.png)
I want you to try something before you read the rest of this.
Open a new tab. Search for anything. Pick a topic you're interested in, a product review, a how-to guide, career advice, a travel recommendation. Read the top five results. Then remove the brand names, the logos, the author bios. Just look at the words.
Can you tell who wrote which piece?
If you can't, you've already experienced the thing I want to talk about today. You just might not have had a name for it.
Researchers call it content homogenisation. The rest of us just feel it, that vague sense that everything online reads the same now. The same structure. The same tone. The same polished-but-empty voice that sounds professional but says nothing memorable.
It's not your imagination. It's measurable. And it's changing the internet faster than most people realise.
The Numbers Behind the Feeling
Let's start with what's actually happening.
Ahrefs analysed nearly 900,000 newly created web pages in April 2025 and found that 74.2% contained AI-generated content. Nearly three out of every four new pages on the internet have AI fingerprints on them.
A Stanford and Imperial College London study found that roughly 35% of newly published websites by mid-2025 were AI-generated or AI-assisted. That figure was essentially zero before ChatGPT launched in November 2022. And here's the finding that matters most for this conversation: pages identified as AI-generated showed semantic similarity scores approximately 33% higher than human-written pages. In plain language, AI-generated content sounds more like other AI-generated content than human writing sounds like other human writing.
Europol warned in a widely cited report that up to 90% of online content could be synthetically generated by 2026. A separate academic analysis estimated that at least 30-40% of text on active web pages already originates from AI-generated sources.
This isn't a future problem. It's a right-now problem. And it affects everyone who reads, writes, or makes decisions based on what they find online.
Why It All Sounds the Same: The Bland Middle
Here's the part most people haven't thought about.
Every major AI language model, ChatGPT, Gemini, Claude, Jasper, and every tool built on top of them was trained on text scraped from the same internet. The same blog posts. The same marketing copy. The same LinkedIn articles. The same industry reports. The same Wikipedia entries.
When millions of people prompt these tools with similar instructions, the output naturally clusters around what researchers call a "statistical mean", a bland professional middle ground that is technically correct, grammatically polished, and completely interchangeable.
A 2025 study published in an academic journal confirmed this empirically. Researchers analysed restaurant marketing content in Milan before, during, and after a temporary ChatGPT ban. During the ban, content showed a 15% decrease in lexical similarity and a 12% decrease in syntactic similarity compared to when ChatGPT was available. Same businesses. Same marketers. Same customers. The only variable was access to the AI tool. Remove the tool, diversity returns.
A meta-analysis aggregating 28 studies involving over 8,000 participants found something equally striking: while humans using generative AI scored slightly higher on individual creativity tasks, AI use substantially reduced idea diversity across the group. Everyone's individual output got a little better. But everyone's output started to converge. The collective became less creative, not more.
This is the paradox at the heart of the homogenisation problem. AI makes each person's content a bit more polished. But it makes everyone's content a lot more similar. The individual wins. The collective loses. And if you're a reader trying to find genuinely different perspectives online, you're increasingly out of luck.
What This Means if You're a Regular Person
You don't need to be a marketer or a technologist to feel the effects of this. Content homogenisation touches everyone in three ways that are worth understanding.
The product reviews you rely on are less reliable. When AI can generate hundreds of plausible-sounding product reviews in seconds, the signal-to-noise ratio collapses. A Gartner survey found that 53% of consumers already distrust AI-powered search results. Only 14% say they fully trust AI-generated content. You're already developing new scepticism habits whether you've noticed it or not, scrolling past the first few results, looking for Reddit threads, checking YouTube for real humans actually using the product. That instinct is your brain compensating for a trust deficit that didn't exist three years ago.
The advice you're getting is narrower than it appears. When AI models are trained on the same data, they tend to converge on the same answers. Researchers call this "knowledge collapse", the narrowing of available perspectives as AI systems regress toward a central tendency. An MIT paper explored how over-reliance on AI-generated information could lead to a collective loss of diverse viewpoints, even as the volume of information increases. More content, fewer ideas. That's the uncomfortable trajectory.
The human voices you used to trust are harder to find. Original reporting, genuine expertise, and authentic personal perspective haven't disappeared. But they're being drowned out by a rising tide of content that sounds authoritative but says nothing original. The internet used to have a quality problem, too much bad content alongside the good. Now it has a sameness problem, too much adequate content that all says the same thing in the same way.
The LinkedIn Effect: You've Seen This Up Close
If you use LinkedIn regularly, you've already witnessed homogenisation in its most concentrated form.
The same post structures. The same "I was walking down the street when I had a revelation" openings. The same bullet-point lessons. The same inspirational closings. The same voice, earnest, slightly breathless, oddly similar regardless of who's posting.
This isn't because LinkedIn users are unoriginal. It's because AI writing tools optimise for engagement patterns that are already proven to work. So everyone's content gravitates toward the same templates, the same rhythms, the same hooks. The algorithm rewards what performs. AI generates what the algorithm rewards. And a platform of millions of professionals ends up sounding like one person with a lot of accounts.
The irony is considerable. LinkedIn is supposed to be the place where your professional voice differentiates you. Instead, AI tools are systematically engineering differentiation out of the platform.
Why This Matters More Than It Seems
Content homogenisation isn't just an aesthetic problem. It's a trust problem, and trust has economic consequences.
When everything sounds the same, nothing sounds credible. When every brand, every professional, every expert sounds interchangeable, the default consumer response isn't engagement, it's scepticism. Research shows that when people suspect content is AI-generated, engagement drops sharply. Not because AI content is necessarily wrong, but because it feels manufactured. And manufactured doesn't earn trust.
There's also a deeper systemic risk. AI models are increasingly being trained on content that was itself generated by AI. Researchers call this "model collapse", a degenerative process where successive generations of models produce progressively more homogeneous, less accurate, and less creative outputs. The internet's content is becoming the training data for tools that produce more of the same content, which becomes training data for the next generation of tools. It's a feedback loop that narrows the range of ideas, perspectives, and expression with each cycle.
As one group of researchers put it: generative systems don't just synthesise existing culture. They actively filter toward homogenisation. The content AI naturally produces when used autonomously is already compressed and generic.
What You Can Actually Do About It
If this all sounds bleak, it's not meant to be. The homogenisation problem is real, but it's also creating a genuine opportunity for anyone willing to do something most people have stopped doing: think for themselves.
As a reader, develop new filters. Look for specificity, personal anecdotes, original data, named sources, concrete examples, opinions that could only come from someone with actual experience. AI-generated content tends to be confidently general. Human expertise tends to be specifically useful. Learn to spot the difference and you'll navigate the internet more effectively than most.
As a professional, your real voice is now your competitive advantage. In a world where everyone can produce polished content in seconds, polished content has no value. What has value is perspective that can only come from your specific experience, your specific thinking, your specific way of seeing your field. The rarest thing online right now isn't quality. It's distinctiveness.
As someone who uses AI tools, use them to think better, not write faster. The people getting the most value from AI aren't using it to replace their thinking. They're using it to challenge their thinking, to stress-test ideas, to find counter-arguments, to explore angles they wouldn't have considered. The output you publish should still sound unmistakably like you. If it sounds like it could have been written by anyone, it will be read by no one.
As a consumer, seek out human-first sources. Newsletters written by people with genuine expertise. Podcasts where real conversations happen. Communities where people share unpolished, honest perspectives. These spaces are becoming more valuable precisely because they're harder to replicate with AI. The premium on authenticity isn't sentimental, it's practical.
The Question I'd Leave You With
The internet was built on the promise of infinite diversity, anyone could publish, any perspective could find an audience, any voice could be heard. For a while, that promise held.
Now we're watching something unprecedented: a technology that gives everyone a voice is simultaneously making every voice sound the same. More content than ever. Fewer genuine perspectives than ever. An ocean of words that is somehow both vast and shallow.
So here's the question I'd put to anyone reading this: in a world where AI can produce adequate content on any topic in seconds, what is the thing that only you can say? What perspective, experience, or insight do you have that no prompt could generate?
If you know the answer, protect it. Develop it. Share it loudly. Because that is the one thing the internet is running short of, not content, not information, not polish, but genuine human thought.
And genuine human thought, unlike content, cannot be produced at scale. Which is exactly what makes it valuable.
Marketing that thinks. Not marketing that trends.
Sources
Ahrefs: AI Content Prevalence Study (900,000 pages; 74.2% AI-generated content, May 2025)
Stanford / Imperial College London / Internet Archive: 35% of new websites AI-generated by mid-2025; 33% higher semantic similarity
Europol: Synthetic media projection (up to 90% of online content by 2026)
arXiv: "Delving Into: The Quantification of AI-Generated Content on the Internet" (30-40% of active web text AI-generated, March 2025)
Graphite: AI content study (65,000 URLs; AI-written articles surpassed human-written by November 2024)
SAGE Journals: "When Artificial Intelligence Makes Everything Similar: The Risks of Content Homogenisation" (March 2026)
SAGE Journals: "The Homogenizing Engine: AI's Role in Standardizing Culture" (December 2025)
SSRN: Milan ChatGPT ban study (15% decrease in lexical similarity, 12% decrease in syntactic similarity)
Meta-analysis of 28 studies (8,214 participants): GenAI reduces idea diversity (Hedges' g = −0.86)
MIT: "AI, Human Cognition and Knowledge Collapse" (February 2026)
arXiv: "Epistemic Diversity and Knowledge Collapse in Large Language Models" (2025)
Gartner: 53% of consumers distrust AI-powered search results (June/July 2025 survey)
NP Digital: Consumer trust survey (61% "somewhat" trust AI content; only 14% fully trust, July 2025)
Attest: 2025 Consumer Adoption of AI Report (43% trust AI chatbot information)
Carat / Anna Campbell: "AI homogenisation" in industry; 2026 as year of differentiation
Linkfluencer: "Why AI Content Isn't Enough in 2026" (February 2026)