3. When Everyone Can Build Anything, Knowing What to Build Becomes Everything

Part 2 of After Intelligence: From Capability to Agency in the Age of AI**

We have spent much of the generative AI era astonished by what machines can create.

Text.

Images.

Music.

Video.

Software.

Presentations.

Research summaries.

Business plans.

Marketing campaigns.

Entire digital products.

Each new capability produces another round of demonstrations showing us that something once requiring considerable specialist skill can now be produced through a conversation with a machine.

And this genuinely matters.

Creation is becoming democratised.

People can participate in forms of making that previously sat outside their technical capability.

Someone who cannot draw can visualise an idea.

Someone who cannot programme can prototype software.

Someone without a production studio can make a video.

Someone without a marketing team can build a campaign.

This is an extraordinary expansion of human capability.

But there is a second-order consequence that receives much less attention.

When everyone can create more, creation itself becomes less scarce.

And that changes where value lies.

AI does not just democratise creation. It commoditises it.

That sounds harsher than I intend.

Creation will obviously continue to matter.

But scarcity helps determine value.

A professionally designed website once represented substantial time, capability and cost.

A polished presentation did too.

So did an illustration, piece of software, promotional video or market analysis.

Increasingly, the mere existence of the artefact tells us less about what went into producing it.

That is not necessarily bad.

It means more people can participate.

The barrier to entry falls.

Ideas that would previously have remained unrealised can be tested.

Small organisations gain capabilities once available primarily to larger ones.

Creators can work across disciplines without mastering every technical tool themselves.

But abundance has consequences.

Consider what has already happened with written content.

The world did not suffer from a shortage of words before generative AI.

Now we can produce billions more of them at negligible marginal cost.

The result is not automatically greater understanding.

Sometimes it is simply more content.

The same may soon apply across many forms of creation.

More images.

More applications.

More videos.

More songs.

More research summaries.

More ideas.

More recommendations.

More everything.

The question becomes:

What deserves to exist among all this abundance?

Scarcity keeps moving

This is the central argument running through both After Scarcity and this series.

Artificial intelligence does not simply remove scarcity.

It relocates it.

This produces a subtle but important change.

For a long time, one of the most important questions was:

Can we make this?

Increasingly, the answer may simply be:

Yes.

Which makes another question more important:

Should we?

And beyond that:

Why this rather than the thousand other things we could make instead?

The prototype before lunch

One of the most compelling claims in Daniel Miessler's Human 3.0 argument is that AI radically compresses the distance between idea and execution.

A person can have an idea in the morning and potentially produce a working prototype by lunch.

That is remarkable.

But imagine what happens when everybody can do it.

Ten ideas become one hundred prototypes.

One hundred become ten thousand.

Suddenly the scarce part is not prototyping.

It is deciding which prototype deserves another afternoon.

This is a pattern we have seen before.

Digital photography made taking photographs effectively free.

The result was not that photography ceased to matter.

We took vastly more photographs.

But the scarcity moved towards attention, selection and composition.

Streaming made access to music abundant.

The problem became deciding what to listen to.

Search engines made information abundant.

The problem became deciding what to trust.

Social media made publishing abundant.

The problem became deciding what deserves attention.

Generative AI extends this logic into creation itself.

More ideas do not necessarily mean better ideas

There is an assumption buried inside much of the excitement around generative AI:

More possibilities must be better.

Sometimes they are.

If AI helps a designer explore 50 alternatives rather than five, perhaps the tenth or 37th contains something genuinely novel.

If it helps a researcher explore different hypotheses, that may reveal a direction previously overlooked.

If it allows a student to experiment rapidly, that can accelerate learning.

But generating alternatives and recognising quality are different cognitive tasks.

Twenty plausible options can help.

Two thousand can overwhelm.

The constraint shifts from generation to selection.

This is particularly interesting because generative systems are extraordinarily good at producing things that are plausible.

Plausible names.

Plausible strategies.

Plausible arguments.

Plausible visual designs.

Plausible innovations.

Plausibility is useful.

But it is not the same as importance.

And a world overflowing with plausible possibilities may actually make it harder to recognise ideas that matter.

The danger of creating because we can

There is a wonderfully human tendency to confuse technological possibility with necessity.

We discover that something can be automated and immediately ask how quickly we can automate it.

We discover that AI can generate 100 personalised messages and immediately imagine sending all 100.

We discover that content can be produced continuously and create systems for producing it continuously.

But nobody particularly wants to receive infinite AI-generated emails, articles, advertisements, videos or social posts.

At some point, abundance encounters another scarcity:

Human attention.

This should make us reconsider a great deal of AI-enabled productivity.

If an AI allows me to produce ten times as much content, but every other person and organisation can also produce ten times as much, have we made communication better?

Or have we simply increased the volume everybody must filter?

Productivity measured purely by output may become a strangely misleading metric in an age where output is nearly free.

Perhaps value moves towards subtraction.

The ability to say:

This one.

Not those nine.

This matters.

That doesn't.

Stop.

Simplify.

Remove.

Do less, better.

Creation without purpose becomes noise

There is an educational implication here.

Universities have traditionally rewarded production.

Write the essay.

Complete the report.

Produce the presentation.

Submit the analysis.

The artefact demonstrates the learning.

But if AI can substantially assist in producing the artefact, perhaps our interest should move upstream.

Why this topic?

Why this method?

Why this conclusion?

Why this audience?

What alternatives did you reject?

What criteria did you use?

What changed your mind?

What did you choose not to include?

What consequences did you consider?

In other words, perhaps we should increasingly assess the intellectual architecture behind creation rather than only the creation itself.

This does not mean essays disappear.

Nor presentations.

Nor reports.

But the finished artefact may become evidence of a deeper process rather than the sole object of assessment.

Creation should become cheaper. Care should not.

There is another risk in treating AI creation purely as liberation.

When production becomes effortless, care can disappear.

Writing something used to impose friction.

Designing something imposed friction.

Making software imposed friction.

Friction was inconvenient.

But sometimes friction forced reflection.

You had to decide whether something deserved the effort required to create it.

AI removes much of that constraint.

That is often wonderful.

But perhaps we need to replace production friction with decision friction.

Before generating the hundredth option:

What are we actually trying to achieve?

Before building the application:

Whose problem does it solve?

Before creating the campaign:

Should this message exist?

Before automating the process:

Is the process itself worthwhile?

Before asking AI for an answer:

Is this the right question?

The scarce capability may become the willingness to stop and ask these questions.

The strange return of taste

Miessler uses a deceptively simple word for part of this:

Taste.

He argues that if AI can increasingly make almost anything, knowing what is worth making becomes an edge.

I think he is right, although I would broaden the idea.

Taste sounds aesthetic.

It makes us think about art, music, writing or design.

But there is also:

• intellectual taste;
• strategic taste;
• ethical taste;
• research taste;
• design taste; and
• problem-selection taste.

A good researcher develops a sense for which questions are fertile.

A good entrepreneur learns which problems people genuinely care about.

A good teacher recognises which explanation will unlock understanding.

A good leader distinguishes important signals from organisational noise.

A good editor knows what to remove.

A good designer knows when another feature will make the product worse.

These forms of taste are difficult to formalise precisely because they emerge from accumulated knowledge, experience, values and judgement.

AI can help with all of them.

But the human still needs some basis for choosing.

From maker to curator

Perhaps one way to describe the emerging shift is:

Yesterday's capable person was primarily a maker.

Tomorrow's capable person may increasingly be a maker-curator.

They can generate.

But they can also select.

They can expand possibilities.

But they can also narrow them.

They know when to ask AI for more.

They also know when enough is enough.

They can move quickly.

But they recognise where slowness improves the decision.

This may become one of the great ironies of the AI era:

The machines give us extraordinary powers of generation just as human value moves towards restraint.

Knowing what to build becomes everything

Imagine two people with access to exactly the same AI capabilities.

Both can generate code.

Both can produce designs.

Both can research a market.

Both can create a campaign.

Both can prototype an idea.

Technically, their capabilities may look increasingly similar.

Yet their outcomes could be completely different.

One creates more.

The other creates better.

Why?

Because the second person understands the problem differently.

They recognise what users actually need.

They notice what others overlook.

They bring experience to the decision.

They understand context.

They have standards.

They reject seductive but weak ideas.

They know when the model's recommendation does not feel right — and can explain why.

This moves us towards the next scarcity.

Because if creation becomes abundant, somebody has to decide what is good.

And that brings us to a deeply human capability that may become more valuable precisely because AI is becoming more capable.

Discernment.

Next in the series

Part 3 — Discernment: AI Can Make Almost Anything. Can We Still Tell What Is Good?

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.