We have always mistaken eloquence for intelligence. It is an easy mistake to make, for the assured use of written language has indicated that someone is in control of one’s thoughts and is able to translate them into language, in a way that makes the working of their minds apparent.
We read logic into narrative coherence, effort into a well-crafted paragraph. Well-written prose carried with it the sweat patch of effort, the creased forehead of concentration. Language communicates at more than one level. Apart from communicating its content, it also tells us something about the user of language. The well-sculpted paragraph implied friction: time, revision, editorial constraint, reputational risk. Even when it didn’t guarantee truth, it suggested that someone had paid something in terms of time and effort to make the thought intelligible.
And then AI has come and spoiled it all. It has made the written word into an indicator of very little except itself. Fluency is no longer a sign of intelligence or, for that matter, even of thought. The paragraph that sounds measured, thoughtful, and complete can be produced at will. Not only can the tone be calibrated: earnest, outraged, wise, playful, sceptical, we can even write in the manner of whosoever we wish to emulate. The coherent argument is now a mass consumer product, one that comes free with ChatGPT.
But when a thing becomes easily reproducible, it stops serving as proof of anything except reproducibility. It is not as if we are exactly new to language detached from meaning. Corporate PR speak, the speeches of politicians, the glowing statements of support from industry leaders after every Budget speech, posts on LinkedIn (which is the Grand Central of polished banalities)— all of these have demonstrated to us that just because a statement is carefully crafted, it does not mean that it is actually saying anything.
What AI changes is scale and indistinguishability. Earlier, detached language still wore recognisable costumes. We knew what a standard press release sounded like; we could strip off the verbiage to get the drift of the message. The costume itself gave the game away.
Now there is no recognisable costume. Any voice can be approximated and appropriated. We can signal earnestness, gravitas, sarcasm, we can turn philosophical or lofty, without putting anything of ourselves behind it. The sentence can be thoughtful without any thinker behind it.
What this does is to take away one way we made sense of the world. For language and its use have become nondiagnostic. One could argue that this is a good thing, for it reduces the power that the glib and the silver-tongued wielded simply because of their mastery of language, and it levels the playing field for everyone.
We use a number of shortcuts to process the world because otherwise the cognitive load of reality is simply too high. Language as a sign works on two levels— through what it communicates on the surface and through what its manner of use signifies. What is being judged is not just intelligence but also sincerity and the intention behind what is being said. When language becomes a grey cloak that paints everyone with the same brush of empty narrative fluency, it stops being a sign of the writer’s intention. Better communication feels like a shortcut to erasing one’s personhood. One is better understood even as one becomes more invisible.
On the receiver’s side, the surface polish of the communication feels like a packaged product. Purpose speaks for itself instead of speaking for the writer. Over time, the polish serves as a slippery surface on which meaning glides rather than rests. When everyone starts speaking in the same register, using the same narrative devices, there is a sense of the world closing in around you. It is a form of prison, this linguistic bland sameness, from where escape seems difficult.
Reading AI-generated text is a little bit like travelling on a too-smooth road; for those unused to it, there is a giddiness that comes from overly smooth motion. AI language performs all the time, and it does so by removing specificity. The things that gave writing a distinct voice— hesitation, long-windedness, and imprecision are lost. In a sense, machines are writing to each other using human beings as an interface. What language thrives on is constraint. Writing after experiencing the world imposes limits on what can be claimed. Certain arguments are foreclosed, certain conclusions unviable. AI has no such boundaries because it has not encountered any limits to its imagination. It goes where the next word takes it. It is fluent because it is unconstrained.
Today, the real test lies not in what is claimed but in how it is defended, argued and built upon. Sounding smart is easy, but working with one’s professed smartness requires one to own one’s thoughts. What matters is how this mind behaves when challenged? Does it revise? Does it clarify? Does it concede? Seriousness may have to be inferred from what happens after the sentence, not from the sentence alone. Truth might have migrated to the land of the extempore response. AI has not impoverished language. It has made language abundant. It has not removed meaning. We will continue to speak, and will speak more than ever. But we are beginning to sound robotic, writing immaculate prose in a metallic voice.
Disclaimer
Views expressed above are the author’s own.
END OF ARTICLE
