AI Will Not Transform the University. The University Has to Transform Itself.

There is a natural temptation with every new technology cycle.

We take what we already do and make it digital.

We digitized forms. We moved applications to the cloud. We created portals and mobile apps. Now we are adding copilots, agents and chatbots.

All of this can be useful.

But putting AI on top of an inefficient process does not necessarily transform it. 

Sometimes it simply gives us a faster inefficient process.

Suppose an academic process requires twelve steps, four approvals, information from three systems and two spreadsheets.

We could use AI to make those twelve steps faster.

Or we could ask a more uncomfortable question:

Why do we have twelve steps in the first place?

• Maybe six are necessary.
• Maybe three could happen automatically.
• Maybe one policy needs to change.
• Maybe two departments are doing essentially the same work.
• And perhaps the spreadsheet should never have existed.

That is where transformation starts.

Not with the technology. With the problem.

I use a fairly simple test.

If an AI initiative cannot demonstrate how it creates meaningful value for learners, faculty or administrators, I would question why we are doing it.

For a learner, value might mean fewer administrative barriers, better academic guidance, more personalized support, faster responses, stronger learning experiences and clearer pathways toward employment.

For faculty, it might mean less repetitive administration, easier access to information and evidence, better support for teaching and research, and more time for students.

For administrators and institutional leaders, it might mean fewer manual processes, trusted data, earlier visibility into risks, better coordination and faster, evidence-informed decisions.

This sounds obvious.

In practice, however, it changes how an institution approaches AI.

Instead of starting with:

What can this technology do?

Start with:

Whose problem are we solving?

Then ask:

What has to change for that problem to actually disappear?

Because sometimes the answer is AI.

Sometimes it is process redesign.

Sometimes it is policy.

Sometimes it is data.

And quite often, it is people and culture.

There is a familiar management saying that culture eats strategy for breakfast. Whatever its precise origin, the principle is particularly relevant to AI transformation.

An institution can have an excellent AI strategy.

It can establish an AI committee, appoint a Chief AI Officer, create governance frameworks and invest significantly in technology.

None of that guarantees transformation.

• If departments protect their data, transformation slows.
• If every decision has to travel through multiple layers of hierarchy, transformation slows.
• If faculty believe AI is something being imposed on them, transformation slows.
• If people are punished for experimentation that doesn’t work, transformation slows.
• If leadership delegates AI entirely to IT, transformation may never really begin.

This is why the work of John Kotter remains relevant. His model of organizational change emphasizes creating urgency, building a guiding coalition, developing a vision, removing barriers, generating short-term wins and ultimately anchoring change in organizational culture [Kotter, 1996].

The technologies have changed enormously since Kotter developed those ideas.

People haven’t changed quite as quickly.

AI transformation therefore requires two architectures.

We spend a great deal of time designing the first architecture.

I think many institutions underestimate the second.

Here is a question I find useful.

If we were designing a university today from scratch, knowing what we now know about AI, data, learning, work and human behaviour — would we design it the same way?

I doubt it.

• Would students still have to navigate multiple disconnected administrative functions?
• Would information still be entered repeatedly?
• Would institutional quality depend so heavily on periodically assembling evidence that already exists somewhere inside the institution?
• Would faculty spend as much time on repetitive administrative activities?
• Would curriculum change at the same speed?
• Would leaders depend on retrospective reports when the underlying data is already being generated in real time?
• Would every department maintain its own version of the truth?

These are not technology questions.

They are operating-model questions.

And that is where I believe the next phase of higher education transformation has to go.

Instead of asking how AI fits into the existing university, we should also be willing to ask:

What should the university itself become now that these capabilities exist?

Universities are very good at talking about student learning.

We spend less time discussing institutional learning.

Yet I think this will become one of the defining capabilities of successful universities.

Consider how much information an institution generates every day:

Admissions behaviour. Student engagement. Learning outcomes. Assessment. Advising. Faculty activity. Research. Curriculum performance. Quality evidence. Employability. Finance. Student experience.

The problem is usually not that the institution has no data.

It has enormous amounts of data.

The problem is that the information lives in different places, under different ownership, interpreted by different teams and often arrives after the moment when it could have changed a decision.

A truly adaptive institution should be able to continuously:

This connects with Peter Senge’s idea of the learning organization — an organization capable of continually expanding its ability to learn and adapt [Senge, 1990].

AI makes this considerably more interesting.

Not because AI should make institutional decisions on our behalf.

But because it can help humans find patterns, connect evidence, identify emerging risks, understand alternatives and act earlier.

I think of this as institutional intelligence.

This distinction matters.

The objective isn’t an autonomous university.

It is a more intelligent institution.

The same principle applies inside the classroom.

For a long time, access to knowledge was scarce.

Universities organized knowledge. Faculty possessed expertise. Students came to learn from those who had spent years developing it.

That world has changed.

A student can now ask an AI system to explain a difficult concept in five different ways, summarize a research paper, generate examples, translate material, debate an argument or provide personalized tutoring at 2 a.m.

That does not make faculty less important.

In some ways, I think it makes good faculty more important.

When information becomes abundant, judgement becomes scarce.

The value of the professor increasingly lies not simply in transmitting information, but in helping students question it.

• Is the answer correct?
• What evidence supports it?
• What assumptions are hidden inside it?
• What is missing?
• What are the ethical implications?
• Can you defend your reasoning?
• Can you apply the idea in a completely different context?

This moves faculty further toward mentorship, contextualisation, inquiry and intellectual challenge.

And it raises an uncomfortable question about assessment.

If AI can successfully complete an assignment, perhaps we should not only question the student’s use of AI. We should also question the assignment.

Academic integrity remains important.

But attempting to preserve every assessment model exactly as it existed before generative AI is unlikely to be a sustainable strategy.

We need assessment that places greater value on reasoning, application, creation, critique, judgement and demonstrated capability.

The objective should not be to educate students in a world where AI does not exist.

It should be to educate people who know how to think when AI does exist.

There is also a boundary I believe universities need to protect.

AI can augment human judgement. It should not eliminate human accountability.

An AI system may identify a student who appears to be at risk.

A human being still needs to understand the context.

AI may surface evidence for an academic decision.

Humans remain accountable for consequential decisions.

AI may recommend.

Humans must retain responsibility.

This is particularly important in higher education because universities are not simply processing transactions.

They are making decisions that can influence people’s futures.

That is why AI governance cannot be reduced to an acceptable-use policy.

Institutions need clarity around purpose, privacy, transparency, fairness, assessment integrity, human oversight, model and vendor risk, monitoring and accountability.

But governance should not become another mechanism for stopping change.

Good governance should make responsible innovation easier, not impossible.

Bringing these ideas together, I would approach AI transformation through seven questions.

That is the model I would use:

And then repeat it.

Because transformation is not an event.

Nobody knows exactly what AI will be capable of five years from now.

Nor can we perfectly predict student expectations, workforce requirements, geopolitics, regulation or new educational models.

So I am becoming less interested in the question:

Can we predict the future university?

And more interested in:

Can we build a university capable of responding intelligently when the future is different from what we predicted?

• How quickly can an institution recognize a problem?
• How quickly can evidence reach the people who need it?
• How quickly can leadership make a decision?
• How quickly can a process change?
• How quickly can curriculum respond?
• How quickly can the institution determine whether the change worked?

And perhaps most importantly:

How comfortable is the institution admitting that something did not work and changing again?

That, to me, is institutional resilience.

There is a danger that we make AI itself the destination.

It isn’t.

The destination is a better university.

• A university where learners encounter fewer barriers and receive better support.
• Where faculty spend more time teaching, mentoring, researching and creating knowledge.
• Where administrators spend less time reconciling information and more time improving the institution.
• Where leaders have trustworthy evidence when they make important decisions.
• Where technology handles complexity without removing human responsibility.
• And where the institution itself continuously learns.

As machines become better at producing information, some deeply human capabilities may actually become more valuable:

Curiosity. Critical thinking. Empathy. Ethics. Creativity. Judgement. Leadership. Wisdom. Responsibility.

The future university should not become less human because of AI.

It should use AI to create more space for humans to do what humans do best.

That is why I don’t believe AI itself will transform the university.

AI is the catalyst. The university still has to do the difficult work of transforming itself.

And perhaps the institutions that ultimately thrive will not be those that deploy the most AI.

They will be the ones that develop the greatest capacity to learn, adapt and change — while remaining clear about the human purpose of education.