Special note: If you’re working at a community college, I posted a roadmap and workbook for you at the end of this post along with a draft Excel dashboard if you’re interested in seeing exactly how I’d lead through our current moment. If it’s useful to you, great! These tools require contextual adaptation, of course, but they reflect the priorities we’re discussing in org transformation. Just be cautious if you print that Excel document; I am a VERY sloppy formatter! Adapt as needed1.
In my last piece, I wrote about MIT’s new report on AI and higher education with a focus on its recommendations that colleges reconsider assessment and explore alternatives to traditional grading among other things.
I spent most of that discussion in familiar territory for me: Pedagogy, grading, power, student motivation, and the complicated reasons students might hand intellectual work over to AI.
But there’s another part of this conversation that requires some discussion, and I would suggest that the whole machine might rise or fall on it: Leadership.
Because if we take the MIT report seriously, then what it describes isn’t primarily a faculty-development problem but an institutional-change problem, and we might not be thrilled to find ourselves here yet again.
Big moments like this one (the pandemic is another example) are exactly why I consistently harp on trust as an organizational technology: When it’s time to engage in Serious Transformation, you cannot — you will not — be successful without the foundation of trust.
Further, I posit that all of that leadership training people have undertaken was meant to prepare them to lead in this moment: When we’re talking about leadership, THIS is what we mean. We find ourselves on shaky footing with unclear outcomes and murky ideas about where we’ll end up, but we know we need to end up on the other side of whatever we’re facing. That’s leadership.
Sometimes, things go awry when context shifts, and higher education has a peculiar habit when confronted with major disruption. Something changes in the environment, org leaders acknowledge that the change is profound, and then the practical response works its way downward until it becomes a new collection of things faculty are expected to do:
· Learn the technology.
· Redesign the assignment.
· Rewrite the syllabus.
· Develop an AI policy.
· Rethink assessment.
· Create more authentic learning experiences.
· Build community.
· Preserve rigor.
· Teach AI literacy.
· Maintain academic integrity.
And, while you’re at it, please make sure enrollment, retention, completion, accessibility, and student satisfaction remain strong. On paper, it might look like institutional transformation, but in reality, it’s just workload transfer.
In my next piece, I’ll address the question, “Workload transfer TO WHOM,” because we have the dynamics of a labor model to discuss as well.
If AI really represents the kind of watershed MIT says it does (I believe so), colleges cannot afford to confuse the two.
A watershed requires more than a workshop
MIT President Sally Kornbluth used an important word in introducing the report: Watershed.
She wrote that it’s rare for a successful institution to need to take a fresh look at so many aspects of its mission but that generative AI presents precisely that situation. I think she’s right.
The implications reach into assessment, curriculum, academic integrity, student support, instructional design, faculty labor, technology, physical space, social learning, and eventually into some of our most basic assumptions about what constitutes a college education.
That scale should tell leaders that a watershed can’t be managed through a Tuesday-afternoon professional-development session, nor can it be managed by purchasing an enterprise AI tool, creating a committee, adding language to an academic-integrity policy, inviting more tech bros to espouse the virtues of AI, and announcing that the institution is now “AI ready.”
The orgs that genuinely evolve will be the ones whose leaders understand that their role is to create conditions in which people can rethink the work, which is much different from telling people to rethink the work.
First, leaders have to create slack
Here’s the least glamorous recommendation I can make, but perhaps the most important.
People need time.
Real institutional learning requires reading, experimenting, talking, failing, observing, revising, and trying again. None of those activities happens especially well when people are already working at capacity.
If orgs want faculty to fundamentally reconsider assessment, that work has to occur somewhere in the workload.
If faculty are expected to redesign courses around meaningful human learning, rethink assignments that AI can now complete, develop new forms of assessment, determine appropriate uses of AI within their disciplines, and learn what these technologies can and cannot do, administrators have to answer a very simple question: When? “Be innovative” is not a workload model.
Leaders who are serious about transformation will need to create actual capacity for it. That might mean course releases, summer stipends, compensated working groups, instructional-design support, smaller pilots, temporary reductions in other service expectations, or dedicated collaborative development time.
The precise mechanism will vary depending on unique contexts but the underlying principle remains the same: If the org benefits from the labor of redesign, the org has to resource the labor of redesign. Period.
I am someone who undertook pedagogical redesign several years ago, and no one asked me to do that. I have intellectual itches constantly, and pedagogy is one of them, so I dug into the research to transform my own approach in the classroom. What served me the most back in those days, quite honestly, was the fact that people just left me alone to do what I wanted to do.
But now, seven or eight years later, we’re in a different context. Now the work isn’t voluntary, and that circumstance requires some additional structure and leadership behaviors.
Protect experimentation from the demand for immediate proof
Leaders also need to become much more comfortable with experimentation. Higher education often says it wants innovation while maintaining systems that punish uncertainty; very often, before we try something, we want evidence that it’ll work. Once we try it, we want measurable outcomes almost immediately. If the outcomes are mixed, we become nervous.
If an experiment fails, we may feel pressure to return to the familiar practice and perhaps conclude that the innovation itself was flawed. But that’s not how learning orgs behave. Right now, nobody knows with confidence what undergraduate education should look like ten years into widespread generative AI use, but we’re going to have to find out.
That means leaders must create spaces in which faculty can pilot new forms of assessment, change course structures, experiment with oral or collaborative work, rethink deadlines, use or prohibit AI strategically, develop new experiential practices, and discover that some of those experiments were terrible ideas.
Failure has to be information, it has to be data; otherwise, faculty will understandably select the safest possible innovations: Those that leave the existing architecture mostly intact. True evolution requires considerably more permission than that.
Stop looking for the AI policy
One thing I appreciate about MIT’s approach is its recognition that AI use must make sense within the purpose of a specific class. The committee recommends clear policies, including rationales, rather than assuming that one universal rule can govern every learning context.
That’s smart. An org-wide policy can establish boundaries around privacy, security, accessibility, academic integrity, intellectual property, or approved tools, but it can’t determine what meaningful AI use looks like in first-year composition, nursing, calculus, automotive technology, philosophy, welding, graphic design, or organic chemistry. Those are pedagogical and disciplinary questions.
Leadership during this period therefore requires resisting the administrative desire for premature standardization. We need guardrails where they’re necessary, but from there, leaders should leave meaningful room for academic judgment, and you already know what I’m going to say:
That requires trust, and our current moment will reveal rather quickly which orgs possess it. I am posting a related piece within this piece if you want to see what it looks like when an org starts to take trust seriously: Check out what happened at Pfizer after it struggled over decades with change processes.
Ask different questions
My next piece of advice is that we ask different questions. Leadership questions matter because questions direct organizational attention.
A weak institutional response to AI asks:
· How do we stop students from cheating?
· How do we detect AI?
· What software should we purchase?
· What should the policy say?
· How do we make faculty comply with it?
Those questions are not entirely irrelevant, but they’re definitely too small. A stronger set of questions sounds more like this:
· What forms of learning do we consider indispensable?
· How will we know whether that learning has occurred?
· Which parts of our current educational model remain valuable even when AI can perform some of the tasks embedded within it?
· Which parts existed largely because the technology of the previous era required them?
· What kinds of human interaction become more valuable as automated cognitive work becomes easier?
· What do students need to know how to do without AI?
· What do they need to learn to do with AI?
· Where should productive friction remain?
· What experiences are sufficiently valuable that we should redesign institutional structures to preserve them?
And perhaps the most uncomfortable question:
· What are we currently doing that AI has exposed as educationally thin?
Those are leadership questions because faculty can’t answer them alone. They require curricular structures, academic policy, resource allocation, scheduling, faculty development, student services, technology infrastructure, governance, and budget priorities to move together.
Treat faculty as knowledge holders
I see another trap here that colleges should work hard to avoid: Do not build the institutional AI strategy somewhere else and then bring faculty in during implementation. Faculty are already conducting the research and development.
Every semester, instructors are watching students interact with these technologies in real time. They’re seeing what works, what fails, where students become confused, where AI accelerates learning, where it substitutes for learning, which old assignments collapse immediately, and which kinds of work become unexpectedly valuable.
That’s org intelligence, and leaders should be harvesting it; not through another survey nobody has time to complete, necessarily, but through ongoing structures that allow people to compare observations, test ideas, share failures, develop practices, and influence institutional decisions.
MIT’s emphasis on communities of practice is important here. Its emerging AI hub explicitly includes working sessions, seminars, shared guidance, and opportunities for instructors and staff to learn from one another. I appreciate the principle they’re operationalizing here: During periods of uncertainty, organizations need lateral learning. People closest to the work need mechanisms for talking to one another, and people with positional authority need mechanisms for listening to them.
Make governance more important, not less
Rapid change often produces a dangerous organizational instinct: We don’t have time for process.
Sometimes that’s true. Emergencies occasionally require immediate decisions.
But “AI is moving fast” can also become a remarkably convenient rationale for bypassing consultation, faculty governance, disciplinary expertise, or the slow work of institutional sensemaking, which would be a mistake. When the environment becomes less certain, orgs need more distributed intelligence, not less.
Faculty governance, curriculum committees, academic departments, instructional designers, librarians, student-affairs professionals, technology staff, accessibility experts, and students all hold different pieces of the problem. The task of leadership is not to eliminate that complexity but to organize it.
Good governance during rapid change will undoubtedly have to become more agile. Processes may need to move faster and pilot authority may need to be delegated. Some decisions may need sunset provisions so orgs can act provisionally rather than pretending that every new policy is permanent.
But agility and centralization are not synonyms. In fact, overly centralized decision-making may be especially brittle in an environment changing this quickly. By the time one central office identifies the problem, develops the policy, approves the policy, communicates the policy, and trains everyone on the policy, the problem may have changed.
Orgs need structures capable of learning closer to the work.
Do not create an innovation penalty
This one is remarkably practical.
If an instructor redesigns a course around AI-aware learning and the result is more labor-intensive, what happens?
If alternative grading produces more substantive feedback conversations, does the org recognize that labor?
If oral assessments take considerably longer than machine-graded exams, does anybody account for that?
If authentic assessment requires smaller classes, additional technology, community partnerships, laboratory access, or different scheduling, will the org provide those things? Or will faculty be told to innovate within the existing resource envelope?
Orgs frequently create what amounts to an innovation tax: The people most willing to improve things are rewarded with considerably more work, and over time, the system teaches people an entirely rational lesson: Do less.
Good leaders will notice where transformation creates labor and redesign structures accordingly; otherwise, the org will slowly exhaust precisely the people it depends upon to evolve.
Let some old things die
This may be the hardest leadership task: Orgs cannot endlessly add.
If AI means we genuinely need new kinds of learning experiences, new assessment structures, additional faculty development, stronger communities of practice, more human interaction, and new forms of student support, something else may need to occupy less institutional space.
· Maybe certain assignments disappear.
· Maybe some policies disappear.
· Maybe some reporting requirements disappear.
· Maybe some committees disappear.
· Maybe some technologies disappear.
· Maybe some assumptions about seat time, course structure, credit hours, scheduling, faculty roles, or assessment eventually need substantial reconsideration.
Transformation without subtraction becomes accumulation, which produces exhaustion, not evolution. Leaders should therefore ask an unusual question whenever they introduce a major new expectation:
What are we going to stop doing so people can do this instead?
I would put that question on the wall of every cabinet meeting for the next five years.
Know that culture is the actual work
One of Kornbluth’s most important observations about the MIT report is that while practical changes matter, the larger challenge is cultural (it almost always is, friends). MIT will need shared understandings about what AI should and should not do within education and about what members of the community owe one another.
That’s precisely right because culture is where leadership becomes especially important. The easiest org culture to create around AI is one organized around suspicion.
· Students can’t be trusted.
· Faculty are falling behind.
· Administrators do not understand teaching.
· Nobody knows what anyone else is doing.
· Everyone needs monitoring.
· Every ambiguity requires another policy.
That culture would be catastrophic for long-term sustainability.
The alternative is not naïveté, though, because academic dishonesty certainly exists. AI will be misused and faculty will make questionable choices. Administrators will also make questionable choices. Vendors will promise things they cannot deliver.
We can’t remove accountability from this equation, but we CAN build systems in which competent adults have meaningful agency, expectations are clear, mistakes can be discussed, disagreement is possible, and people don’t need to hide uncertainty in order to appear compliant or competent.
AI is going to require enormous amounts of organizational learning; org learning requires psychological and professional safety. People have to be able to say:
· I tried this and it failed.
· I don’t understand this technology yet.
· My previous assignment no longer works.
· I changed my mind.
· My students taught me something.
· I need help.
· I disagree with the current approach.
Those sentences are signals of a learning organization, and leaders should pay very close attention to whether people feel safe enough to say them.
This is what leadership is for
As a leadership scholar, I’ll say that the MIT report interests me for reasons beyond AI because it describes the kind of moment organizations routinely claim to be preparing leaders to navigate:
· Uncertainty.
· Rapid technological change.
· Competing values.
· Incomplete information.
· High stakes.
· Distributed expertise.
· No obvious endpoint.
This is the work. We don’t know what the future will look like after generative AI, so our focus must be on building orgs capable of finding out.
That process requires resources, time, experimentation, distributed decision-making, meaningful faculty participation, tolerance for failure, attention to workload, and enough institutional humility to recognize that some longstanding practices may need to change.
It also requires resisting the temptation to treat transformation as something leaders announce and everyone else implements. If the model of education is changing, the org has to change with it, and that includes the people at the top, perhaps especially the people at the top.
Free resources are below for ya, and next time, I’ll address what this transformation means in the context of some very real labor constraints, one of the least convenient parts of this conversation: How do we sustain real pedagogical transformation when the people doing much of the teaching are employed contingently?
Free Resources: If you like what you’ve read here and what to know how exactly I would lead an org through this process AND you’re at a community college, take this free roadmap and workbook below. I built them with you in mind! Excel dashboard available, too (sloppy formatting! Use caution when printing!). These are loaded in the order in which I would use them: Roadmap first, then workbook.
Thank you in advance to the people who I know will reach out to me to say I should monetize these resources for my own good, but please know that I truly believe in the democratization of knowledge. Your government supported my attendance at every level of my education through its student loan policies at the time (RIP); this information and the benefits of these skills are just as much yours as they are mine.
