Editor’s note: This is the 15^th article since May 20, 2026 in an ongoing series by Dr. Andrew Maxwell, the Bergeron Chair in Technology Entrepreneurship in the Lassonde School of Engineering at York University. Every week – and occasionally every other week – we’ll present a new article by Maxwell, in a series whose wide-ranging and incisive themes encompass: Canada and innovation policy; productivity and industry; innovation frameworks; AI and higher education; research and intellectual property; technology adoption; entrepreneurship and commercialization; universities and higher education; entrepreneurship education; and AI and the future of work. 

 This is Part 3 of a five-part series looking at the future of universities: Why are universities so difficult to change? What does AI mean for learning? If AI can teach, what is the professor for? And if some of the constraints around which universities were designed are disappearing, what might we design instead? Part 1 was published on August 26, 2026, Part 2 on September 2.

Higher education has a diagnosis it does not want to hear.

Universities are responding to AI as if this were a problem of academic integrity, tool adoption  and faculty training.

It is not. It is a problem of educational relevance.

And the institutions that fail to understand that are going to discover something uncomfortable: in a competitive market, students do not reward denial. They walk.

That is the part too many universities still do not understand.

Every year, they have to attract the next cohort of students. Every year, students decide what is worth paying for, what kind of educational experience seems likely to matter, and which institution feels most likely to prepare them for a changing world. If one institution offers a more meaningful, more challenging, more relevant, more future-facing learning experience than another, students will increasingly vote with their feet.

That means doing nothing is not an option. It is a strategy – a bad one.

 The comfortable lie

The comfortable lie in higher education is that AI is just another tool.

Something to manage.
Something to regulate.
Something to fit into the existing model.

So institutions do what institutions always do when they do not want to face a deeper problem. They create committees. They update policy language. They offer workshops. They debate plagiarism. They tell faculty to “experiment.” They hope the underlying model remains intact.

It will not.

The printing press changed distribution.
The internet changed access.
AI changes cognition.

But each of those shifts went much deeper than function.

The printing press did not just make books easier to reproduce. It redistributed power. It weakened centralized control over knowledge, helped new religious and scientific ideas spread, and contributed to profound changes in politics, public discourse, and eventually democracy itself.

The internet did not just improve access to information. It upended gatekeepers, rewired markets and communication, and shifted influence toward networks, platforms and individuals.

AI goes one step further. It changes not just who can access knowledge, but how people generate, interpret, test, refine and act on it. It alters the relationship between knowledge, expertise, judgment and action.

And when cognition changes, education changes.

That is not a minor upgrade. It a systemic shock to the logic of education.

Because once students can access explanation, feedback, iteration, examples, summarization, drafting support and coaching on demand, the old educational bargain starts to wobble.

And that forces a much harder question than most universities seem ready to ask: If AI can increasingly teach, what exactly is the professor for?

Not the romantic answer.
Not the defensive answer.
Not the committee-approved answer.

The real answer.

Because “content expert” is no longer enough.

“Lecturer” is no longer enough.

“Subject matter authority” is no longer enough.

Those roles were built for a world where explanation was scarce. We no longer live in that world.

 The symptoms are already here

Students are already using AI to explain difficult concepts, summarize readings, generate examples, debug code, improve writing and reflect on learning.

They are not waiting for official permission.

They are not waiting for a senate committee.

They are already redesigning parts of their own learning experience from the bottom up.

Meanwhile, many universities are still offering courses designed around an older logic: knowledge is scarce, the professor is the gateway, and learning consists largely of receiving, reproducing and being judged on command.

That mismatch is not theoretical anymore.

It shows up everywhere.

It shows up in assessment. If AI can help produce first drafts, routine analysis, basic coding, formula support and competent summaries, then a lot of what universities have historically assessed starts to look suspiciously procedural.

So let’s ask the uncomfortable question: What are we actually measuring?

It shows up in teaching. If students can get personalized explanation instantly, in multiple forms, with more patience than most professors have time to provide, then the value of simply delivering content has already dropped.

It shows up in student expectations. Students are increasingly asking not, “How do I absorb what this course gives me?” but, “What is the fastest and best way to make sense of this?”

And it shows up in the wider market. AI-native learning platforms, modular credentials, peer learning communities, corporate academies and direct-to-employer pathways are all moving into territory universities have long treated as theirs by default.

That should worry people.

Because the danger is not that universities suddenly disappear.

The danger is that they remain standing while becoming steadily less central, less compellingand less worth choosing.

 The diagnosis universities are avoiding

The diagnosis is not that professors are obsolete.

The diagnosis is that too many universities are still organized around an outdated theory of value.

That theory made sense in a world where knowledge was scarce, expertise was concentrated, careers were stable, learning happened once, and credentials served as one of the few accepted signals of capability.

In that world, the professor as gatekeeper made sense.

The institution stored knowledge, transmitted it, assessed it and certified who had mastered it.

That model worked brilliantly.

Past tense.

Because AI does not just automate tasks. It shifts the value of human capability.

As procedural work becomes easier to generate and explanation becomes easier to access, the centre of gravity moves.

What becomes more valuable is not routine execution.

It is judgment. Discernment. Problem framing. Critical evaluation. Reflection. Ethical reasoning. Intellectual courage. The ability to work through uncertainty rather than hide behind fluency.

This is why universities cannot keep pretending that content delivery is the same thing as education. It is not.

And professors cannot continue defining their value as ownership of knowledge that students can increasingly access elsewhere.

The old role of the professor was built partly on scarcity.

The new role must be built on significance.

Less gatekeeper of knowledge. More architect of learning.

Less transmitter of answers. More designer of challenge, process, and rigor.

Less content monopoly. More builder of intellectual community and disciplined inquiry.

That is not a diminished role. It is a much harder one.

Which is precisely why some faculty will resist it.

 Not all AI is the same

This is where the conversation gets lazy.

People talk about “AI in education” as though all uses of AI are educationally equivalent. They are not.

A student casually using a general-purpose Large Language Model (LLM) is not the same thing as a professor intentionally designing an AI-supported learning journey.

A generic LLM is built to be broadly helpful. It is optimized for fluency, speed, plausibility and responsiveness. That can be useful. But left alone, it often pulls toward convenience, premature closure, conventional wisdom and the illusion of understanding.

That is not pedagogy. That is assistance. And there is a big difference.

Because real education is not just about arriving at an answer. It is about learning how to think. It is about understanding why diagnosis should precede treatment, why method matters, why process matters, why evidence matters, why assumptions need to be surfaced, why solutions should not be confused with insight.

General LLMs do not reliably teach that.

Professor-designed agents can.

That is the opportunity most universities are still underestimating.

The future is not simply “students using AI.”

The future is professors designing AI-supported learning experiences that force better thinking: holding students in the problem space longer, separating symptoms from causes, requiring reflection, structuring inquiry, reinforcing sequence and connecting learning outcomes to process rather than just outputs.

In that world, the professor is not replaced.

The professor becomes the designer of the cognitive journey.

And that is a much more important job than reading from slides.

Learning outcomes are about to become embarrassing

There is another implication universities are not discussing nearly enough.

If AI changes learning, it must also change learning outcomes.

Many current learning outcomes still reflect an older educational logic: recall the content, demonstrate procedural competence, perform standard analysis, produce the required artifact.

But if AI can increasingly support explanation, summarization, first drafts, routine coding and basic analysis, then many traditional learning outcomes become weaker as indicators of real capability.

That does not mean foundations do not matter. They do. But the center of gravity has shifted.

Learning outcomes now need to put much greater emphasis on:

• problem framing
• judgment
• evidence evaluation,
• critical challenge of AI output
• integration across domains
• reflection on process
• decision quality
• working through uncertainty
• the discipline to follow sound method rather than chase quick answers.

In other words, the point is no longer just to produce answers.

It is to develop the maturity to know which questions matter, which evidence holds, which outputs deserve skepticism, and how to move from information to insight.

That is a very different ambition.

And it means universities cannot simply sprinkle AI onto an unchanged curriculum and call it innovation.

That is not redesign. That is decoration.

 The questions universities really do not want to ask

Once learning outcomes shift, an even bigger question comes into view: What is education actually for?

And related to that: What exactly is a degree supposed to certify?

For years, degrees have served as bundles of signals: persistence, capability, disciplinary exposure, achievement, the ability to survive complexity, and a socially recognized marker of readiness.

But if AI changes how students learn, how they demonstrate competence, what support they can access, and which human capabilities matter most, then the meaning of the degree becomes much less obvious.

What is it certifying?

Content mastery? Process discipline? Judgment? Intellectual maturity? The ability to cope with ambiguity? Readiness for work? Readiness for citizenship?

Universities do not have to fully resolve that tomorrow.

But they do have to stop acting as though the answer is self-evident.

Because if institutions cannot explain what their credential means in an AI-transformed world, employers, students, and alternative providers will do it for them.

And the answer may not flatter the incumbents.

 The treatment

The treatment is not to ban AI.

And it is not to surrender to it.

The treatment is to redesign education around the forms of value that do not auto-generate.

That means courses built around meaningful intellectual work, not just content coverage.

It means assessment that rewards reasoning, critique, framing, reflection, iteration and process discipline, not just competent output.

It means professors using AI not to replace student thinking, but to intensify it.

It means investing in purpose-built agents and pedagogical architectures, not just generic AI access.

It means rethinking learning outcomes at the level of courses, programs and degrees.

It means helping faculty redesign the learning journey, not just learn a new tool.

And it means recognizing that this is now a competitive issue, not merely an educational one.

In many markets, the institutions that create more meaningful, more developmental, more future-relevant learning experiences will attract more students.

The institutions that do not will not simply remain unchanged. They will lose ground.

There will be side effects.

Some faculty will feel exposed, because AI reveals how much of their historical value came from controlling explanation rather than designing transformation.

Some courses will look outdated. Some assessments will need to die.

Some institutions will discover that many of their most cherished practices are educationally weaker than they assumed.

Students may become less patient with passive teaching.

That is not a bug. That is a sign that they are correctly identifying low-value educational design.

There will also be noise, confusion, poor implementation, and a lot of performative innovation theater.

Again, not a reason to do nothing. A reason to do better.

 Prognosis if treatment is delayed

If universities do not adapt, they will not all vanish overnight.

But that should not comfort anyone.

The real mechanism of decline is simpler. Students will vote with their feet.

Every year, institutions have to win the next cohort. In competitive environments, especially urban ones, students can compare programs, formats, experiences, outcomes and perceived value. If some institutions redesign around real intellectual development, experiential learning, richer feedback and AI-enhanced rigor, while others continue offering tired, outdated academic experiences, students will sort themselves accordingly.

The leaders may grow.

The laggards may shrink.

And once enrollment starts to fall, the spiral can become self-reinforcing.

Lower enrollment means lower revenue.
Lower revenue means fewer resources.
Fewer resources mean weaker programs and less innovation.
That makes the institution less attractive. Which leads to more decline.

That is how disruption often works in higher education.

Not through sudden death. Through differential adaptation, student sorting, and compounding erosion.

The diagnosis is not extinction.

It is enrollment decline, institutional stratification, and slow marginalization.

Final note from the innovation doctor

The wrong question is whether AI threatens professors.

The right question is whether universities are willing to let AI expose what was already becoming obsolete.

If AI can explain the lecture, summarize the reading, support routine work, and help students generate acceptable outputs, then the professor’s value does not disappear.

But it does migrate.

Toward curiosity.
Toward discernment.
Toward courage.
Toward judgment.
Toward identity.
Toward wisdom.

These do not auto-generate.

Neither does the kind of educational experience that develops them.

That is why the future role of the professor is not smaller. It is sharper.

And that is why doing nothing is not neutrality. It is surrender.

 Prescription: redesign the learning journey, not just the tool stack.
Warning label: delayed action increases the risk of declining relevance, declining enrollment and institutional erosion.

R$