A Map for the AI Conversation Your Campus Needs to Have

Walk into almost any institution right now and you will find the same split running down the middle of the building. In one office, someone is piloting an AI tutor or an advising chatbot and moving fast. Down the hall, someone else is quietly worried about academic integrity, about FERPA, about a vendor contract nobody in governance has read. Both people are right. And the gap between them is the actual work of the next few years.

That gap is what every serious AI convening in higher education is now trying to close. Spend a season in the sector’s conferences, webinars, and working discussions, and the same question keeps surfacing: How do institutions build responsible, mission-aligned approaches to AI, rather than a scattered pile of pilots?

Across those convenings, the same five themes keep showing up:

• Aligning AI with mission and values. Strategies for tying AI initiatives to institutional purpose, not adopting tools for their own sake.

• Teaching, learning, and assessment. What good pedagogy and honest evaluation look like in an AI-enabled classroom.

• Governance, policy, and ethics. The rules, oversight, and ethical guardrails that make responsible implementation possible.

• Coordinating across divisions. How institutions align academic and administrative efforts instead of duplicating them.

• Practical lessons from real campuses. What actually worked, what broke, and how approaches get refined over time.

What is striking is the convergence. Very different institutions and organizers keep arriving at the same questions. What the sector has not agreed on is the answers.

That pattern is what got me building a framework. I kept hearing the same themes in different rooms and noticed they did not stay in one place. The same questions were being worked out in my own department, again at the system level in the UC and California Community College systems, and again at EDUCAUSE. Watching the same conversation restart from scratch in every room is what got me thinking about how to have it once, as one connected model.

In my observation, most campuses do not have a connected model yet. What follows is a way to think about it, organized around those five themes. Consider it a field guide: something to read before your next convening, a set of places to start after, and for colleagues who cannot get to those rooms at all, a way to work through the same questions from wherever you are.

Before getting into the themes, one thing plainly. The deepest learning at a gathering does not come from any framework, including this one. It comes from the room: the provost who tells you over coffee exactly which policy blew up and why; the librarian and the faculty member who discover they have been solving the same problem in isolation; the peer who asks the question that reframes your whole approach. If you can be in that room, be in it.

But not everyone can be in that room. These conversations happen at conferences and convenings that take days to attend, and not every budget, calendar, or campus can absorb that. This piece is for them especially. It will not replace the hallway. What it can do is give you a running start, so whether you are in the room or reading this from your desk, you are not starting from a blank page.

A strategy is not a longer list of pilots. It is the shared logic that decides which pilots matter, how they are governed, and how they connect to mission.

Every campus already has a mission statement. Very few have asked what it implies for AI. An access-focused institution and an elite research university should make different AI decisions about which tools they adopt, which risks they accept, and which efficiencies they refuse because they would undercut the point of the place.

This is where alignment stops being a slogan and becomes a filter. A community college weighing an AI system to optimize financial-aid packaging has to ask a hard question: does optimization serve access, or quietly undercut it? Mission is the tiebreaker, but only if you have made it concrete enough to actually decide with.

That is the job of the first two pillars of the framework: turning campus mission and vision into a testable AI vision, and translating it into AI principles that give people decision norms rather than platitudes. When you need to weigh a whole portfolio of initiatives against that vision, the Strategic Compass scores each one for strategic fit before anyone commits budget.

Where to start: Draft a one-page AI vision, then pressure-test three initiatives against it. The Campus AI Framework walks through this in the Mission & Vision and Strategic Compass sections.

If Theme One is about institutional direction, this one is personal. Faculty feel it first and most directly, and it is where blanket policy fails hardest. “Ban it” and “embrace it” are both wrong, because the right answer changes by assignment, by discipline, and by what you are actually trying to teach. An AI writing assistant that erodes skill-building in a first-year composition course might be exactly the right scaffold in an upper-level seminar.

The useful move is not a rule. It is a vocabulary. Instructors need shared language for disclosure, for where human judgment is non-negotiable, and for what “assessment” means when a model can produce a competent first draft. Real examples help more than abstractions: AI tutoring, automated feedback, early-alert systems that flag struggling students. Each carries a different risk profile and deserves a different level of scrutiny.

Where to start: Look at specific use cases for teaching, learning, and assessment. Each one should name the value, the risks, and a proportionate review path. The framework’s Use Cases section does this by domain.

Once you know your mission and your classroom approach, you need the rules. And here is where most campuses make a mistake: they respond to AI by writing a brand-new “AI policy” from scratch. Usually that is the wrong instinct. Most AI questions are new inputs to processes that already have rules: privacy, procurement, records, academic integrity. The stronger move is to govern the behavior and the data, not the product, so your policy survives the next model release instead of being obsolete by spring.

What genuinely is new (disclosure, human oversight, accountability when a system gets it wrong) deserves focused guidance and a named owner with a review cycle. Match scrutiny to stakes: a low-risk research summarizer and a high-stakes admissions screener should not go through the same review. Ground it in recognized standards (NIST AI RMF, OECD principles, FERPA, ISO/IEC 42001) so your approach is defensible to a board, an accreditor, or a worried faculty senate.

Write around data, behavior, and risk. Never a single tool or vendor. Durable language survives model churn.

Where to start: Audit your existing policies for AI gaps rather than building from scratch. The framework’s Governance Guide covers the oversight structures, and you can check any approach against recognized standards in the standards alignment.

Good policies and good classroom practices still fail if nobody coordinates them. The most common failure mode is not a bad decision. It is the same decision being made five times in five offices that never talk to each other. IT vets a vendor the provost’s office is already piloting. The library builds AI-literacy programming the teaching center does not know exists. Student affairs deploys a chatbot that touches data governance has never classified. Nobody did anything wrong. There was simply no operating model.

Coordination in higher education is its own discipline, because campuses run on shared governance, not command structures. You cannot just issue an org chart. You need inclusive participation that respects faculty authority, clear ownership so “everyone’s responsible” does not become “no one is,” and a way for academic and administrative sides to see each other’s work.

Where to start: Map who is currently making AI decisions on your campus and where the gaps are. The framework’s Roles & Responsibilities and shared governance sections address this directly.

All four themes above are easier to discuss than to execute. So the fifth theme is the most practical: what have campuses actually tried, and what happened? The best part of a conference is rarely the keynote. It is the hallway conversation where someone tells you what actually broke when they tried it. That is the knowledge worth capturing: not “AI is transformative,” but “here is the pilot we killed, and why,” and “here is the review step we skipped that came back to bite us.”

Practical progress tends to follow a shape: diagnose honestly where you are, pick a proportionate path rather than boiling the ocean, and treat AI as sustained operations, resourced, monitored, improved, not a launch you can walk away from.

Where to start: Run a quick maturity assessment, then map your first 90 days against the gaps it surfaces.

These five themes are not separate problems. They are connected, and behind all of them is a single operating model built around eight pillars spanning governance, enablement, and capability: campus mission and vision, AI principles, policies and guidelines, governance and risk and compliance, engagement and collaboration, campus readiness, roles and responsibilities, and implementation and operations. Whether your institution is still assessing what AI means for it or refining an approach already live, the same map applies. You enter it at a different point.

The full framework, every pillar, guide, and tool, is free and open at campusaiframework.com.

The future of AI in higher education will not be decided by a vendor or a model. It will be decided by the ordinary, consequential choices faculty, staff, and administrators make this year: which pilot to fund, which policy to write, which risk to refuse. That is not a reason to wait. It is the reason to get the structure right now, so those choices add up to something coherent instead of a hallway full of good intentions that never met.

If you are heading to a convening, go with a plan for what you will bring home. If you cannot be there, start with the framework and take the themes back to your own colleagues. A working group over lunch on your own campus can do a surprising amount of what a conference does. You just have to start it.