Khaberni - Many users note that despite the development and acceleration of capabilities of artificial intelligence tools, they still fail to fully adopt their personal style, even after extensive training. The reason, as revealed through an American journalistic experiment, is deeper than it appears.
For several months, Amanda Caswell, a journalist at "Tom's Guide", tried to coax chatbot AIs to perform what seemed like a simple task: to write in her style. Caswell says she fed the models hundreds of her articles, text messages, and personal emails, corrected the tone, tweaked the phrasing, and explicitly requested imitation of her literary voice.
But the result, as she describes, was always the same: "The text is technically good, the sentences are clear and well-crafted, but the rhythm is off, and the emotions are missing. The writing is polished, but soulless."
And although models like Gemini 3 are remarkably fast in performance, they still stumble over the finest stylistic nuances: the difference between confidence and aloofness, between warmth and sarcasm, and between simplicity and pretentiousness.
Imitating the form.. but not the hidden decisions
Caswell explains that she tested five of the most popular automated chat programs, not to delegate the entire writing task, but to help her with quick daily tasks: drafting an email in her tone, or turning bullet points into a Slack message that genuinely seems to come from her.
And she says, "If you've ever asked artificial intelligence to rephrase a message or emulate your style, you've likely felt that strange disparity, even if you couldn't name it."
Here, precisely, lies the paradox. Artificial intelligence is adept at mimicking sentence length, vocabulary, and even the general rhythm, but it fails to understand what Caswell calls "invisible decisions," those precise choices a human writer makes in every line: what to say, what to omit, and where to stop.
Over-explaining and removing the soul of the text
Comparing her original drafts to the AI versions, Caswell noticed two clear gaps:
First: The problem of omission. Driven by a desire to assist, artificial intelligence tends to over-explain and fill in the blanks, whereas a human writer knows when to stop and leave space for the reader.
Second: A gap in style. Artificial intelligence, by default, tends to adopt a tone that's too polite or overly formal. Caswell says, "It sanitizes the sentences to the extent that it loses their essence, I call this removing the soul of the text."
Great writing breaks the rules.. and artificial intelligence doesn’t dare
Caswell points to a more fundamental issue: while artificial intelligence is trained on rules and patterns, good writing is often defined by a writer's ability to consciously break those rules to achieve an artistic or emotional effect.
And she adds, "The problem isn't a lack of intelligence, but an absence of responsibility."
The human writer feels the weight of words, and the risks of misunderstanding or boredom, while artificial intelligence feels nothing. As a result, it often chooses the safe formula and predicts what a typical writer might say, not what the real writer intends to say.
A literary experiment reveals the difference
To test this practically, Caswell asked artificial intelligence to write a scene from a science fiction novel in which a technician discovers that the AI on a spacecraft sends fake reports to Earth using her voice. The result? "Gemini wrote a summary of the plot, while I wrote the scene."
The AI version, as she says, was functionally correct but cold, resembling a police report: it tells you what happened, but it doesn't make you feel the weight of it. Her version, on the other hand, relied on irony and the latent horror in logic, and on metaphors that can't be taught by data alone.
Artificial intelligence.. just a frozen meal at best
Caswell sums up the experience with a striking analogy: "Writing with artificial intelligence is like heating up a frozen meal. It gets the job done quickly, but you can always tell it apart from a carefully prepared meal."
And she asserts that artificial intelligence isn't useless, as she relies on it for organizing plans, generating ideas, and writing summaries. However, the problem starts when the entire text is left in its hands.
The most important skill in the age of artificial intelligence
With the rapid spread of artificial intelligence, Caswell believes the real skill won't be knowing how to use it, but knowing when to avoid it.
While artificial intelligence may help you work faster, it still lacks the ability to determine what deserves to be said or how it should be said.
