10 AI Mistakes University Leaders Keep Making

Most universities now have an "AI strategy." Far fewer have a good one.

The gap rarely comes from bad intentions. It comes from a handful of predictable mistakes — the kind that look reasonable in a leadership meeting and only reveal themselves a year later, when adoption has stalled, faculty are frustrated, and the strategy is a PDF no one opens.

Here are ten of the most common, and what to do instead. None of these are about a specific person or institution. They're patterns — and the first step to avoiding them is recognizing them.

AI gets handed to the technology office, a vendor is selected, a license is bought — and leadership considers the job done. But AI in a university isn't a software rollout; it's a change in how teaching, assessment, and research happen. Procurement is the easy 10%. The pedagogy and culture are the other 90%, and they don't come in the contract.

A common reflex is to draft a campus-wide AI policy fast, to "get ahead of it." But policy written before anyone understands how students and faculty actually use these tools tends to ban what can't be enforced and permit what should be examined. Watch real practice first. Policy that follows evidence survives; policy that precedes it gets ignored.

A values statement about "responsible AI" is not a strategy. A strategy says who does what, with what budget, by when, and how you'll know if it worked. Many institutions publish the former and believe they've done the latter.

Decisions about AI in teaching get made by committees that don't teach. Faculty then experience AI as something done to them, not with them — and the predictable result is quiet resistance. The people closest to the classroom have the best information about what will actually work there.

The instinct to prohibit feels safe and decisive. But blanket bans push usage underground, teach students that the institution is behind, and forfeit the chance to build judgment. The harder, better path is teaching students how to use these tools well — including when not to.

If your degree's assessments can be completed by a chatbot in thirty seconds, the problem was visible before AI — AI just made it impossible to ignore. Leaders who treat this as a cheating crisis miss the real opportunity: to redesign assessment around the things that actually demonstrate learning.

Licenses are easy to approve because they're a line item. Professional development is harder because it's time, and time is the scarcest resource on campus. So institutions buy access and skip the capability-building — then wonder why adoption is low. A tool no one is trained to use is shelfware.

There's pressure to be seen doing something with AI — a center, a task force, a press release. Visible activity is mistaken for progress. But a launch event is not an outcome. Ask what will be measurably different for students and faculty in a year; if there's no answer, it's theater.

When AI access depends on who can afford the paid tier, or who already arrived knowing how to use it, existing gaps widen. Leaders focused on innovation can miss that the same tools can deepen inequality unless access and instruction are deliberately leveled.

Strategies built around the capabilities of today's tools are obsolete by the time they're approved. The pace of change means the durable questions aren't "which tool" but "what do we want a graduate to be able to do," and "how do we help our people keep learning." Build for adaptability, not for a snapshot.

Look back over the list and a pattern emerges. Almost every mistake comes from treating AI as a thing to manage — a tool, a risk, a policy — rather than a shift in how the institution learns and teaches.

The leaders getting this right aren't the ones with the biggest AI budgets or the flashiest pilots. They're the ones asking a quieter question: what is the university actually for, and how does this technology serve that — or threaten it?

That question doesn't have a vendor. It has to be answered by people who understand the institution. Which is to say: by you.

That's the Attento take. Strategy isn't about keeping up with AI. It's about staying clear on what matters while everything around it changes.

Leading AI strategy at your institution? Reply and tell me where you're stuck — the hardest problems often make the best future issues. I read every email.

See you Tuesday, Rina