The Words Were Never the Whole Value — Editorial Intelligence

AI has separated the act of writing from the judgement, synthesis and sense-making that writing was hiding.

For most of my career, the finished article was the visible proof that I had done my job.

People saw the headline, the introduction and the carefully edited paragraphs. They did not see everything that happened before the writing began.

The interview where the important answer arrived halfway through an unrelated question. The research document that contained dozens of findings but only one genuinely useful tension. The decision to leave out an impressive statistic because it distracted from the argument. The moment several disconnected pieces of evidence suddenly formed a coherent narrative. The conversations required to understand what an organisation was really trying to say, rather than simply what it wanted to publish.

All of that was compressed into the final piece.

The article was visible. The thinking behind it was not.

For a long time, we treated those things as the same capability. We called the whole bundle “writing”.

Then generative AI arrived and began producing competent prose almost instantly. The immediate reaction was understandable: if a machine can write, what happens to writers?

I think that is the wrong question.

The more significant thing AI has done is not replace writing. It has separated the act of writing from the value that writing was hiding.

Writing was never one capability

A strong piece of editorial work draws on many different forms of intelligence: research, listening, pattern recognition, synthesis, subject knowledge, narrative judgement, structural thinking, audience understanding, editing, taste and the ability to recognise what is important rather than merely what is available.

Writing was the point where these capabilities became visible.

Because they appeared together in one finished piece, it was easy to assume that the value lived in the production of the words. But producing the words was often not even the hardest part.

Once I know what I am trying to say, writing moves quickly. A clear argument creates its own momentum.

The difficult work happens before that: deciding what the piece is actually about.

That distinction was easy to ignore when writing remained slow, specialised and expensive. The finished output acted as a convenient proxy for everything underneath it.

AI has broken that proxy.

It can produce the visible layer without necessarily performing the deeper work that once accompanied it. That makes the difference between words and understanding much harder to ignore.

Competent prose is no longer proof of strong thinking

A well-written article once suggested that somebody had spent time researching, thinking and editing.

That is no longer a safe assumption.

The prose may be polished while the argument remains thin. The structure may be clear while the evidence is generic. The sentences may sound authoritative while the piece says almost nothing that could not have been produced from the first page of search results.

This is why so much AI-generated content feels strangely empty.

The problem usually is not that it is badly written. Often, it is written perfectly well.

The problem is that there was no meaningful act of judgement behind it.

No original evidence.

No difficult choice.

No tension resolved.

No reason for this particular organisation to be the one publishing it.

AI has made competent prose abundant. It has not made understanding abundant.

By making the surface layer easier to produce, it has made the absence of deeper thinking more obvious, not less.

Where the value actually sits

This is what I have come to call Editorial Intelligence: the capability that turns scattered organisational knowledge into useful narratives and repeatable editorial systems.

It begins before the writing, with evidence: research, customer conversations, product expertise and performance data.

It then requires sense-making: finding patterns, identifying tensions and deciding what the evidence means.

From there, it becomes narrative: a clear idea that can organise many pieces of content without forcing every channel to repeat the same words.

Only then does production begin.

When I worked on the Hidden Hours research, the value did not come from turning statistics into paragraphs.

It came from recognising what those statistics collectively revealed: that accountants were performing significant work outside the formal boundaries of their role, often unbilled, and that this hidden contribution was creating both personal and commercial pressure.

The narrative was not sitting inside one statistic waiting to be copied.

It emerged from the relationship between the evidence.

AI could help examine that relationship. It could not decide what the organisation ultimately chose to say.

I found the same thing while building an editorial destination around football club stories and business lessons.

The challenge was never writing the individual club profiles. It was recognising that the stories needed a shared editorial idea, or the page would become a collection of unrelated content.

The work was deciding how different examples — alignment, continuous improvement, pathways and community — could become evidence for a broader argument about the conditions organisations create for success.

Even the wireframe became part of the thinking.

Seeing the content laid out spatially revealed where the narrative was strong, where it repeated itself and where the relationship between the individual stories and the larger thesis needed more work.

Again, the visible output was a webpage.

The value was the sense-making that allowed the webpage to exist.

Two very different kinds of writing work

This is not an argument that every writing role is safe.

Some work genuinely was centred on production: rewriting existing information, creating minor variations of familiar pages, turning a straightforward brief into serviceable copy and producing volume against a predictable template.

AI can now do a significant proportion of that work faster and more cheaply.

Pretending otherwise does not help writers.

But other roles were only described as production roles because organisations had no better language for the intelligence involved.

A writer might have been researching unfamiliar topics, interviewing experts, identifying contradictions, shaping arguments and helping stakeholders clarify what they actually believed.

The job title said writer.

The actual work was much broader.

AI has not eliminated that value. It has made it easier to see.

It has also created an uncomfortable divide.

Writers whose value depended primarily on producing acceptable words will face increasing pressure.

Writers who can work with evidence, exercise judgement and build systems around knowledge may become more valuable.

But they may also need to stop presenting themselves only as writers.

The editor’s role is moving, not disappearing

As the production of prose becomes cheaper, value moves upstream towards the decisions that determine what should be produced in the first place.

Which evidence matters?

What is the argument?

What has been overlooked?

What should not be published?

When first drafts were expensive, editors often spent a large part of their time repairing them: correcting structure, removing repetition and clarifying sentences.

AI can handle more of that baseline work now. That should free editors to spend more time on the decisions machines cannot own.

Is the argument true?

Is the evidence sufficient?

Does this reflect what the organisation genuinely knows?

What has been missed?

What should not be published?

AI raises the baseline.

Editors raise the ceiling.

The strongest editors will not simply polish AI-generated copy. They will design the conditions from which better ideas emerge: building evidence systems, connecting expertise across teams and deciding how knowledge should be expressed and improved over time.

That is not less editorial work.

It is editorial work with more leverage.

Writing still matters

None of this makes writing irrelevant.

The act of writing can sharpen thought and reveal uncertainty in ways that did not exist before the sentence was attempted.

Rhythm matters.

Precision matters.

Voice matters.

But writing can no longer carry the full burden of proving value on its own.

Beautiful prose without insight is still empty.

A recognisable voice without a worthwhile idea eventually becomes performance.

The craft remains important.

It is simply no longer sufficient by itself.

The opportunity for writers

For experienced writers, this can feel like a loss of status.

A capability that took years to build can now be imitated by anyone with access to a model.

But I think there is another way to see it.

AI is revealing how much more many writers were doing all along.

The writer who knows how to ask the question that changes an interview was never merely producing words.

The writer who can find the one meaningful pattern across a hundred pages of research was never merely producing words.

Those capabilities were previously hidden inside the act of writing.

Now they can become the work itself.

That demands a shift in identity: from writer to editor, from content producer to sense-maker, from completing briefs to shaping them.

It will not happen automatically.

Organisations need to recognise these capabilities, but writers also need to become better at naming, demonstrating and taking responsibility for them.

The opportunity is real.

So is the requirement to evolve.

The question is no longer whether AI can write.

It clearly can.

The more useful question is what writing was doing for us that we failed to recognise.

It was giving form to research.

It was making invisible connections visible.

It was forcing decisions.

It was carrying judgement.

AI can help with all of that.

But assistance is not ownership.

Someone still has to decide what matters and take responsibility for what the work claims.

The words were never the whole value.

They were where the value became visible.

Writing was the container.

Understanding was what it carried.