Opinion | The Great AI Writing Trap: Why Everyone Suddenly Sounds The Same

I was at a business conference recently, listening to some of India's brightest minds speak one after another. It was the kind of event where you come prepared to be bored and awed in the same day. One after another, speeches written with corporate jargon, waiting for that one moment in each speech that was clearly written to be remembered.

But this time, I found myself getting bored faster than I used to. Not because the speakers were bad, they really weren't, but because somewhere in the middle of it all, I realised every speech was following a pattern. Every speaker sounded way more similar than they used to.

I started paying attention after that to this pattern. Speeches, articles, social media posts by people with verified accounts and decades of experience - all of it carried a rhythm, like it's a template - the same way of building tension and arriving at a conclusion. So, I did something that we all do these days when we have a thought: I asked ChatGPT why this was happening. The irony of doing that is not lost on me.

The answer, to its credit, was way more honest than I expected. It admitted that it is obsessed with the contrast framing of sentences. It is rhetorical and takes control of the reader's mental model.

AI has become a genuinely useful writing tool. It fixes grammar, tightens arguments, and helps organise thought. For people writing in a second or third language, it does something more valuable still, bridging a gap that used to require years of immersion or an editor with the time and patience to help. That is a real benefit, and an underappreciated one. But the same tool is producing a problem that is harder to see when you're writing, but obvious to a reader or a listener: it is making intelligent people sound identical.

The evidence is in the constructions we can all now identify. "This is not a challenge; it is an opportunity." "India is not a market, it is a mission."

"Immigration is not about borders. It is about identity." "AI is not about technology. It is about power."

"Manufacturing is not about factories. It is about sovereignty."

"The question is no longer whether, but how."

Once you start noticing these formulations, you cannot stop. They are in speeches, opinion pieces, annual reports, social media threads, and television debates. The first time, they sound persuasive. By the hundredth time, they are furniture. And increasingly, they are something worse than furniture. AI writing is no longer just identifiable. It has become something people laugh at, and the people producing it are often the last to know.

The reason is an unexpected one. The problem with AI writing is not that it is bad. It is that it is too perfect. Too free of the hesitations and instincts that make writing feel like it came from a person working something out. Human thinking and, by extension, writing, is a product of friction. The odd stumble, the sentence that goes somewhere unexpected, the argument that admits its own uncertainty. AI writing has none of this. Every sentence arrives fully formed. Every problem becomes a clean contrast. Every paragraph strains toward profundity whether or not it has earned it. Readers have started to feel this frictionlessness as a tell, and once they feel it, the mockery follows naturally.

Underneath the style problem is a thinking problem, and this one is harder to paper over.

AI-assisted writing does not just reveal writing habits; it reveals intellectual habits. A person with strong ideas will use AI to sharpen and organise them, and the output will be better for it. A person with weak ideas will use it to dress them up, and the dressing will not hold. The machine amplifies whatever is already there. If the underlying thought is thin, robot-written linguistic polish alone can't carry it for long.

What ChatGPT very honestly explained to me was this perfection in speech is part of how AI functions and not something you can override with the help of prompts, as some LinkedIn influencers would like you to believe. The reason this happens is mathematical. Large language models are prediction engines. At every step, they calculate the most probable next word based on billions of examples. If millions of speeches use the construction "not just X but also Y", the model learns it is a reliable way to sound persuasive. If thousands of articles end with "the real question is", the model learns that is how opinion pieces close. Words like "transformation", "purpose" and "future-ready" accumulate high probability scores because influential people keep using them, and the model keeps feeding them back. It is not trying to be original. It is trying to be statistically correct.

The kind of conference where you arrive expecting to hear something that could only have come from a particular person, a specific observation, an argument built from lived experience, is exactly where this AI writing falls apart.

The best thinkers and writers have always worked against statistical probability. They choose the unexpected detail, a raw emotion, the description of reality as they see it rather than as the pattern predicts. That instinct is not something AI can replicate. It can only reflect what has already been said, more smoothly than before. Which is why the people who use it will never let it speak for them; they let it assist them but not drive them. AI can be an editor, a researcher, a tool for helping you think or even sharpen your ideas with the right words. But those who let it do the thinking will find its limits soon enough, probably in the form of an audience that has quietly stopped believing them.

Before hitting generate, there is one question worth sitting with: what do I know, or see, or understand, that the machine cannot access? That is the only part that is original. The rest is AI homogenisation, full of cliches and, finally, stylistic collapse.

(Sanjay Pugalia is the CEO of AMG Media Network)

Disclaimer: These are the personal opinions of the author