Glass Half Empty
9/9/2025
I found myself hesitating the other day, my fingers hovering over the keyboard before posting a comment on a forum. It was a perfectly innocuous comment—a few sentences strung together with what I hoped was a modicum of wit and a correctly placed semicolon. But a strange, creeping dread washed over me, a feeling entirely new to the digital age: Will they think a bot wrote this?
This is the peculiar purgatory we now inhabit. For decades, the Turing Test was a benchmark for machine intelligence: could a computer fool a human into thinking it was also human? Now, in a bizarre and tragicomic inversion, we humans find ourselves desperately trying to pass a reverse Turing Test every time we write. We are no longer the judges; we are the suspects, frantically trying to prove our own authenticity in a world awash with algorithmically generated text.
It’s a chilling thought, particularly for those of us who have spent a lifetime wrestling with language. The act of writing—of choosing one word over its near-synonym, of structuring a sentence for rhythm and impact, of labouring over a comma—was once a fundamentally human pursuit. Now, it risks becoming a liability. The well-turned phrase is no longer evidence of a thoughtful mind, but a potential signifier of a sophisticated language model. The very skills we cultivated are now the fingerprints of the ghost in the machine.
The common refrain from the technically-minded is that this is all overblown. “It’s just a Markov chain on steroids,” a friend told me, dismissively. And he’s not entirely wrong, though the steroids in question are of Olympian potency. At its most rudimentary, a Markov chain is a model where the future state depends only on the present. In language, this means the next word is predicted based on the last word. A simple first-order Markov model might see the word “rolling” and predict “stone” with a certain probability, based on a training corpus. It’s a simple chain of conditional probability, say, the probability of the next word given the current word :
Of course, today’s Large Language Models are vastly more complex. They don’t just look at the last word; through the sorcery of transformer architectures and attention mechanisms, they consider the context of every word that came before, weighing relationships across entire paragraphs. They are not mere chains, but vast, interconnected webs of statistical likelihood. And yet, the core principle remains. They are breathtakingly sophisticated prediction engines, calculating the most probable sequence of tokens to follow a prompt. They are not thinking; they are completing a pattern.
But whence comes our despair? It’s not merely the fear of being replaced in the workforce, though that is a palpable and legitimate anxiety. No, the deeper horror is one of meaning. The “worse of the worst-case scenario” is not that we become unemployed, but that we become indistinguishable. It’s the moment a writer pours their soul into a paragraph, agonising over its cadence and clarity, only for a reader to shrug and mutter, “ChatGPT could’ve written that.”
This is the true hollowing out. It’s the devaluation of effort, the erosion of the tacit agreement between writer and reader that these words are a missive from one human consciousness to another. A lifetime of reading, of accumulating experience, of feeling joy and grief and boredom, all distilled into a unique voice, is now competing with a machine that has “read” the entire internet and can mimic any voice it’s shown. The genuine article now has to worry about being perceived as the phony fake.
And for what? For what did one endure a laborious English medium schooling, build an industrious career over fifteen years in the US, only to return and be reduced to something not only no longer required but a character whose very articulation is suspect? Pile on top of that the crushing weight of bipolar depression and the skinless, kinless, inhospitable calculus of an NRI’s return, and you have a pain so specific, so textured, that no algorithm could ever synthesise its depths. It is a pain only I can viscerally feel, a unique confluence of circumstances that becomes the last bastion of my authenticity.
So, wherefore do we go from here? Hence, what is the writer to do? Perhaps the path forward is not to try and write “better” than the machine in the technical sense—more grammatically perfect, more syntactically complex. The machine will likely always win that game.
Perhaps, instead, the future lies in leaning into the very things the machine cannot replicate. It lies in the messy, the idiosyncratic, the vulnerable. It lies in the specific, ungeneralisable truth of lived experience. It lies in the deliberate imperfection, the slightly-off phrasing that betrays a human hand, the logical leap that isn’t statistically probable but feels emotionally true. Our grammatical “glitches,” our stylistic quirks, our raw and unpolished honesty—these may become our most reliable watermarks of humanity. We must write not like perfect machines, but like flawed, feeling, and gloriously fallible people. Because we are not just stringing probable words together; we are trying to tell someone what it is like to be alive. And that, for now, is a story only we can tell.
And people often take my ability to delineate the raw reality I see around me in impeccable English as their right for labeling me as a person who seems to see the world as a glass half full, which is a curious inheritance for an expression because the glass itself is such a middle-class American object of contemplation, not a Bengali one. The phrase assumes you are seated at a safe table, that your drink is potable, and that you are neither stuffed and satiated like an opulent mercenary nor ravenous and parched like a mangy Bengali mendicant, but rather in that middling zone of appetite where “half” feels like a philosophical choice rather than a bodily condition. From a Bengali perspective the whole thing is comic. In a land where a famine has been within living memory, the idea that someone has the leisure to argue whether the vessel is half full or half empty is almost absurd. For the man in the desert, half a glass is salvation. For the man who has already overeaten, half a glass is a burden. Only for the comfortable is it a diagnostic of optimism or pessimism.
That is why the metaphor feels suspiciously bloodless, a bit like weak tea that has been brewed too many times. It has no edge, no paradox, no dark humor. Zen koans at least rattle your brain by asking the impossible. Bengali sayings will sting you with irony: if you try to balance yourself in two boats, you will fall into the river. But “glass half full” is an instrument of polite psychology, invented somewhere between Dale Carnegie and post-war American motivational seminars, where people wanted to be told they were optimists rather than cynics. It works as cultural shorthand, but it isn’t ancient, it isn’t gnomic, and it isn’t profound. To those who have thirsted, it feels dishonest; to those who have eaten too much, it feels trivial.
Most people do this misappropriation casually, almost cheerfully, as if phrases and sayings are talismans whose mere repetition confers wisdom, when in truth the bulk of them are fragile glass baubles that only sparkle if held in the right light. Out of context they become nonsense, yet we keep deploying them because they sound like inherited truths. “Glass half full” works in a suburban kitchen but not in a famine field; “time is money” rings in a stock exchange but falls flat in a village where time is an unbearable weight; “survival of the fittest” was never meant as a guide for CEOs but is flung around boardrooms like Darwin was writing business manuals. The misapplication is the trick, the sleight-of-hand—people smuggle phrases into situations where they do not belong, and the surprise fit creates the illusion of wisdom.
Language is a con game at times, and the risk is that if we are not vigilant the very words we use to steady ourselves end up turning on us. That’s when we get the humiliating sense of having our own feces rubbed on our faces, realizing we’ve parroted something that disintegrates under scrutiny. The old trick was done by clever rhetoricians, priests, politicians, all with a practiced hand. The new trick will be automated: robots will be able to misapply language more deftly, more routinely, than any human, endlessly remixing clichés until we drown in recycled phrases. We will get whole oceans of glass metaphors and Zen-lite koans churned out by machines, each one half-shiny, half-foolish.
And if we aren’t careful, if we stop interrogating the fit between phrase and moment, we risk being reduced to linguistic dogs licking nonexistent anuses, savoring the ghost-tricks of words that once carried meaning but are now only gestures, repetitions, echoes. It is not optimism or pessimism that will define us then, but our willingness to notice when a phrase no longer belongs.