big portion of that.

The ASU announcement makes one thing clear, many observers say—colleges and universities can’t afford to ignore generative AI, and need to be figuring out ways to leverage it to better meet students’ needs. That includes focusing on the durable skills employers say are often lacking in workers, such as critical thinking skills, creativity, and the ability to collaborate.

“How can we take the good parts of AI and actually create customized learning profiles, create customized pathways and have personalized tutoring for learners, all while equipping them with the skills and competencies they need to be able to work with AI in the future?” asks Lisa McIntyre-Hite, executive vice president and chief operating officer of the Competency-Based Education Network. 

“We can’t even predict what that’s going to look like yet.” 

Expanding the Fold

Figuring out what that’s going to look like will require trial and error. And there’s an urgent need to make sure selective institutions aren’t the only ones running those experiments, McIntyre-Hite and others say.

That’s where private funders can have a tremendous impact, says Amanda Bickerstaff, co-founder and CEO of AI for Education. 

“A lot of funders are sitting on the sidelines trying to figure out what’s going to happen next,” she says. “Providing access to students who would not traditionally have access to these best models could be a beautiful way to start building their capacity.” 

Bickerstaff says it’s also important for higher education institutions to understand that we’ve only just begun to experiment with AI in the classroom. People tend to overestimate the short-term impact of technology but underestimate the long-term implications.

It probably will take three to five years before the tools impact teaching so significantly that faculty can shift to teaching durable skills. As more faculty and staff become literate in AI, and as the tools themselves improve, “the opportunities will be massive,” she says.

“But if we’re not prepared for those opportunities, then we won’t see that shift.”

Parting Thought: The ASU and OpenAI initiative has its skeptics, however, particularly among those who see generative AI as a threat to quality higher education rather than an enhancement. 

John Warner, who comments frequently about composition pedagogy, has been particularly critical of ASU’s plan to use ChatGPT-4 for freshman English composition courses. In a column for Inside Higher Ed, Warner wrote that the partnership advances ASU’s goal “to reduce or eliminate the cost of human labor as it relates to the teaching of general education courses such as first-year writing.”

In a post on X, Warner said that ASU faculty, “even those not directly affected, should resist with everything they have.”

Download the Work Shift Guide to Understanding New Collar Apprenticeships