There’s a kind of text that’s become omnipresent this past year. It isn’t bad. It’s well structured, it has no typos, it answers the question. And it’s exactly the same as a thousand others. You recognise it instantly: it sounds like everyone and no one. It’s the default prose of a machine that has read the whole internet and learned to hand back its average.
The problem isn’t that this text is poor. The problem is that it’s now free. Correctness —writing competently— has stopped being an advantage, because anyone with an account and a good prompt has it.
The mistake of competing on the wrong ground
Faced with that tide, the reflex is to produce more of the same, better optimised. More articles, more keywords, more flawless structure. But you don’t win a sameness contest by being a little more the same as everyone else. Optimising the same text everyone writes, only harder, just turns you into one more indistinguishable result inside a list of indistinguishable results.
Scarcity has moved. It’s no longer information that’s scarce, not even decent prose. What’s scarce is a point of view. A voice. The willingness to say something an averaging machine would never say, because averaging is, by definition, filing down everything that stands out.
Four signals of where the web is heading as it fills up with automated content.
74%
of new pages already carry AI-generated content
Ahrefs · 900,000 pages, 2025
68%
of Google searches end without a click to the open web (it was 60% in 2024)
SparkToro · Similarweb, 2026
20%
of a page's text the average visitor reads
Nielsen Norman Group
−15/25%
organic traffic has fallen in many sectors
Bain & Company, 2025
The more everything looks alike, the less people click and the less they read. What's scarce —and what earns the citation— is the voice that isn't mistaken for the average.
What it really means to be different, brave and divergent
They aren't decoration. They're ways of thinking, and each one has consequences.
Finding the angle nobody took: a frame that reorders how the reader sees the subject, not a fancier synonym for the usual.
Holding a claim you'd have to defend. Saying "not this" when the consensus says "this." Taking a stand.
Thinking outside the rails: seeking the edge of the subject instead of its centre, cross-wiring concepts that rarely sit together. It isn't rambling; it's triangulating from different places to reach a new one — what makes a text memorable and worth re-reading.
The myth worth dismantling
Here comes the usual objection: “that’s lovely, but such a distinctive voice doesn’t rank.” It’s false, and it’s worth understanding why.
Google has spent years rewarding exactly this. Its experience, expertise, authoritativeness and trust criteria —the well-known E-E-A-T— and its declared war on mass-produced content point in one direction: demonstrable originality and expertise at the top, uniformity at the bottom. A voice with a point of view isn’t a problem for the search engine; it’s what the search engine says it’s looking for.
And generative answers —what we already call GEO, optimisation for AI engines— work by citing. The AI synthesises many sources and attributes the sentence that deserves attribution: the sharp one, the authoritative one, the one you can put in quotation marks. Soft text dissolves inside the synthesis and disappears. Text with an edge gets cited by name. Being citable is, today, being findable.
Put differently: thinking that diverges —that crosses borders and dares to hold a thesis— isn’t a tax on visibility. It’s what earns the link and the citation, which is the new currency.
The honest caveat: different isn’t chaotic
It’s worth not getting confused. Being divergent isn’t rambling. Being brave isn’t being reckless. Being different isn’t being odd for the sake of it. A voice without craft is just noise, and noise doesn’t rank either.
That’s why the text’s own trajectory matters: that it has structure, that the reader gets somewhere, that each section genuinely answers a question. That discipline —headings that say something, paragraphs that breathe, one idea per block— is what lets the voice land instead of scattering. It’s the same clarity the reader appreciates and the machine needs to understand you. Voice and method don’t compete: they hold each other up.
Stories have shape
There’s a lovely example of all this. Kurt Vonnegut argued, half in jest and half seriously, that stories have shapes you can draw on a graph: time on the horizontal axis, the protagonist’s fortune on the vertical. And out come recognisable silhouettes —the “man in a hole,” who falls and climbs back up; “boy meets girl”; Cinderella, who rises in leaps until she goes off the top of the chart.
His idea, brought here, is simple: a text isn’t a flat line. It has a trajectory the reader notices even without being able to name it. That’s the arc —and it’s worth hearing him tell it himself.
Kurt Vonnegut explains the shapes of stories. Open on YouTube ↗ Playing the video will load content from YouTube, which may store information in your browser. Read our cookie policy.
And where does AI fit in?
In the place it belongs: that of an extraordinary tool. It speeds up research, proposes drafts, sorts, summarises, saves you the blank page. Use it without hang-ups.
What it can’t give you is judgment about what’s worth saying. The angle. The courage to chase the unexpected crossing when the easy thing would be to go straight. The ear for the line that closes it. AI hands you the average of everything ever written; your job starts where that average ends. The intelligence that makes the difference is still human.
Where to start
If you publish content for your organisation, try reversing the order. Don’t start by asking “how do I rank for this keyword.” Start with another one: what do we know, what have we seen, what do we believe no one else can tell the same way. That’s where the voice is. The rest —the clear structure, the well-answered question, the text the machine finds and cites— comes after, and it’s built without giving up anything.
Being brave and being found used to look like opposite paths. In 2026, they’re increasingly the same move.
Sources
- Ahrefs, Ryan Law, Xibeijia Guan and Tim Soulo, 74% of New Webpages Include AI Content (Study of 900k Pages), 19 May 2025. 900,000 new English-language pages, one per domain, detected in April 2025; 74.2% contained AI-generated content.
- SparkToro, Rand Fishkin, In 2026, Less than One Third of Google Searches Still Send a Click, 9 June 2026. Similarweb clickstream panel, January to April 2026, United States only: 68.01% of Google searches end without a click, up from 60.45% in 2024.
- Nielsen Norman Group, Jakob Nielsen, How Little Do Users Read?, 2008. 45,237 page views; the average visitor reads about 20% of the text. It is an old measurement and remains the reference because it has not been repeated at that scale.
- Bain & Company, Goodbye Clicks, Hello AI: Zero-Click Search Redefines Marketing, 19 February 2025. Bain-Dynata survey, December 2024 (1,117 consumers); the 15% to 25% drop in organic traffic is the authors’ estimate, not a measurement.
- Google Search Central, Creating helpful, reliable, people-first content, current guidance on experience, expertise, authoritativeness and trust (E-E-A-T).
- Kurt Vonnegut, The Shapes of Stories, recorded lecture.