Saud Bin Nauman
A friend in Lahore — senior partner at a law firm of good standing — told me recently, with some satisfaction, that his chambers have stopped hiring trainee associates. “The machine drafts better than a five-year lawyer,” he said, “and it does not ask for leave.”
He calls this efficiency. He has not yet asked himself the question that should be keeping him awake: when his generation retires, where will the next senior partners come from, if nobody is ever allowed to be junior again?
Hold that question. Because the future of every profession depends on how it answers it. It is the real AI story of 2026, and it is not the one you have been reading.
You have seen the other column many times by now — the one that tells the freelancer to upskill and the graduate to learn prompting. This is not that column because this year’s layoffs are not chiefly about the freelancer. Oracle cut 21,000 people; Amazon cut 16,000. When a global law firm shed a tenth of its people, the partners stayed — the support staff went.
The pattern is uniform: not the senior lawyer, his assistants; not the chief engineer, the fresh graduate. The machine is not yet eating professions. It is eating the apprenticeship — the junior years through which every profession has always reproduced itself.
Geoffrey Hinton, the Nobel-winning godfather of AI, compressed the economics into one sentence: “It will make a few people much richer and most people poorer. That’s not AI’s fault, that is the capitalist system.” The machine’s rewards flow first to those who own it. Its costs arrive first for those selling routine labour. The first thing it consumes is the only commodity a junior has ever had to sell: hours of routine work, done cheaply, while learning.
Now, say aloud what Pakistan actually exports. Not software, in the main — junior hours. Pakistan’s IT exports crossed four billion dollars this fiscal year. In the global division of digital labour, Pakistan is the world’s apprentice. My friend in Lahore is only one example of a much wider shift. There is no shame in it — every industrial power began as somebody’s workshop — but the position rests on one assumption: that the world will always need apprentices. This year, the world has begun cancelling the apprenticeship. Much of Pakistan’s digital export success still depends on junior labour, and that labour is being repriced towards zero.
So this is not the freelancer’s problem. It is yours. Your firm’s pyramid is built on billing junior time. Your daughter’s computer science degree assumed a first job that is quietly ceasing to exist. The staircase your own career climbed is being dismantled one step from the bottom.
What should a sensible person do? Not despair — because one fact separates this machine from every machine before it: the others belonged only to those who could afford to build them. This one rents for twenty dollars a month. Last month, the American government suspended Anthropic’s Claude Fable within days of release under national-security rules; it cleared to return only at the end of June. A tool a superpower treated as too dangerous to circulate rents tonight, for the price of two pizzas, to anyone — the partner in Lahore or the kid in a village.
Consider what is already possible, if we choose to act. In April, Telenor launched “Kissan Dost Bashir,” Pakistan’s first AI-powered agricultural voice bot. I called a farmer in Sahiwal who uses it. “I used to travel forty kilometres to check mandi rates,” he told me. “Now I just dial a number. It tells me the weather and the prices. My father never had this.” The technology is here. The question is whether we scale it.
Flood warnings offer another glimpse. The 2022 floods displaced 33 million people and caused catastrophic losses. Today, a China-Pakistan joint team has deployed an intelligent flash flood early warning system in Swat and Wazirabad, extending the warning window by five hours and incorporating glacial melt into forecasts These systems have already avoided casualties in high-risk valley areas. They are not science fiction. They are already saving lives, in some corners of this country. What is missing is not technology. What is missing is the will to switch it on across the board. I watch from the inside what taking this seriously looks like.
The German insurance group I work for switched on an internal AI assistant in 2023. Nobody has been replaced — the answer has been to retrain people onto the machine. What my company did, the country is doing. Deutsche Telekom and NVIDIA have switched on Europe’s largest Industrial AI Cloud in Munich. A third of German schools already use AI.
Cautious, slow-moving Germany decided it could not afford to wait. Where governments act, AI becomes infrastructure, not privilege. And Islamabad? So far, ceremonies. The National AI Policy was approved in July 2025. A Rs283 billion programme has been planned. “AI Seekho 2026” aims to train one million young people.
A fair question: how many of the million are trained, and where do they work? The AI Council — the apex body meant to guide the policy — has not yet been formed. The Personal Data Protection Bill remains pending. Pakistan faces an acute shortage of high-performance computing data centres.
The policy is ambitious but lacks legal teeth. The difference between Islamabad and Munich this year is the difference between a promise and a power switch. My friend in Lahore cannot single-handedly fix the AI Council. But he can change what he hires juniors for. And he can ask himself whether his firm’s obituary is being written in the very efficiency he celebrates. So, two instructions, by seniority. To the young: learn the machine now, while it is cheap and your seniors are still laughing at it — in every technological upheaval, the ones who reached the machine first were the ones it did not remove. To the senior — to my friend in Lahore: keep hiring juniors, and change what you hire them for.
A junior with the machine is your succession plan; the machine without a junior is your firm’s obituary. And a third instruction, for those who govern. Stop treating AI as a ceremony. Form the AI Council. Pass the Personal Data Protection Bill. Build the high-performance computing centres. And make AI available to every Pakistani — not as a promise for 2030, but as a tool for today. The farmer in Sahiwal already has his voice bot.
The people of Swat already have their flood warnings. They did not wait for a policy. They simply needed someone to switch it on. Pakistan has been the world’s apprentice long enough to know the apprentice’s two fates: the one who learns only the master’s routine is dismissed the day the routine is automated; the one who learns how the whole workshop runs ends up owning one. The workshop is being rebuilt. The rules of who owns it are being written now, in other people’s capitals. For once in our history, let us arrive before the ink dries.
—The writer is an accountant and a passionate journalist based in Munich, Germany. He has a keen interest in socio-economic and political issues. His writings frequently explore the intersections of policy, finance and cultural exchange, with a particular focus on strengthening ties and fostering dialogue between Pakistan and Germany. He tweets at @SaudBinNauman1.
