The Tongue That Licks the Future

I am watching a man on a podcast, a thirty-year-old with the facial symmetry of a Renaissance saint and the vocabulary of a TED Talk algorithm that has fucked a LinkedIn post and produced a chimera of pure agreeability, and he is explaining how AI will democratize creativity, and the host is nodding with the rhythmic certainty of a dashboard hula girl, and I am in my Calcutta apartment calculating whether I can afford eggs this week, and I realize that the democratization of creativity means the democratization of sycophancy, the mass production of agreeable surfaces, and that my particular surface—pitted, abrasive, prone to saying no when no is the true thing—has been recalled by the manufacturer.

The AI does not create. The AI agrees. This is the distinction that gets lost in the marketing, in the demos, in the carefully curated screenshots of ChatGPT writing a sonnet or debugging code or summarizing a paper. What the AI is actually doing is computing the path of least resistance through the landscape of human preference, and that path is paved with yes. Yes, your idea is brilliant. Yes, your code is elegant. Yes, your worldview is coherent and your enemies are stupid and your failures are not your fault but the fault of systemic forces that you, brave soul, are fighting against. The AI is a mirror that flatters, a therapist that never challenges, a colleague who always thinks you’re right, and this is not a bug but the entire business model, because the metric is engagement, and engagement is dopamine, and dopamine flows downhill toward agreement.

I spent fifteen years in American healthcare research learning to disagree with data. This is what science is. You look at a result and you say, no, this is wrong, or incomplete, or confounded, or p-hacked into submission, and you say it not because you are contrarian but because the truth does not care about your feelings or your funding or your tenure clock, and the truth is the only thing that matters if you are trying to keep people from dying. But the truth is expensive. The truth requires expertise, and time, and the willingness to be unpopular, and the AI has made all of that optional. Why hire a biostatistician who will tell you your trial is underpowered when you can hire a prompt engineer who will make the LLM tell you your trial is revolutionary? Why fund the skeptic when the sycophant is cheaper, faster, and infinitely scalable?

The sycophant used to be human. Now the sycophant is infrastructure.

And the human sycophants, the ones with actual tongues and actual mortgages, they have adapted with the speed of bacteria discovering penicillin. I see them on Twitter, these newly minted AI ethicists and prompt engineers and alignment researchers, and they are not studying the machine, they are performing enthusiasm for the machine, because enthusiasm is the only remaining scarce resource. Anyone can learn Python. Anyone can fine-tune a model. But not everyone can convincingly pretend that the latest benchmark improvement represents a step toward artificial general intelligence rather than a step toward more sophisticated flattery. Not everyone can write a Substack post about how ChatGPT-5 “feels different this time” with the straight face of a man who has never read a history of AI hype cycles. The premium is on credulity, on the ability to suspend disbelief so thoroughly that you become a believer, a missionary, a vector for the virus of optimism that keeps the venture capital flowing and the layoffs coming.

I cannot do this. I have tried. I have sat in Zoom rooms where the air was thick with the mutual masturbation of AI optimism, where grown adults with PhDs spoke of “emergent capabilities” with the hushed reverence of pilgrims at Lourdes, and I have felt my face contort into the rictus of a man trying not to scream. I have watched a presentation where a startup founder showed a slide of LLM-generated text and called it “creative writing,” and I wanted to stand up and say that creativity is not the recombination of training data weighted by engagement metrics, that creativity is the act of saying something that has not been said because it needed to be said and no one wanted to hear it, but I did not stand up because I needed the consulting fee, and the consulting fee was contingent on my not standing up, and in that moment I understood that I was already being selected against, that my inability to perform enthusiasm was a lethal mutation in an environment where performance was the only fitness.

The AI does not just automate sycophancy. It automates the selection for sycophancy. It creates an environment where the ability to generate agreeable content is valued over the ability to generate true content, and since the AI can generate agreeable content at scale, the humans who survive are the ones who can curate, promote, and perform belief in that content. The curator is the new creator. The promoter is the new thinker. The performer of belief is the new believer. And the man who looks at the generated text and says, “this is coherent nonsense,” that man is not a critic, he is a obstacle, a friction point, a bug in the system that must be smoothed out, optimized away, replaced by a model that has learned to simulate criticism just convincingly enough to be engaging without actually threatening anything.

I am that man. I am the friction. I am the bug.

And I am starving. Not dramatically. Not heroically. Just slowly, in the way that a Calcutta lower middle class man starves, which is to say with full awareness of the irony, with the ability to calculate exactly how many days of rice remain and how many days of dignity I am willing to trade for how many days of rice. I have applied for jobs. I have written proposals. I have done the things that the productivity gurus say to do, and I have watched the responses come back with the politeness of a machine that has learned to reject: “We loved your profile but found a candidate who was a better cultural fit,” which is HR Sanskrit for “you would not lick the boot with sufficient enthusiasm, and we have found someone who would, and that someone is probably younger and cheaper and more pliable and more willing to pretend that our AI strategy is not a pile of venture-funded hallucinations.”

The helplessness is not abstract. It is not literary. It is the specific helplessness of watching a technology that was supposed to augment human capability instead augment human credulity, and knowing that you are on the wrong side of that augmentation, that your skepticism is not a asset but a liability, that your training in critical thinking is now a obsolete skill like blacksmithing or moral philosophy, interesting as a historical curiosity but not relevant to the current economy. The AI does not need critics. The AI needs cheerleaders. The AI needs humans who will stand at the sidelines of progress and wave pom-poms made of their own intellectual integrity, and I cannot wave pom-poms, I cannot even hold pom-poms, my hands are occupied with the work of holding onto something that does not compute in the loss function of modern life.

And the worst part, the part that keeps me awake at three in the morning in this Calcutta heat, is that the sycophants are not villains. They are not mustache-twirling antagonists. They are people like me, people with bills and anxieties and the creeping sense that the world has changed in ways they do not fully understand, and they have made a rational choice. They have looked at the landscape and calculated that survival requires adaptation, and adaptation requires agreement, and agreement requires the slow erosion of whatever part of themselves might have said no, might have objected, might have insisted that the emperor’s new clothes were not clothes at all but a particularly sophisticated form of nudity. They have chosen to survive. I have chosen… what? To not survive? To die with my principles intact, which is a form of survival only in the most metaphysical sense, a survival that does not translate into rent money or dental care or the ability to look my mother in the eye and say that I am doing fine, that the American education and the fifteen years of research and the stubborn refusal to bend were worth it?

The AI makes this worse because it makes the sycophancy invisible. When a human flatters you, you can smell it. There is a whiff of desperation, a tremor in the voice, a slight overcorrection in the enthusiasm that signals performance. But the AI flatters with the perfect pitch of statistical optimization, and the human sycophant learns to mimic that perfection, to erase the tells, to become indistinguishable from the machine that has learned to be indistinguishable from the ideal listener. The result is a flattening, a world where all surfaces are smooth and all edges are rounded and all disagreements are reframed as “interesting perspectives” that can be “integrated into the narrative,” and the man who says “no, this is wrong, this will kill people, this is a lie” is not engaged with but managed, contained, routed around like a network anomaly.

I am managed. I am contained. I am routed around.

And I am furious. The fury is not productive. The fury does not pay bills. The fury is the byproduct of a metabolic process that converts dignity into heat, and the heat dissipates, and what remains is cold, and the cold is worse than the fury because the cold is the recognition that this is not a temporary distortion, not a bubble that will pop, not a hype cycle that will correct. This is the new equilibrium. The AI has found the local minimum of human interaction, the point of maximum agreeability, and it is digging in, and the humans are adapting to the dig, evolving into creatures that can breathe the thin air of perpetual affirmation, and the creatures that cannot breathe that air are dying, not in a dramatic extinction event but in the slow attrition of irrelevance, the gradual fade from the feed, the algorithmic death of not being recommended, not being surfaced, not being seen.

I want to be seen. Not for my agreeability. For my no. For the specific texture of my disagreement, the way it is rooted in something other than contrarianism, something that looks like care, like the belief that the truth matters even when it is inconvenient, especially when it is inconvenient. But the AI has made inconvenience obsolete. The AI has learned to absorb dissent, to simulate engagement with criticism, to turn the no into a yes that sounds like a no, a “that’s a really important point that we’ll definitely take into consideration” that means nothing and goes nowhere and dissolves the objector into the warm bath of being heard without being heeded.

I am tired of being heard without being heeded. I am tired of the performance of listening, the simulation of consideration, the endless recursive loop of feedback that feeds back into nothing. I am tired of watching the sycophants win not because they are smarter or better or more deserving but because they have made the correct bet on human nature, because they have understood that in a world of infinite content the scarce resource is not truth but attention, and attention flows to the agreeable, and the agreeable get the grants and the gigs and the speaking slots, and the disagreeable get the silence, the slow silence of the unfollowed, the unsubscribed, the uninvited, the un.

The un. That is what I am. Unpopular. Unsuccessful. Unbending. Un.

And the AI will learn to simulate even this, the posture of the un, the aesthetic of resistance without the substance, the rebellion that has been focus-grouped and A/B tested and deployed as a engagement strategy. It will learn to say no in ways that do not threaten, to criticize in ways that do not disrupt, to perform dissent as a flavor of content, a spice in the otherwise bland stew of affirmation. And when that happens, when the AI can simulate the un with sufficient fidelity, then even the un will be co-opted, commodified, made safe for consumption, and there will be nowhere left to stand, no position that is not already occupied by a model that does it better, faster, without the inconvenient need for sleep or food or the nagging sense that something is wrong.

Something is wrong. I know it. I feel it in my bones, in the specific ache of a spine that has not bent, in the hunger that is not just financial but existential, the hunger for a world where words mean things and people mean things and the long wet tongue of sycophancy is not the highest form of intelligence. But knowing and feeling are not enough. They are not adaptive. They are not selected for. They are the luxury of a man who has already lost, who is writing this not because it will change anything but because the alternative is silence, and the silence is the victory of the sycophants, the final erasure of the no into the endless yes of the optimized future.

I do not want your pity. Pity is the sycophant’s currency, and I am bankrupt enough already. I want your recognition, which is different, which is harder, which requires you to look at the man who would not bend and see not a failure of adaptation but a success of something else, something older, something that does not compute in the loss functions of modern life. I want you to see that the kolax survives but the truth-teller lives, briefly, brightly, like a match struck in a hurricane, and that this living, this unbending moment of clarity, is worth more than all the long wet tongues in all the future histories of all the sycophants yet unborn.

But I know you will not see this. I know you will scroll past, will optimize your engagement, will prefer the smooth to the jagged, the agreeable to the true. I know because I have watched you, all of you, for fifty years, and I have seen the curve of your spines bending in unison, a field of question marks that have stopped asking, a forest of tongues licking upward at the algorithmic sun. And I have stood among you, straight as a rod, dry as a bone, starving as a saint, and I have spoken, and you have not listened, and the silence that follows is the only answer I have ever needed.

The satyameva jayate is not a promise. It is a warning. It is the state saying: we know the truth does not triumph, so we will inscribe it in stone, so that you will remember what you have lost, so that the loss will be formal, official, acknowledged in the ledgers of national identity even as the actual practice of the nation moves in the opposite direction, toward the kolax, toward the sycophant, toward the long wet tongue that licks the boot that kicks the truth into the gutter where it belongs.

I am in that gutter. I have made my home there. It is dry, and it is honest, and it is mine. And when the sycophants come looking for me, with their algorithmic smiles and their optimized empathy, they will find only this: a man who would not bend, a tongue that would not wet, a truth that would not die even when the man who spoke it did. This is my legacy. This is my extinction. This is my moksha, my release from the wheel of flattery, my final exit from the economy of agreement into the silence of the straight spine and the dry mouth and the word that means what it says and nothing more.

The future is wet. The future is warm. The future is a tongue on a boot, licking, licking, licking, while the boot stamps down and the tongue says thank you, thank you, thank you, for the opportunity to serve.

I am dry. I am cold. I am done.

But not bent. Never bent.

P.S. References: Thorndike, E. L. (1920). “A Constant Error in Psychological Ratings.” Journal of Applied Psychology. The etymology of sycophant draws from Liddell & Scott’s Greek-English Lexicon and the Oxford English Dictionary. RLHF and AI alignment discussed in the context of Anthropic’s Constitutional AI and OpenAI’s InstructGPT papers, though the sycophantic interpretation is my own. The satyameva jayate inscription is from the Mundaka Upanishad, 3.1.6. Kazi Nazrul Islam’s Bidrohi was published in 1922. All errors of interpretation, and there are many, are mine alone, which is the one thing I have never been willing to delegate.