Everyone writes better now. That is the problem.

Generative AI has made everyone a competent writer. Grammar is cleaner. Sentences are smoother. Vocabulary is broader. And none of it matters, because when everyone sounds the same, sounding good is no longer a differentiator. It is camouflage.

The paradox is almost too clean to believe. The tools designed to improve professional writing have, in aggregate, made most professional writing worse. Not grammatically worse. Structurally identical. The sentence rhythms match. The hedging patterns match. A generation of professionals now writes with the fluency of a competent copywriter and the distinctiveness of a hotel welcome card.

That is not an improvement. That is a new kind of noise.

The Batch That Broke

I noticed it first during a product brief review at a company where I was advising. Three product managers had submitted briefs for three different initiatives. Different markets, different user segments, different strategic priorities. But the briefs read like they had been written by the same person. The same sentence structures. The same way of introducing a problem statement. The same careful, polished, entirely unremarkable prose.

I asked one of the PMs whether she had used a writing tool. She had. She said it made her briefs "more professional." She was right. The prose was professional. But it was also indistinguishable from the other two. The professionalism had come at the cost of the one thing a product brief needs to do: convey a specific person's understanding of a specific problem. Her thinking was somewhere inside that document. But it had been smoothed into something no one would argue with, which meant no one would remember it.

I started watching for the pattern, and once I saw it, I could not stop seeing it.

LinkedIn posts from product leaders began reading like they were generated from the same prompt. Slack updates across teams I worked with developed an eerie uniformity. Investor updates from founders I mentor started arriving in prose so clean and measured that I could not tell whether the founder was excited, worried, or merely complying with a quarterly obligation. The voice had been polished away. What remained was text.

The Sameness Tax

I call this the sameness tax. It is the price you pay when your communication is technically correct but personally absent. The sameness tax is not measured in errors. It is measured in attention. When every product brief, stakeholder update, and internal memo reads like every other one, readers stop reading for meaning and start scanning for compliance. They skim. They check boxes. They move on. The writing has become furniture. Present, functional, and entirely invisible.

But the sameness tax compounds. When your writing sounds like everyone else's, you lose the ability to signal what you actually think, as distinct from what a competent professional might think. A product leader's job is not to produce correct prose. It is to produce clear thinking in a voice others can trust, challenge, and respond to. When the voice disappears, so does the trust signal. You are not communicating. You are filing documents.

The thing nobody warned us about AI-assisted writing is that it optimises for the average. It makes your worst writing better and your best writing blander. It lifts the floor and lowers the ceiling simultaneously. But most professionals only notice the floor lifting and mistake that for progress.

What Voice Actually Is

Voice is not style. It is not vocabulary choices or sentence length preferences. Voice is the residue of a specific mind thinking through a specific problem, and it shows up in the patterns that no tool would generate because they are too personal, too shaped by individual experience to be predicted by a model trained on the aggregate.

At Freshworks, there was a product director who wrote the most distinctive stakeholder updates I have ever read. They were not literary. They were not polished in the way that AI prose is polished. But they were unmistakably hers. She had a habit of starting updates with the single most uncomfortable number from the previous quarter, no context, no framing, just the number. Then she would spend two paragraphs explaining why that number was the one that mattered. Her updates were the only ones that senior leadership actually discussed in meetings rather than acknowledged and moved past.

Her voice was not a personality trait. It was a strategic asset.

But here is what has changed. Before generative AI, distinctive voice was nice to have. Now it is the primary signal that separates human thought from generated text. When a founder sends me an investor update that reads like every other update I have seen that month, I do not think "this person writes well." I think "this person used a tool and did not edit for voice." The polish itself has become the tell.

Voice as signal is the concept that matters now. Your distinctive way of communicating is no longer a stylistic preference. It is the primary evidence that a human being with specific knowledge and specific judgement produced this text. Without it, your writing joins the growing pile of competent, indistinguishable prose that nobody reads twice.

The Intern Problem

I ran into this most starkly while mentoring interns. I had asked three of them to write up findings from a user research sprint. Different products, different user segments, different insights. The write-ups arrived within an hour of each other.

They were identical in everything but the specifics.

Same sentence structure. Same transitional phrases. Same way of hedging uncertainty. Same slightly formal register that no twenty-two-year-old actually speaks in. I asked them directly. All three had used AI tools to "clean up" their drafts. The cleaning had removed every trace of the original thinking. One of them had written a rough first draft that was messy, specific, and full of observations that surprised me. The AI-polished version was correct and forgettable. She had taken something interesting and made it professionally invisible.

But I could not blame them. Every signal in the professional world tells junior people that polished equals competent. Nobody had told them that polished and indistinguishable equals ignored.

The most valuable writing skill in a world of AI is the one AI cannot replicate: a point of view.

The Premium on Being Recognisable

At Schneider Electric, I worked on building automation products, thermostats and energy management systems for facility managers who were not technical users. One of the hardest design challenges was making our interface distinct from competitors using the same component libraries and the same interaction patterns. Everything looked competent. But nothing looked specific.

Professional communication is approaching the same problem. When every tool draws from the same training data and produces prose in the same register, the result is communication where nothing is wrong and nothing stands out. But standing out is not vanity. It is function. A product brief that sounds like every other product brief will be processed like every other product brief. A brief that sounds like a specific person thought hard about a specific problem will be read, argued with, and remembered.

The premium is not on writing well. The premium is on being recognisable.

But recognisability requires risk. It requires leaving in the rough edges that a tool would smooth away. It requires writing a sentence that is slightly too direct, slightly too shaped by your own experience to be mistaken for generated text. It requires trusting that the reader will value your thinking over your grammar.

The tools will keep getting better at producing competent text. But competent text is not what leadership communication requires. It requires conviction, specificity, and the willingness to sound like exactly one person in a world that is rapidly learning to sound like everyone.