The New Normal, Episode 5: President Mark Milliron

Five takeaways from the president of a university built for military and working students on keeping AI tied to mission, and why it has stopped short of agentic systems.

National University was founded in 1971 by a Navy captain who watched service members coming out of Vietnam try to use their GI Bill inside institutions that were never built for them. Today, Dr. Mark Milliron leads National University as president and CEO, after prior stints at the Bill & Melinda Gates Foundation and Western Governors University.

Milliron is the fifth guest on The New Normal, OutRival’s interview series with leaders deploying AI in production. The conversation, recorded on National University’s San Diego campus, covers how the university keeps AI tied to its mission, and why it has stopped short of agentic systems.

Here are the five biggest takeaways from his conversation with OutRival co-founder Ruben Harris.

1. Design for ANDers, and treat it as a feature

Milliron refers to the students NU serves as “ANDers” because they are both students, and something else: students and parents, students and employees, students and deployed.

“If you ask them to choose between their ands, you’re gonna lose,” he says.

At NU, they don’t treat that “and” as a complication; as Milliron puts it, it’s a feature, not a bug.

An example of this is the four-week undergraduate term at NU. Five courses across sixteen weeks is too high a load for a working parent, and if the demands of work or family are too high, the student won’t just drop one class: they’ll drop all of them.

NU’s solution to this was to go down to one class a month. A student who needs to take a month off can, without having to drop their classes entirely. It remains NU’s most popular undergraduate format.

2. The thing breaking is navigation, not academics

When low-income and first-generation students fall off higher education pathways, they usually leave in good academic standing. They didn’t fail the coursework; instead, they’re failing to navigate the institution and balance it against the rest of their lives.

NU’s leadership academy runs an exercise that looks at this same problem.

Twenty-five rising leaders, some new and some thirty-year veterans, are asked to list every resource the university has for a student facing an academic, psychosocial, or logistical crisis. By the end, they have quite a long list.

Then Milliron asks how many just learned about something NU offers that they did not know existed. Every hand goes up.

This gap is exactly what Wingman is built for. The initiative, named for NU’s military identity and its wings-up hand sign, rolled out over the past year to put an AI wingman alongside every employee and, by extension, every student.

A faculty member or advisor gets prompted with what else might apply: a veterans center resource, a library service, or a single-parent support group that might previously have been missed.

3. No chief AI officer, and 150 people leading initiatives

NU has no chief AI officer. Instead, they have what Milliron calls a “whole-of-institution approach.” Their AI council is split into four rings, each with its own leaders: student experience at the center, academics and operations on the next ring out, then infrastructure, then governance.

More than 150 people now lead AI initiatives in advising, enrollment, the library, and the academic support centers, each tied to strategy, a budget, and instrumentation. Their method is try, test, tune: make the bet, instrument it, then do more of what works and stop what doesn’t.

Subscribe to The New Normal here.

4. Not agentic yet, and that’s the point

When asked about what he wants to do with AI that NU hasn’t done yet, Milliron starts with a caveat. He says that anyone claiming to have AI figured out is lying, and what he wants is to be around people getting real about what works instead of just selling features.

He then sorts the work into levels. Automations handle specific tasks. Chatbots are wildly successful for quick questions. Wingman is working in advising and enrollment. But true agentic AI, systems operating on their own, is not something NU has deployed just yet.

That hold is deliberate. Agentic systems will require what he calls a stretch of training and trust, the period where an institution learns what the technology does unsupervised.

“Our students are precious cargo,” he says.

5. Novel to normal

Milliron’s first vice presidency involved getting an entire campus onto email and microcomputers, training faculty who wanted nothing to do with either. He has watched each new technology go from novel to normal, and he runs the same playbook each time, AI included: give innovators support and ask them for data, then use that evidence to bring in the early majority.

The late majority, though, needs more than evidence. They need playbooks and real value, or you get the AI version of death by PowerPoint: deployed everywhere, useful nowhere.

Behind all of this is his read on who he is up against. “My competitors aren’t institutions,” he says. “My competition’s poverty.”

Listen to the episode

Check out the full episode to hear:

• Why Milliron thinks every graduate professor in America has been working with agents for hundreds of years, and what they were called
• RAISE5, the five-level framework NU faculty built to code how much AI is allowed on any given assignment
• The Datsun 280Z that taught Milliron why institutions should be in “better-with” conversations instead of “better-than” ones
• Milliron on what higher ed owes students beyond AI literacy: “Not educating them in durable skills, like, you’re not doing your job”