2. Universities Prepared Us to Find Work. What If the Future Is About Creating It?

Part 1 of After Intelligence: From Capability to Agency in the Age of AI**

For generations, one of higher education's implicit promises has been relatively straightforward.

Study hard.

Acquire knowledge.

Develop expertise.

Earn a qualification.

Find meaningful work.

It is a model that has served universities, students and economies reasonably well.

And despite some of the more breathless predictions about artificial intelligence, I do not think employment is about to disappear.

People will still become teachers, engineers, accountants, nurses, managers, researchers, designers and public servants. Organisations will still need talented people. Expertise will still matter.

But something underneath this model is beginning to shift.

Artificial intelligence increasingly allows an individual to do things that previously required a specialist, a team, considerable capital or years of technical training.

That changes more than productivity.

It changes the relationship between a person and possibility.

The question may gradually move from:

Who will give me an opportunity?

towards:

What opportunity could I create?

And that suggests that one of the most important graduate capabilities of the AI age may be something universities have historically found difficult to teach or even define.

Agency.

The worker-to-creator argument

Daniel Miessler describes a more radical version of this transition in his essay on what he calls Human 3.0.

His argument is that industrial society built educational systems primarily to prepare people to become workers: learn established knowledge, follow processes, enter organisations and perform defined roles.

AI, he argues, radically reduces the cost of becoming a builder or creator. Someone can now move from an idea to a prototype with a speed and level of technical assistance that would have been unimaginable only a few years ago.

Miessler therefore contrasts the traditional question:

“Who will hire me?”

with a newer one:

“What should I build?”

There is something powerful in that shift.

But I think creator is too narrow a word.

Not everyone should start a company.

Not everyone needs to build an app.

And we should be wary of turning one Silicon Valley ideal — the entrepreneur — into the universal model of human flourishing.

The more interesting concept is agency.

Agency can exist anywhere.

A nurse noticing a broken process and redesigning it is exercising agency.

A teacher developing a better way to help students understand a difficult concept is exercising agency.

A researcher pursuing a question others have overlooked is exercising agency.

A public servant connecting previously separate services around citizen needs is exercising agency.

A student identifying a genuine community problem and testing a solution is exercising agency.

None needs to become an entrepreneur.

All are doing something more than simply completing an allocated task.

They are shaping what happens next.

What do we mean by agency?

Agency is easy to confuse with independence, confidence or entrepreneurship.

It is broader than any of those.

At its simplest, agency is the capacity to recognise that the world is not entirely fixed and that you can participate in changing some part of it.

It combines several things:

• noticing;
• questioning;
• choosing;
• initiating;
• experimenting;
• learning; and
• taking responsibility.

Someone with agency does not necessarily reject institutions or authority.

They do not simply ignore instructions.

Agency is not rebellion for its own sake.

Rather, an agentic person can operate within a system while still recognising that the system contains choices.

They can ask:

Why are we doing it this way?

What problem are we actually trying to solve?

Who is missing from this conversation?

What else could we try?

What happens if this assumption is wrong?

And increasingly:

What becomes possible now that AI can help me do things I could not previously do?

That final question matters.

Capability used to constrain imagination

Imagine a student ten years ago who had an idea for a digital service.

Before testing the idea, they might have needed:

• programming skills;
• web design capability;
• database knowledge;
• hosting;
• graphic design;
• money to hire specialists; or
• collaborators who possessed those capabilities.

The distance between idea and experiment was substantial.

That distance acted as a filter.

Many ideas never moved beyond imagination because the cost of testing them was too high.

AI compresses that distance.

The student may now be able to describe the idea, develop an initial interface, generate code, create visual assets, analyse potential users, test alternative business models and produce a functioning prototype with AI assistance.

It will not necessarily be good.

The student may still lack important knowledge.

The prototype may be technically flawed.

AI may confidently recommend nonsense.

But something fundamental has changed:

The cost of asking “What if?” has fallen.

And when experimentation becomes cheaper, agency becomes more consequential.

From employability to agency

This does not mean universities should stop caring about employability.

It means our definition of employability may need to expand.

There is an important difference between being able to perform and being able to initiate.

Universities are generally quite good at structuring performance.

We create subjects.

Learning outcomes.

Assessment briefs.

Rubrics.

Deadlines.

Templates.

Students become very skilled at determining what is required.

That is understandable.

But there is a paradox here.

The more completely education structures the problem, the less opportunity the learner has to practise finding the problem.

We tell students:

Here is the case.

Here is the question.

Here are the criteria.

Here is the length.

Here are the readings.

Here is the deadline.

Now demonstrate your capability.

Then, after graduation, we tell them that employers value initiative, curiosity, adaptability and problem-solving.

Perhaps we should notice the contradiction.

AI makes the contradiction harder to ignore

Generative AI is exceptionally good at responding to structured tasks.

Give it a clearly defined assignment and it can help brainstorm, research, outline, analyse, draft and polish the response.

That creates obvious assessment challenges.

But perhaps it also reveals something about assessment itself.

If the most important intellectual challenge in an assignment is following instructions well enough to produce a competent artefact, AI may increasingly be very good at the same game.

The interesting educational work moves elsewhere.

What if students had to determine:

• which problem was worth investigating;
• whose perspective mattered;
• what evidence would change their mind;
• which AI tools were appropriate;
• when not to use AI;
• what assumptions the model introduced;
• what should actually be built;
• whether the result created value; and
• what they would do differently next time?

That is a much harder assignment.

It also resembles the world outside university much more closely.

Agency requires knowledge, not the absence of it

There is an important trap here.

Celebrating agency should not become an argument against expertise.

Quite the opposite.

Agency without knowledge can become misplaced confidence.

A novice with AI can now produce something that looks remarkably sophisticated.

That does not mean they understand the domain.

A student may be able to generate a legal argument without understanding law, create financial projections without understanding finance or produce software without recognising security risks.

The ability to act therefore needs to be paired with the ability to judge.

This was one of the arguments in my earlier After Scarcity series.

When AI makes expert-looking outputs easier to produce, foundational knowledge may become more important because humans need enough understanding to interrogate what the machine produces.

Agency therefore does not replace disciplinary knowledge.

It gives disciplinary knowledge somewhere to go.

We might think of it this way:

Knowledge tells you what has been understood.

Judgement helps you determine what makes sense.

Agency allows you to decide what to do next.

Perhaps universities should create more unfinished spaces

One implication is that students may need more educational experiences in which the destination is not completely predetermined.

Problems without single correct answers.

Projects with real stakeholders.

Opportunities to initiate something.

Interdisciplinary challenges.

Experiments where failure is informative.

Assessments where students must define part of the question before attempting the answer.

Situations where an educator can genuinely say:

“I don't know what the best solution is either. Show me what you can discover.”

This can be uncomfortable.

Students understandably want clarity.

Teachers understandably want assessments that can be administered consistently.

Institutions understandably want reliability.

But agency requires some encounter with uncertainty.

You cannot practise making meaningful choices in an environment where all meaningful choices have already been made for you.

From “What do you want me to do?” to “What is worth doing?”

Perhaps this is ultimately the educational shift.

For much of the industrial era, institutions created opportunities and individuals competed to enter them.

AI does not abolish that system.

But it creates a parallel possibility.

Individuals can increasingly assemble extraordinary capability around themselves.

They can research.

Analyse.

Design.

Code.

Create.

Communicate.

Translate.

Prototype.

They can do so faster and more cheaply than before.

The important graduate question may therefore become less:

Can this person perform the work we give them?

and increasingly:

Can this person recognise worthwhile work and help make it happen?

That is a very different conception of capability.

And it leads to the next problem.

Because if millions of increasingly agentic people gain access to tools that allow them to create almost anything, something else changes.

Creation itself becomes abundant.

And when everyone can make things, simply making something stops being the advantage.

Next in the series

Part 2 — Creation: When Everyone Can Build Anything, Knowing What to Build Becomes Everything

References

Farmer, I. (2026). After Scarcity: Rethinking the University in the Age of AI. Frontiering.

Miessler, D. (2025). The Problem with Human 2.0 and the Promise of Human 3.0. Daniel Miessler.