5 Lessons from the First Chief AI Officers

Gerry Bayne: Welcome to EDUCAUSE Exchange, where we focus on a single topic from the higher ed IT community and hear insights, perspectives, best practices, and more. Over the past couple of years, colleges and universities have begun to formalize the chief artificial intelligence role in response to generative AI's impact on research, teaching and learning and operations. In recent conversations with two early incumbents, we heard some interesting reflections on what the job entails and what the job means. Their insights point to five lessons about how AI leadership and higher education is taking shape. The first lesson is about the chief AI officer as a transitional role.

Mark Daley: An ultimately successful CAIO can leave the role and there is no successor.

Gerry Bayne: That's Mark Daly. He's the chief artificial intelligence officer for Western University.

Mark Daley: I think this is a moment in time where we have a transformative general purpose technology that is changing our institutions, and so it's helpful to have leadership and some capacity to get through that transition. But when the transition's done, there's going to be AI everywhere. We don't have a chief running water officer, but everybody uses running water, whether you're a sports team or a chemistry lab. But I think a lot of the functions that I'm covering right now will end up being taken by line managers who exist. So the VP academic and provost or our VP international. Just because it's international with AI doesn't mean it's not still international at the core.

Amarda Shehu: There is a sense that if you limit your view to integration of technologies, just building infrastructure, the moment when this becomes a normal technology and we kind of know what we're doing, at that moment, if that's your view of a chief officer, you don't need them anymore. You need a chief information officer.

Gerry Bayne: That's Amarda Shehu. She's the chief artificial intelligence officer for George Mason University.

Amarda Shehu: So those two will come together. And a chief information officer or CIO will be sufficient when this AI technology becomes normal technology. I would say we are maybe more than five years away from that. And the reason I say it is because the technology is evolving fast, the regulations, the policies are not catching up. So it probably won't be a real normal technology for a while.

Gerry Bayne: The second lesson on our list is coordination as the primary mode of AI leadership.

Amarda Shehu: It is an executive position that is very demanding because baked in it is you want to move to animate a very complex system with a complex system, but you don't get to control all the variables. So you have to be very astute to understand with what do you connect? How do you coordinate so you can move the system?

Mark Daley: I don't want any FTEs. I want zero. Everything I do, every project I deliver, I'm going to partner with a line manager who has budget and they're going to have to spend their own budget on it. So I know they're invested, so I know it's sustainable. And from a political standpoint, you know what everyone in the academy hates? New fiefdoms inside the central bureaucracy. So we avoid the political dimension. Everything is a partnership and now everything is sustainable because that budget item isn't one-time funding from the president or the provost. That budget item is align on the registrar's budget, on the Dean of Science budget.

Gerry Bayne: The third item on our list is action without waiting for a standalone AI strategy.

Mark Daley: It's our knee-jerk reaction in the academy. Oh, we need a strategy. And it takes us about a year of consultation to create a strategy. And so by the time we publish our AI strategy, it's comically out of date. It doesn't even look quaint. It looks embarrassing. So you're in a technology domain that is moving really, really quickly is really, really high impact. And trying to generate a strategy gives you a false sense of security. We generate strategies sometimes because we're self-soothing. Hey, if I've got a strategy, I've got it under control. What you actually need for AI transformation is a north star and a compass. So your north star is, where are we going to be in two, five, 10 years? Because it's not going to be exactly where we are now. Our teaching mission is going to change. Our students already expect that. Our research mission is going to change.

Mark Daley: There are already peer-reviewed papers that were end-to-end generated by AI with no human involvement. The compass is what are our core values as an institution? Where are the hard red lines beyond which we are not going to step? What are acceptable uses of AI and what are unacceptable uses of AI? And if you've got that in place, then the strategy becomes the institutional strategy.

Amarda Shehu: Think about, what do I want for the members of my community? And then you can answer the question, what are the tools that we need in order to help members of my community to get to, to get to literacy, to engage with AI thoughtfully? How should AI reshape higher education and how can we do this responsibly and make sure that we bring everything together, the academic, the technical, the governance dimension. So this role basically says to one person you're responsible for a cohesive strategy that drives forward, not just supports, but drives forward all the key pillars of a higher ed, students and research and instruction and learning and everything that's under that umbrella.

Gerry Bayne: Our fourth item is governance through adaptation, not proliferation.

Amarda Shehu: I put together an AI visioning task force representation from all academic and non-academic units. We really, I think, balance the ability to mobilize fast with accountability. So it's a shared governance model, and I'm a big believer in a shared governance model because then when you say, okay, here are the policies and guidelines, we can point to the people, to the faculty, to the staff, to the students that came together and designed those guidelines collaboratively. There's no question on us versus them.

Mark Daley: So if you study governance, there is essentially no governance textbook, no case study, no theoretical paper that says it is a great idea to create governance around a single technology. Our facilities team uses chainsaws. We don't have a chainsaw policy. We don't have a ballpoint pen policy. Yes, AI is new and exciting. And in our institutions, our knee-jerk response is, "Oh, we need a committee and a white paper and governance and a policy." And that's wrong. And it's especially wrong in the case of a fast-moving general purpose technology. So our approach to governance is I've been working for two years now with policyholders of existing policies and updating those policies for the academic age. So the use of AI in the classroom, that's not something we're going to govern with an AI policy that the next paragraph down is talking about snow removal, which is ridiculous.

Mark Daley: I work with our vice provost academic programs and we updated our academic policy and our academic governance mechanisms for the AI age.

Gerry Bayne: And the final item monolist is AI as an institutional mission.

Amarda Shehu: You want to align the academic, the administrative with the technical efforts because you want to make sure that this AI thing is not seen as just your PET project, but it's seen as a shared institutional project. Collaborating with other colleges and seeing through the institute the other opportunities beyond engineering, beyond computing was very key to having this bigger, wider vision for what should be the AI strategy in a higher ed.

Mark Daley: If I'm talking to our Dean of Science, the AI strategy for the Faculty of Science is just the Faculty of Science Strategy and AI is an enabler. And in some places it'll change the strategy, but the Faculty of Science has to decide that. The AI office can't tell deans what the strategy should be.

Gerry Bayne: I'm Jerry Bain for EDUCAUSE. Thanks for listening.