Why does AI write like that?

You have probably already read its work today. A product description that feels generic, a brand newsletter that sounds close to that one from last year, just a bit different. Its prose is smooth and competent, but flat. This is the sound of AI writing. It is not ‘bad’ writing in a traditional sense. It is something else, a new dialect. And it’s becoming so commonplace that it’s affecting the way we speak and the way we sell. This dialect is becoming the sound of the modern marketing industry, and its implications run deeper than efficiency

It is only natural to find something alien so odd. Artificial Intelligence, as a language model, is trained on input data, scouring human language for the most common words. It learns probability, not meaning. It masters pattern, not purpose. Therefore, its voice is not a voice at all. It is average. It is the linguistic means of countless blog posts, annual reports, and website copy. The French novelist Gustave Flaubert famously tortured himself searching for le mot juste, the single exact word. AI operates on a different principle. It selects le mot probable, the most statistically likely word. The result is a text without a source, a message without a messenger. It has the shape of communication but not the heat.

Art versus annotation

This synthetic output finds a perfect antithesis in literature’s celebration of the messy self. Consider the narrator of Fyodor Dostoevsky’s Notes from Underground. He is spiteful, illogical, and vibrantly human. His voice is a weapon against a world demanding rational, polished conformity. An AI could never generate his chaotic brilliance. It could only produce a clinical summary of his condition. The former is art. The latter is an annotation. This distinction matters profoundly for marketing, a field built on connection. We connect to voices, not to summaries.

The volume trap

Why then does this average voice proliferate across agencies and content plans? In an industry that rewards output and quotas, it is not surprising that so many of us are turning to AI to offer a helping hand. Writing is hard. Research from the Stanford Institute for Human-Centred AI indicates most professionals now use these tools to manage overwhelming content demands and to bypass the paralysis of a blank page. The tool offers a lifeline in a sector obsessed with scale and speed. The first time I used AI, Chat GPT specifically, was during an internship in France to parse complex financial articles I couldn’t understand, but had to brief clients on over email. It was horrifically easy, and instantly stopped the anxiety swelling in my stomach that I wouldn’t be able to finish the task on time, or to a good enough degree. The ethical unease begins here, not with the use itself, but with the conditions that make it indispensable. When the metric is volume, the average becomes the standard.

Two quiet costs follow this adoption. First, a narrowing of discovery. When we ask an AI to summarise, we accept its algorithmic curation of sources. It reinforces the dominant narrative, the already popular idea. It does not stumble upon the obscure reference, the contradictory study, or the niche perspective that sparks true innovation. Our research becomes a closed loop, polished and insular. Second, and more critical for marketing, is a trust deficit. Data from a 2024 Tessian report shows most UK consumers instinctively distrust brand communications that feel generically automated. The audience senses the void. We are wired to seek the fingerprint of human thought, the slight imperfection that signals authenticity. Its absence triggers disengagement, not delight.

Thinking out loud

The central question is therefore not whether the writing is good. It is about what we surrender in the process. Language is not merely a vessel for thought. The struggle to shape a sentence is the struggle to shape the idea itself. To outsource this compositional act is to outsource a part of thinking. Can a process designed to reassemble existing fragments ever produce a genuine new whole? The form is not just the container. It is the thought.

This synthetic average is already altering our landscape. A study published in Science

noted a detectable shift towards homogenised language in scientific and marketing

preprints, a subtle erosion of lexical diversity. The accent is spreading.

Yet a stubborn human preference remains. Observe the recent public reaction to two

campaigns. Coca-Cola's "Masterpiece" advertisement, a spectacle of AI-generated

imagery was met with global apathy. It was a technical feat that resonated as a cold trick. Contrast this with the warmth generated by Apple’s "Underdogs" commercial for the Macintosh, which celebrated tangible, hands-on creation with practical effects. The audience did not just see the ad. They felt it. One was manufactured. The other was made.

Reserving the soul

The future path is not to reject the tool but to define its role with precision. Let this synthetic voice handle the genuinely repetitive, the data-driven update, the first draft of the generic. But we must consciously, even fiercely, reserve the spaces that matter for the human voice. The brand story that builds a legacy, the creative concept that requires a leap, the copy that needs a soul, these are not tasks for the average. They are tasks for a person. The goal is not to make the machine sound more like us. It is to use its capabilities to free us to do what it cannot. People do not crave flawless perfection. They crave the authentic spark of another mind.

And that, for now, remains a uniquely human offering.