Let’s start with an uncomfortable irony.
LinkedIn — the platform that an enormous number of marketers and business owners now use to publish AI-generated content — has built a system specifically designed to suppress it. In January 2025, LinkedIn’s engineering team published a research paper on arXiv introducing 360Brew: a 150-billion-parameter AI model that now powers the platform’s feed ranking. Unlike the old algorithm, which counted likes and tracked hashtags, 360Brew understands language. It reads your post the way a human editor would — and it has become highly effective at identifying content that lacks a genuine human point of view. [1]
Posts that once pulled thousands of impressions are now sitting at a few hundred. It’s not a glitch. It’s a deliberate architectural decision by one of the world’s largest professional platforms.
The tools everyone rushed to adopt are now being penalised by the very platforms they’re posting on.
If that doesn’t make you pause, it should.
Because it points to something deeper than a platform policy update. It points to a fundamental problem with how most businesses are using AI to write their copy — and why, if you’re one of them, you’re probably getting average results and wondering why.
The Convergence Problem Nobody Is Talking About
Here’s what AI actually does when it writes copy.
It doesn’t think. It doesn’t create. It identifies patterns in vast amounts of existing human writing and produces output that statistically fits those patterns. When you ask it to write a high-converting landing page, it draws on every high-converting landing page it has ever been trained on and produces something that reflects the most established conventions: a strong headline, a pain point, social proof, a call to action.
Which, on the surface, sounds useful.
But think about what happens when every business owner, every marketing team, every freelance copywriter does the same thing. They all feed similar prompts into similar tools and receive output shaped by the same underlying patterns. The landing pages start to sound alike. The email sequences blur together. The LinkedIn posts become indistinguishable.
This isn’t just a quality problem. It’s a competitive one.
Distinctiveness is what makes copy work. It’s what makes a reader stop scrolling, feel something, and take action. And AI is structurally biased against distinctiveness, because distinctiveness by definition sits outside the established patterns AI is trained to reproduce.
You’re not just getting average copy. You’re getting the same average as everyone else.
The Emotion Gap: Where AI Copy Falls Apart
This is the part most people sense but can’t quite articulate.
You’ve probably read a piece of copy — maybe an email, a social post, a product description — and felt that something was slightly off. Not wrong, exactly. Technically fine. Grammatically correct. Structured well. But somehow hollow. Like something was missing, even if you couldn’t name it.
What was missing was felt emotion.
AI can replicate the language of emotion. It has processed millions of examples of human writing about grief, joy, frustration, hope, pride — and it can produce text that contains all the right words in all the right places. But it has never felt any of those things. It has no body. No relationships. No memory of failure or success. No stake in any outcome.
So when AI writes emotionally, it is pattern-matching on how humans have described their experiences — not drawing on experiences of its own.
AI emotion is always borrowed, never owned, never experienced.
And here is the critical point: humans are the sum of their experiences. Every reader brings their own emotional history to the words they read, and at some level — often subconscious — they can sense when the emotion on the page was assembled rather than felt. When the empathy was constructed rather than genuine. When the story was generated rather than lived.
That gap is what roboticist Masahiro Mori called the uncanny valley — a concept he first introduced in 1970 to describe the unsettling feeling people experience when a robot appears almost, but not quite, human. The closer it gets to genuine humanity without reaching it, the more deeply wrong it feels. Mori was talking about machines with faces and moving limbs. But the psychological mechanism is identical when applied to words on a page.
There’s a certain poignancy in the timing. Mori died in January 2025, at the age of 97 — just as AI writing tools reached the point of mass adoption and his valley became one of the defining problems of digital communication. The man who first identified the gap between almost-human and truly human didn’t live to see it become a mainstream marketing crisis. But he named it perfectly, half a century before it arrived.
AI copy falls into that valley every time. It looks right. It reads right. But it doesn’t land the way human copy does, because the emotional raw material simply isn’t there.
Here’s the thing though — this isn’t exclusively an AI problem. It’s a borrowed voice problem.
Think about what happens when hundreds of people complete the same copywriting course and flood LinkedIn with identically structured posts. Hook, insight, three bullet points, call to action. Every time. Or the writer who so thoroughly studied someone else’s style that their copy mimics its surface rhythms without any of its underlying conviction. Readers feel that hollowness in exactly the same way. The uncanny valley isn’t triggered specifically by AI — it’s triggered by the absence of genuine perspective and felt experience, whatever the source.
AI is simply the most powerful and scalable version of that problem we’ve ever seen.
This Isn’t an Argument Against AI
If you’re expecting me to tell you to delete your ChatGPT subscription and hire a copywriter, you’re going to be disappointed.
AI is a genuinely powerful tool. The problem isn’t that businesses are using it. The problem is that most are using it for the wrong part of the process.
AI is exceptional at research — processing large amounts of information quickly and surfacing relevant patterns. It’s good at structure — building frameworks, outlines, logical flow. It’s useful for drafting — turning rough ideas into readable prose. And it’s brilliant at refinement — editing, tightening, improving what you’ve already written.
What AI cannot do is supply the one thing that makes copy actually connect: a genuine, original point of view, rooted in real experience and felt emotion.
That has to come from you first.
The businesses getting poor results from AI copy aren’t failing because the tools are bad. They’re failing because they’ve outsourced their thinking. They’re using AI as a replacement for perspective rather than a tool for expressing it.
A leader doesn’t use tools to do their thinking. They think for themselves, and then use tools to communicate those thoughts more effectively.
And the commercial stakes here are significant. According to the 2025 Edelman-LinkedIn B2B Thought Leadership Impact Report — which surveyed nearly 2,000 management-level professionals globally — 95% of hidden buyers (the internal stakeholders in finance, legal, procurement and operations who quietly shape purchasing decisions) say that strong thought leadership makes them more receptive to sales and marketing outreach. 79% say they’re more likely to advocate for a vendor during the RFP process if that vendor consistently produces high-quality thought leadership. [2]
High-quality. Not high-volume. Not AI-assisted at the expense of a genuine point of view.
A Framework for Copy That Actually Connects
Start with your own thinking. Before you open any AI tool, write down – messily, without editing – what you actually think and feel about the topic. What’s your genuine opinion? What experience have you had that’s relevant? What would you say to a client over coffee that you’d never put in a formal brief? That raw material is irreplaceable. No AI can generate it for you.
Use AI for research and angles. Once you have your own perspective, use AI to stress-test it. Ask it what the counterarguments are. Use it to find data, case studies, and references that support or challenge your thinking. Let it inform your point of view — don’t let it replace it.
Build a structure, not a draft. Ask AI to help you structure your argument, not write it. A strong outline with your ideas in it is far more valuable than a polished draft with AI’s ideas in it.
Write the emotional core yourself. The story. The personal anecdote. The moment of genuine frustration or pride or realisation. AI can help you polish the language afterwards, but the substance needs to come from your actual experience. This is the part readers will remember.
Use AI to refine, not replace. Once you have a draft that reflects your real thinking and experience, AI is brilliant for refinement — improving flow, tightening language, checking for clarity. This is where it earns its place: in service of your voice, not as a substitute for it.
In a market where AI tools are democratising the ability to produce copy at scale, the most powerful thing you can do is something AI structurally cannot: sound like yourself.
The businesses that win the next decade won’t be the ones who used AI most. They’ll be the ones who used it without losing their voice — the ones who understood that AI is a production tool, not a thinking tool, and kept the thinking relentlessly, unapologetically human.
Your lived experience is not a soft asset. In a world of converging AI copy, it is your single greatest competitive differentiator.
Use it.
One final thought. I developed the ideas in this piece through a conversation with an AI — using it to stress-test my thinking, challenge my assumptions, and fact-check my reasoning. But every argument, every opinion, and every formulation I’m proud of? That came from me. That’s exactly the point.
References
[1] 360Brew: A Decoder-only Foundation Model for Personalized Ranking and Recommendation — LinkedIn FAIT team, arXiv, January 2025: https://arxiv.org/abs/2501.16450
[2] 2025 Edelman-LinkedIn B2B Thought Leadership Impact Report: https://www.edelman.com/expertise/Business-Marketing/2025-b2b-thought-leadership-report