When higher education talks about teaching AI, the conversation usually jumps straight to new majors and shiny technical programs. 

In Arizona, it starts with how students think and make decisions, and with where AI fits into the work they’ll eventually be trusted to do.

Grand Canyon University President Brian Mueller has been watching the frenzy from the sidelines, and he’s not convinced the fastest schools win anything meaningful.

“There’s this race to establish technical AI degrees as soon as possible,” he says. “We hope to use AI to supplement learning, not replace it.”

That distinction — AI as a thinking partner, not a shortcut — has become the backbone of Arizona’s approach.

DEEPER DIVE: Arizona, California, Nevada reach historic agreement to protect Colorado River

INDUSTRY INSIGHTS: Want more news like this? Get our free newsletter here

A philosophy that predates the hype

At the University of Phoenix, the shift didn’t begin with ChatGPT. Leo Goncalves, vice president of the workforce solutions group, says the university was already deep into machine‑learning work in 2021, long before the hype.

They’d lived through the “hallucination era of GPT‑3.5.” They’d already learned not to trust the tools blindly.

“Today’s workers are seeking a structured approach,” Goncalves says. “They want to integrate AI into their jobs without losing judgment.”

Universities must meet that need if they want their graduates to be taken seriously, he adds.

Where ethics actually lives

That same philosophy shows up at the University of Arizona’s College of Information Science — not as a standalone ethics course, but as a thread running throughout all classes.

Diana Daly, an associate professor and associate dean of graduate academic affairs, says the college’s roots matter. 

“The first iteration of what our college is now was a library science department,” she says. 

“Both of those are very focused on humanity, on tool use, and the arts. So, I would say there are pretty human‑centered roots that we have.”

Ethics has been there from the beginning, she says. The University of Arizona established the Office of Responsible Artificial Intelligence, which aligns AI integration with long-term sustainability goals and ethical principles. When Daly teaches students to identify bias in AI systems, she doesn’t start with code. She starts with behavior.

“You can test the systems just as a user,” she says. One of her go-to examples is translation prompts — the kind that reveal gender bias instantly. 

“You’ll get the system to say that the doctor must be a male, that a nurse must be a female,” she says. “Those pattern‑based behaviors, things you can pick up just by dialoguing with these systems.”

The college’s new AI and society minor, launching this fall, formalizes what faculty have already been doing: teaching students to interrogate systems, not worship them.

AI that teaches students to think, not copy

GCU’s AI tutoring ecosystem — Isaac for math, Mira for health sciences, Bloom for education exam prep, Aiden for scenario‑based learning — was built with the same guardrails.

“It was designed to embed concepts, guide problem‑solving, and mimic real‑world decision‑making,” Mueller says.

These aren’t answer‑generators. They’re mentors. Students must disclose when they use AI, and they’re graded on how well they manage it. Ethics isn’t a chapter; it’s an expectation.

Across more than 90 AI‑integrated courses, students learn how to use AI inside their field, not as a generic tool. Healthcare students examine cases through a data‑governance lens. Education majors practice classroom scenarios. Business students earn Microsoft credentials tied to AI‑driven decision‑making.

The goal isn’t to turn everyone into machine‑learning engineers. It’s to graduate people who know how to think.

Mueller and Goncalves keep circling back to the same point: AI can widen or close opportunity gaps. The outcome “depends on the design.”

“AI is not inherently good or bad,” Goncalves says. “It’s a force multiplier for whatever systems are already in existence.”

Arizona’s human‑centered approach doesn’t start in college. It’s already reshaping kindergarten to eighth-grade classrooms.

At Novatio, an East Valley school, students use AI daily to personalize learning. Core academics happen in a focused two‑hour morning. The rest of the day is spent building businesses, developing apps, collaborating — the work no machine can do.

“Our teachers don’t end up traded for AI — it frees them to help us prepare our kids,” says Karissa Ham, Novatio’s head of school. She calls it “AI as a coach, not a ghostwriter.”

Unbound Academy head Michael Goto sees the same dynamic. Schools that ignore AI are preparing students for a world that no longer exists, he says.

“These students are not trained how to ask what they need and how to prompt,” he says. “By not having the right AI implementation, they self‑teach to cheat and shortcut.”

A Tucson teacher recently told him that when high schoolers were asked about AI, they didn’t talk about cheating. They said they wished their schools trusted them enough to train them.

“Students without tools to use it create their own purpose of AI,” Goto says. “Those can vary from basic academic shortcuts to outright cheating.”

A statewide model built on judgment

Across institutions, the throughline is the same: Arizona is betting on judgment, not speed.

GCU’s Purposeful Pedagogy model helps instructors integrate AI transparently and responsibly. The University of Phoenix built a culture where validation and ethical use matter more than novelty. 

At the University of Arizona, Daly’s students learn to interrogate systems — sometimes with nothing more than a translation prompt and a sharp eye.

“We are definitely making some of that curriculum as we go,” Daly says. But the human‑centered roots are the constant.

For example, students train their own simple machine-learning classifiers with Google Teachable Machine, then reflect critically on how classification systems work and fail. The assignment asks:

“Describe your experience training a machine in meaningful classification … What does this micro-version of machine learning demonstrate about this manner of processing information? What are potential impacts… in society?” Daly explains.

She cited another major exercise that asks students to reverse-engineer recommendation and personalization systems such as TikTok, YouTube, Amazon, Google Search, Spotify or library discovery systems.

At the University of Arizona, students are also specifically asked to analyze: “Signals: What inputs does the system seem to use? (clicks, watch time, skips, searches, dwell time, saves, likes, location, device, time of day, etc.)” And “Inferences: What does it appear to infer about the user — interests, intent, proficiency, mood, sensitive proxies, etc.” 

Mueller says it simply.

“The schools that rise above will be those using AI to revolutionize the way students learn and the way they’re supported.”

Arizona’s graduates may stand out not for technical specialization, but for judgment, leadership and integrity.

“They are going to know how to think, how to interrogate what AI produces, and how to apply it in a way that matters,” Mueller says.