The AI Sermon and the Wet Knot
AI: Artificial Intelligence, meaning software systems that can generate, classify, summarize, translate, recommend, code, search, and automate by learning patterns from data.
GPU: Graphics Processing Unit, a specialized computer chip originally used for graphics and now widely used to train and run AI systems.
CSV: Comma-Separated Values, a simple table-like file format where data sits in rows and columns.
TED: Technology, Entertainment, Design, a conference format famous for polished talks about big ideas, sometimes useful, sometimes smelling faintly of expensive optimism.
My laptop is blowing hot air onto my thigh like a pigeon recovering from bronchitis, and somewhere on the internet a man with excellent lighting is telling the unemployed to use AI.
That is the whole comedy now.
Use AI.
Build software.
Launch a business.
Make agents.
Automate your life before your life notices it has already been repossessed.
This advice arrives wearing a fresh shirt. It has white teeth, a ring light, a newsletter, and that peculiar startup confidence which makes even a sentence about disaster sound like a gym membership offer. It tells the clerk, the coder, the painter, the copywriter, the engineering graduate waiting at home, the girl with a portfolio and no payments, the middle-aged fellow with acidity and a cracked plastic chair, that the future is available if only he or she learns to prompt correctly.
A very neat word, “use.”
As if use contains rent.
As if use contains clients.
As if use contains a quiet family, paid bills, a working fan, and a father-in-law who has finally stopped making that small, throat-clearing noise of judgment.
The lie is not that AI is useless. That would be too easy, and also foolish. AI is useful. A knife is useful. A spreadsheet is useful. A pressure cooker is useful. But if you treat a pressure cooker like a temple idol, it will eventually redecorate your ceiling with dal.
The lie is that tools erase the world around them.
A tool can help you make something faster. It cannot guarantee that anyone wants it, trusts you, finds you, pays you, remembers you, or does not already have fourteen similar things open in fourteen browser tabs while also worrying about gas cylinder prices and cholesterol.
This is the part the cheerful people skip.
They show the making. They do not show the selling.
They show the app. They do not show the customer support.
They show the logo. They do not show the market.
They show the AI-written children’s book. They do not show the ten lakh other hopeful people uploading the same glossy little animal story into the same flooded marketplace, all of them told by the same guru that passive income is waiting politely near the checkout counter.
The old advice was: get a job.
The new advice is: become a company.
This is a breathtaking escalation. Yesterday you were asked to send a résumé. Today you are expected to become founder, marketer, designer, coder, accountant, performer, distribution expert, customer service desk, brand strategist, and emotional rhinoceros.
And if you cannot do it, you are “not adapting.”
That phrase has become a small stick. It is tapped on the knuckles of anyone frightened by a machine that can do in thirty seconds what they were once paid to do badly, slowly, humanly, with tea breaks and eye strain. Not adapting. Not upskilling. Not leaning in. Not building in public. Not finding a niche.
Niche is a word used by people who have never stood in a ration shop queue behind a man arguing over two rupees.
There is a difference between encouraging a person and pushing him toward a cliff while shouting motivational slogans. Entrepreneurial advice is especially cruel when given to people who cannot afford repeated failure. Failure is charming only when there is a cushion under it. In Silicon Valley stories, failure is a badge, a TED-ready seasoning, an avocado smear on toast. In ordinary life, failure first eats your savings, then your sleep, then your confidence, then your phone contacts. After a while people do not say you failed. They say you are “going through a phase,” in that careful voice used for fever, drinking, and furniture damaged by rain.
Thick skin is not a downloadable plugin.
Some people have family money. Some have social networks. Some have fluent English that opens doors before they knock. Some have savings. Some have visas, mentors, cushions, cousins, uncles, school friends in useful places, and the casual bravery of those who have never had to count loose change before ordering eggs.
The rest are told to be bold.
Boldness is cheap advice when someone else is paying for the bruises.
AI makes this worse because it gives the advice a technical shine. Earlier the motivational man merely told you to believe in yourself. Now he tells you to build an agent. The sermon has upgraded its vocabulary. The old snake oil came in bottles. The new one comes with a dashboard.
And yes, the dashboard may work.
That is the trick.
The tool may be genuinely impressive. It may write decent code, draft competent emails, summarize documents, generate images, make outlines, suggest business names, turn a messy note into polite English, and help a tired person produce a better first draft than his own exhausted brain could manage at midnight.
But usefulness is not livelihood.
Capability is not income.
Output is not demand.
This is the missing screw in the whole machine. AI shifts the problem from “Can I make the thing?” to “Can I sell the thing when everyone else can also make the thing?” That second problem is colder. It has no friendly demo video. It involves attention, trust, timing, luck, money, patience, reputation, stamina, and the ability to survive long enough for randomness to mistake you for talent.
That is not prompt engineering.
That is life with a leaking roof.
The modern internet hates this because it prefers clean diagrams. Life, unfortunately, is not a clean diagram. It is not a CSV file with neat rows and polite columns. Poverty is not a column. Loneliness is not a column. A sick mother is not a column. Rent, debt, tooth pain, heat, bad sleep, class shame, bad schools, family pressure, medical bills, unpaid invoices, cheap plastic chairs, corrupt offices, vanished entry-level jobs, and the smell of a small room at 3 p.m. when the fan is moving hot air from one wall to another like a bored clerk moving files — these things do not line up neatly for an algorithm.
They form a knot.
Not a decorative sailor’s knot.
A drain knot.
Wet hair, soap scum, old water, and regret.
AI can nibble at pieces of the knot. It can help a student understand a difficult topic. It can help a doctor draft a note. It can help a small shopkeeper write clearer product descriptions. It can translate. It can search. It can summarize. It can reduce some drudgery. It can make certain people much faster.
But it cannot repair the social contract by autocomplete.
It cannot invent fair wages. It cannot make clients honorable. It cannot make schools sane. It cannot make healthcare affordable. It cannot make public institutions less decayed. It cannot make a market less crowded. It cannot give beginners the apprenticeship ladder after the first rung has been greased and polished with automation.
That first rung matters.
People talk about job loss as if it happens in one dramatic cinematic sweep. Robots march in. Humans march out. End credits. But real job loss is sneakier. The ladder remains visible. The first step becomes slippery. Entry-level workers are told that AI can do their beginner tasks. Mid-level workers are told that productivity has improved, which means they now do the work of three people with the cheerful face of a hostage in a birthday photograph. Senior people become managers of ghosts: tools, dashboards, contractors, agents, and one remaining junior named Rohan who has gastritis and a Canva subscription.
Everyone is told not to worry.
Whenever a powerful system tells you not to worry, worry has already been transferred to someone weaker.
This is why the consumer AI gospel feels less like progress and more like a civic fraud in cheerful fonts. People are being hassled into hope. They are being harried into optimism. They are being told that if the economy has become cruel, the solution is to become a better little machine inside it.
Wake earlier.
Learn faster.
Post daily.
Build publicly.
Ship constantly.
Fail faster.
Even despair now has to wear sneakers and attend a webinar.
The old factory had whistles. The new factory has notifications.
Once, Frederick Winslow Taylor stood over workers with a stopwatch and sliced human motion into profitable pieces. Now the stopwatch has moved inside the skull. Every quiet moment becomes waste. Every hobby becomes content. Every skill becomes a product. Every person becomes a small business waiting to disappoint himself.
This is not freedom. This is a job description written by people who confuse society with a pitch deck.
And somewhere behind all this sits the giant glowing furnace: GPUs, data centers, venture capital, cooling systems, electricity, water, cables, chips, benchmark boasting, and executives speaking of the future with the moist confidence of men who have never had to bargain with a plumber. Some parts of the AI business will make money. Of course they will. The people selling shovels usually do better than the people digging with blisters.
But the fantasy being sold to ordinary people is different. It says that millions of individuals can pay monthly for tools that help them produce millions of similar products in already flooded markets, and somehow everybody will earn.
This is not economics.
This is a pyramid built out of panic and subscription fatigue.
The tragedy is that a more serious conversation is possible. AI as infrastructure. AI as assistant. AI in medicine, logistics, accessibility, translation, research, documentation, education, fraud detection, software maintenance, and boring office plumbing. Useful, limited, expensive, dangerous, impressive, and in need of adult supervision.
But adult conversation does not go viral. Adult conversation has back pain. It asks dull but necessary questions. Who pays? Who loses? Who owns the data? Who carries the risk? What happens to beginners? What happens when production becomes cheap but trust stays expensive? What happens when everyone can generate everything and nobody knows what is worth looking at?
These questions do not give the listener a dopamine firecracker. They do not say, “You are early.” They say, “You may already be late, the bus is smoking, and the conductor is selling a masterclass.”
That is bad for engagement.
So the evangelists keep smiling. They say democratization. They say unlock your potential. They say personal brand. They say one-person unicorn, which is a phrase so ridiculous it should be taken to a veterinary clinic and observed overnight.
A one-person unicorn.
Why stop there?
A one-person municipality.
A one-person sewage board.
A one-person Bengal monsoon with localized flooding near the underpants.
The cheery AI prophets should not be beaten or censored. That would only make them heroic, and there is nothing a stage-friendly optimist enjoys more than persecution with good lighting. They should simply be seated under a flickering tube light with a steel glass of water and a few inconvenient people.
An accountant.
A schoolteacher.
A nurse.
An unemployed graduate.
A small business owner.
A mother pricing medicines.
A poet whose work has been scraped into digital soup.
A man with a rice cooker and enough unpaid experience to smell nonsense before it enters the room.
Then let the prophets explain the future.
No wireless microphone.
No stage lights.
No “community.”
No “moat.”
No “creator economy.”
Just the old plastic chair, the sweating upper lip, and one question: if everybody can generate everything, who is left to pay anybody for anything?
The end will not look like chrome skeletons marching through fire. That is too organized for us. Our version will be sillier and more bureaucratic. A helpdesk ticket marked resolved while the building burns. Dashboards showing green. Workers making AI-generated posters about resilience while their professions are quietly fed into a machine that sends back invoices. A smiling man saying “This is an opportunity” while the floor gives way.
AI will matter. It already matters.
That is why we should stop talking about it like drunk futurists at a sponsored buffet. Use it where it helps. Distrust it where it flatters. Price it honestly. Govern it seriously. Do not sell it to frightened people as a private lifeboat in a public flood.
And if someone tells you that all you need is AI, grit, and a personal brand, ask him for his savings, his family cushion, his client list, his distribution channel, his rent, his medical bills, his third failure, and the names of the people who still answer his calls.
Then ask whether his thick skin came bundled with the software.