The largest four-year public university system in the United States has bet its future on generative AI, licensing OpenAI's tools for all 460,000-plus students, faculty and staff across its 22 campuses, according to the system's top academic technology official.
Dr Leslie Lange Kennedy, Assistant Vice Chancellor for Academic Technology Services at the California State University (CSU) Office of the Chancellor, spoke to Samaa Digital in an exclusive interview on the sidelines of the Ai4 2026 conference in Las Vegas, where she also co-leads CSU's Generative AI initiative, the body steering the system's academic response to AI adoption.
Asked to lay out the scale of the system she oversees, Dr Kennedy said CSU's size is what makes its AI experiment consequential well beyond California's borders. "One in 10 of our graduates are employed in the state of California and one in 20 across the United States," she said, adding that CSU's four-and-a-half-million alumni "represent more than the population of 23 states."
The system's 22 campuses cover engineering, business, nursing and healthcare, alongside liberal arts and social sciences, and offer both master's and doctoral programmes.
"It's quite a regional, comprehensive, instructional environment," she said, describing an institution built to serve a workforce pipeline rather than a narrow academic niche.
OpenAI For All
Regarding the decision to license OpenAI's tools system-wide, covering every student, faculty member and staff employee rather than a pilot group, Dr Kennedy said the rationale was twofold: teaching the CSU community to use the technology responsibly, and preparing students for a labour market already being reshaped by it. She described the scale of change now underway inside classrooms as considerable. "The transformation of instruction is significant right now," she said, noting that the initiative began roughly a year ago and remains a work in progress. "We're on that journey, because we started a year ago, and we're still learning and exploring the best ways to do that."
Training Trainers
On the question of how CSU is preparing its instructors, a system employing roughly 29,000 faculty members, Dr Kennedy said training has proven harder than distributing software licences. The university has set up several faculty development programmes and attached funding to them, she said, because most instructors will not absorb the additional workload without support. "A lot of times the issues are they don't have enough time or capacity, so they want to be paid extra. So we don't pay all of them," she said. Instead, CSU runs a grant process that faculty can apply to individually.
Asked whether adoption has been even across the faculty, Dr Kennedy said it has followed a predictable pattern. "Our faculty, just like anywhere, represent the spectrum of adoption, which is a bell curve," she said. "The first part of a bell curve is usually the early adopter group... they're usually jumping right in .... and eventually we hope to train everyone."
She said the grants also fund faculty-led research into how AI should reshape instruction within individual disciplines, since the needs of a writing instructor and a computer science instructor diverge sharply in practice. "What a writing professor does versus a computer science professor can be very different in their assessments and how they present the material, how they ask or allow the students to use AI as well," she said.
Reworking Exams and Grading
Turning to assessment, which is one of the more contentious fronts in higher education's response to generative AI, Dr Kennedy was asked directly how much exams and grading have been affected. Her answer was unambiguous: "Significantly."
She said faculty pursuing genuine measures of student learning can no longer rely on assessment formats used for decades, because "essays are not necessarily as valid as they would have been before". Although, she added a caveat, noting that in some cases essays "might not have been" fully valid even prior to AI's arrival.
Multiple-choice testing, she said, is similarly exposed: "Even multiple-choice questions can be easily invalidated." In response, she said CSU is experimenting with project-based and experiential learning formats across departments to see which hold up under the new conditions.
AI Workforce Acceleration Board
Asked about the widening gap between what universities teach and what employers actually need, and how CSU plans to close that gap, Dr Kennedy pointed to a body the system created specifically to keep both sides talking: an AI Workforce Acceleration Board, which includes representatives from AI companies and meets quarterly.
The board's purpose, she said, is to generate "potential solutions of what we should be doing for our youth students and what the industry is expecting." She described it as a two-way channel: CSU hears directly what employers expect from graduates, while companies hear what students and faculty are actually experiencing on campus.
Evergreen Skills
On what CSU is doing beyond that board, Dr Kennedy said the system's broader goal is to equip students with skills that outlast any single technology cycle, given how much of the job market AI is expected to redraw before today's students graduate.
"We're doing as much as we can to help students be prepared to be able to have those analytical, critical thinking, problem-solving skills that they're going to need... for jobs we don't even know exist yet," she said.
That uncertainty, in her assessment, rules out narrow technical training as a strategy. "You can't necessarily train for one thing. You've got to be able to learn and grow and adapt," she said, adding that the World Economic Forum's top 10 workplace skills are now being built directly into CSU's training curriculum.
Asked for a final comment as the interview closed, Dr Kennedy said she considered her presence at the conference a chance to represent the scale of the challenge facing public higher education.
