AI has solved a problem the creative industries spent decades complaining about.
It can make more.
More copy. More images. More layouts. More concepts. More edits. More variations. More versions of the variation requested twenty minutes before the presentation.
We have achieved abundance.
Congratulations.
Now what?
Because once making becomes dramatically easier, the value of making begins to move somewhere else.
Towards choosing.
Towards editing.
Towards taste.
Towards judgement.
Towards knowing why one idea deserves another hour and nineteen others deserve the bin.
The question is no longer whether machines can produce creative material.
Clearly, they can.
The question is whether production was ever the most valuable part of creativity in the first place.
We are confusing output with ideas
Generative AI is extraordinarily good at output.
Give it a sufficiently clear brief and it can generate a hundred headlines, dozens of visual directions, scripts, storyboards, naming territories or campaign routes in minutes.
Sometimes one will be good.
Sometimes seventeen will look good at 11:45 p.m. and considerably less good the next morning.
Creative people will recognise this experience. We had it before AI. It was simply slower.
The machine changes the scale of possibility.
It does not remove the need to choose between possibilities.
Canva's 2026 State of Marketing & AI research, based on 1,415 marketing leaders and 3,547 consumers, found that 97% of marketing leaders use AI in daily creative work and 99% plan to increase AI investment during 2026.
So the adoption debate is essentially over.
The more useful conversation starts afterwards.
What exactly are we doing with the additional capacity?
The average has become very, very good
Generative AI has an unusual effect on creative standards.
It raises the floor.
Someone with relatively little technical experience can now produce a competent visual, a respectable presentation or perfectly serviceable copy in a fraction of the time it once required.
That is a real democratisation of production.
It is also why so much work is beginning to feel strangely familiar.
Not bad.
Familiar.
Generative systems learn from enormous bodies of existing material. Given a conventional brief, they often converge on familiar patterns.
And those patterns have become remarkably polished.
The result is a new creative danger:
professional-looking average.
Canva's research captures the audience side of that problem. 70% of consumers surveyed said AI-generated advertising often feels as though something is missing. 78% would still rather see advertising made by people, while 87% believe the best advertising requires a human touch.
People are not necessarily identifying a technical flaw.
They are responding to a feeling.
Something too complete.
Too expected.
Too correctly assembled.
The visual equivalent of a person at dinner who says all the right things and somehow leaves no impression whatsoever.
Taste is not decoration
We often talk about taste as though it were an aesthetic accessory.
Nice fonts. Good photography. Knowing which chair is currently appearing in every Copenhagen apartment.
Taste is much more useful than that.
In creative work, taste is a decision-making system.
It is the accumulated ability to recognise the difference between:
Taste is built through exposure, experience, curiosity, mistakes, arguments, rejected work and paying attention for a very long time.
It is why two people can use exactly the same tools and produce profoundly different results.
John Berger opened Ways of Seeing with four words:
“Seeing comes before words.”
The line was written in 1972, but it feels unexpectedly useful now.
Before we articulate why something works, we often recognise that it does.
A creative director looks at twenty routes and pauses at one.
A strategist hears a sentence and knows there is an idea inside it.
A designer removes the element that everyone else has been trying to improve.
That judgement may eventually be explained rationally.
But its source is rarely a checklist.
AI can rank. Someone still has to define relevance.
AI can evaluate options.
It can score work against criteria, analyse performance data, identify patterns and help compare possible directions.
We use it for exactly these things.
But relevance always contains an objective.
An AI system can optimise against the criteria it receives. The strategic act is deciding which criteria deserve to govern the work.
That decision belongs upstream.
The World Economic Forum's Future of Jobs Report 2025 makes an interesting companion point. AI and big data are among the fastest-growing skills, but creative thinking, analytical thinking, curiosity, leadership and other human capabilities are also expected to grow in importance through 2030.
The future is not shaping up as a contest between technical ability and human judgement.
It increasingly requires both.
Plausibility is not the same thing as truth
Something can sound convincing without being worth saying.
Philosopher Harry Frankfurt examined exactly this territory in On Bullshit. He distinguished lying from communication produced with indifference to whether it corresponds to reality. Princeton's summary of his work describes this as an “indifference to the truth.”
He was not writing about generative AI.
Obviously.
But the distinction has become rather useful.
Language models are exceptionally capable of producing plausible language. Unless grounded in reliable information and properly directed, plausibility can easily be mistaken for knowledge.
The same thing happens creatively.
Something can look like a campaign.
Sound like a brand manifesto.
Read like thought leadership.
Contain all the correct ingredients.
And still have absolutely nothing at stake.
This is why the prompt is only part of the work.
Someone still has to know enough to notice when the answer is empty.
Consumers are noticing too
The trust question is becoming difficult for brands to ignore.
Gartner reported in March 2026 that 50% of US consumers surveyed would prefer to buy from brands that do not use generative AI in consumer-facing content. 68% said they frequently wonder whether the content and information they encounter is real.
That does not mean brands should stop using AI.
It means invisible efficiency and visible syntheticity are not the same proposition.
The question should not be:
Where can we use AI?
The answer is increasingly: almost everywhere.
A better question is:
Where does AI genuinely improve the work, and where does it remove the thing that made the work worth noticing?
Efficiency should buy thinking time
One of the strangest possible outcomes of the AI revolution would be to save enormous amounts of time only to spend that time producing enormous amounts of additional mediocre content.
Yet this is entirely possible.
If AI turns a four-hour production task into forty minutes, there are at least two ways to use the remaining three hours and twenty minutes.
Make six more things.
Or think harder about the first one.
We are much more interested in the second option.
Use AI to explore territory faster.
Use it to interrogate research.
Use it to test assumptions.
Use it to prototype.
Use it to make adaptation less painful.
Then spend the time you saved on the parts that still benefit from attention.
The brief.
The idea.
The uncomfortable question.
The unexpected reference.
The edit.
The decision to start again.
AI, with direction.
At Recreative Digital, we do not see AI as a separate creative discipline.
It has already become part of the environment in which creative work happens.
We use it across research, strategy, ideation and production.
But there is an important distinction between using AI to expand possibility and asking it to replace direction.
The former is powerful.
The latter tends to give you exactly what you requested.
Which is sometimes the problem.
The advantage of the next generation of creative businesses will not come from having access to AI. Access is rapidly becoming universal.
It will come from knowing where to point it.
Knowing what to keep.
Knowing what to reject.
Knowing when the technically impressive answer is strategically irrelevant.
And knowing when the best prompt is followed by:
“No. Again.”
AI can produce more.
That part is settled.
What matters now is who is deciding what matters.