AI arriving in Pakistan as mirror

Pakistan’s relationship with artificial intelligence (AI) is usually framed as a question of adoption: how fast can the country acquire the tools, train people to use them, and claim a place in the AI economy? That framing is too small.

The harder question is whether Pakistan can use AI with judgment - in governance, in learning, in research, in business, and in public policy.

Even more fundamentally, we need to ask whether the structure of education itself, built around fixed-duration degrees and one-time credentials issued by institutions designed for a pre-AI world, still makes sense.

The answer should make us uncomfortable. AI is not arriving in Pakistan as a magic escalator. It is arriving as a mirror. It will show us where our education system is strong, but more likely it will expose where it is weak: rote learning, poor literacy, weak numeracy, thin research habits, narrow degrees, bureaucratic universities, and a public conversation that too often rewards status over thought.

The most important reform Pakistan can undertake is therefore the complete revamping of its education system. Tax reform matters. Energy reform matters. Export policy matters. Investment promotion matters. But each of these eventually runs into the same wall: a population not equipped to read deeply, reason independently, use technology intelligently, solve problems, or adapt to a changing world. Education is not one reform among many. It is the reform that determines whether the others can work.

Daron Acemoglu and Pascual Restrepo provide a useful framework. Their task-based model of automation distinguishes between technologies that simply replace human labour in existing tasks and technologies that create new tasks requiring judgment, oversight, creativity and complementary skill. Which effect dominates is not decided by the machine alone. It depends on the institutions around it - schools, universities, firms, regulators and the state - and on whether people are prepared to use the technology productively.

This is Pakistan’s danger. If a memorization-based system hands AI tools to students, teachers and officials without changing how people learn, think and work, the result will not be transformation. It will be a so-so outcome: faster reproduction of mediocre work. Faster mediocrity is still mediocrity; now it just arrives with better formatting.

Acemoglu and Simon Johnson have warned about so-so technologies - innovations that automate tasks without raising productivity or creating higher-value work because the workforce and institutions around them are never reorganized to capture the gains. That is precisely the risk here. AI can help a serious student explore ideas, test assumptions, draft arguments and improve learning. But in a weak system it can also help a lazy student produce an assignment that looks polished while containing little thought. It can assist a good teacher, but it can also camouflage a bad classroom. It can strengthen governance, but it can also generate smarter-looking files for the same old broken decisions.

This is why the current education crisis cannot be treated as a background issue. Pakistan is not starting from a position of strength. UNICEF estimates that around 25 million children aged 5 to 16 are out of school. The World Bank’s learning poverty data shows that a very large share of children at late-primary age are not proficient in reading, once out-of-school children are included. In plain language, millions are outside the classroom, and many inside the classroom are not learning enough. That is the dismal platform on which the AI age has arrived.

The fiscal signal is no better. The federal Budget in Brief 2026-27 places Education Affairs and Services current expenditure at roughly Rs117.7 billion. UNICEF has noted that Pakistan’s overall public education spending has recently been around 0.8 percent of GDP, far below the 4 to 6 percent of GDP benchmark often associated with serious education investment. This is not an education emergency response. It is fiscal tokenism dressed as policy. No country can enter the AI age on bargain-bin education and then wonder why the future chose another address.

Pakistan must first bring people into the learning fold, but not by dragging them through the same slow, book-heavy, examination-heavy pipeline that has already failed too many. The country needs a parallel national learning pathway that is flexible, practical and modern from the start: community learning hubs, evening and weekend classes, mobile classrooms, digital lessons, accelerated literacy and numeracy, AI-assisted tutoring in Urdu and local languages, and certification based on competence rather than seat time.

This is where Pakistan can leapfrog. Children and adolescents who have never had proper schooling should not be asked to spend years catching up to a broken model. They need strong foundations in reading, numbers, language, reasoning, communication and digital confidence, while being introduced early to tools that help them learn independently. For older learners, education should connect directly to livelihoods: agriculture, repair work, logistics, healthcare support, design, freelancing, data work, small enterprise, and, where relevant, coding. The objective should not be to manufacture degree-holders. It should be to build capable people.

AI also changes the importance of experience. In a world where information can be generated instantly, lived judgment becomes more valuable, not less. A person who has worked across sectors, dealt with people, understood failure, negotiated constraints and seen policies collide with reality can ask better questions of AI than someone who only knows how to prompt it neatly. AI rewards wide exposure. It helps those who can connect agriculture with water, energy with exports, education with employment, and technology with society. The narrow specialist is not finished, but the specialist who cannot see beyond the silo is in trouble.

This also makes the old credentialing model increasingly inadequate. The traditional path - years of schooling culminating in a fixed degree, treated as a lifetime credential - was designed for an economy in which skills depreciated slowly. That world is fading. Generative AI is shortening the half-life of many specific skills. A certificate showing what a person knew at twenty-two says less and less about what that person can do at thirty-five.

Pakistan does not need the same certifications stretched longer or shortened at the edges. It needs a different relationship between education and certification. Foundational education must be protected and strengthened: literacy, numeracy, reasoning, language, ethics and civic understanding. Beyond that foundation, the system should move toward modular, stackable and renewable certification. People should be able to re-enter learning throughout life, acquire new competencies, prove them, and be re-certified as task content changes.

This matters because the labour market is no longer organized around stable job titles alone. Acemoglu and Autor’s work on labour-market polarization showed even before generative AI that routine cognitive tasks were under pressure. AI accelerates that pressure. The safer bet is not a longer race toward the same old degree. It is a system in which people keep building verified capability over time. A degree may still matter, but it should no longer be treated as a permanent passport stamped once at the gate of adulthood.

Erik Brynjolfsson’s warning about the ‘Turing Trap’ is relevant here. Economies that chase automation by making machines imitate humans may concentrate gains narrowly and miss the larger opportunity: augmenting human beings so they can do new and more valuable work. A one-time credentialing system reinforces the trap. It certifies people once and leaves the burden of adaptation entirely on the individual. A modern system would treat learning as continuous public infrastructure, not a private struggle after graduation.

Universities should be central to this transition. They should be the places where society learns how to use AI with judgment: how to frame better questions, test assumptions, combine technology with ethics, and connect science, economics, literature, history, business and public policy. Their role is not simply to award degrees. It is to produce citizens and professionals who can think across boundaries.

Pakistani universities are currently ill-equipped for this task. The problem is not merely underfunding, though underfunding is real. The deeper problem is structure. Too many universities operate like bureaucratic machines rather than intellectual communities. Governance is often dominated by appointees, administrative hierarchies and public-sector habits. Vice-chancellor selection is frequently routed through political and bureaucratic networks. Boards and syndicates too often reflect officialdom rather than scholarship. Promotion systems can reward seniority, loyalty and safe compliance more than original work, good teaching or serious research.

This is not incidental dysfunction; it is an architecture of control dressed in the language of quality assurance. Accreditation and regulation often focus on inputs and processes - buildings, forms, titles, committees, paperwork - rather than outcomes such as learning, research quality, employability, intellectual independence and contribution to society. Nadeem Ul Haque and colleagues have written extensively on the weakness of Pakistan’s university research culture and the need to build institutions that reward ideas rather than compliance. That argument becomes even more urgent in the age of AI.

The reason is simple. If AI can reproduce approved knowledge faster than any lecturer, then universities that merely reward the reproduction of approved knowledge lose their reason for being. The value of higher education will lie in cultivating judgment: the ability to know which question matters, which assumption is weak, which source is unreliable, which model is incomplete, and which answer may be technically neat but socially foolish. That capacity cannot be cultivated by institutions organized around status, fear and conformity.

The reform required is not cosmetic. Pakistan needs universities with genuine academic search processes for leadership; boards that include scholars, innovators, employers and independent thinkers; promotion and funding criteria tied to research quality, teaching quality, peer review, student learning and public contribution; and protection for academic freedom and debate. Universities must be treated as autonomous centers of inquiry, not as government departments that happen to award degrees.

There is also a cultural failure that needs to be named. Pakistani universities do not exchange enough with one another. Literary societies, policy forums, journal clubs, visiting lectures, inter-university debates, translation projects and serious public seminars are too limited, too episodic or too ceremonial. We have campuses, but not enough conversation between campuses. We have departments, but not enough dialogue between disciplines. A living intellectual culture requires movement: of students, teachers, books, ideas, criticism and disagreement.

The same is true of the broader intellectual class. Too much public debate is performative: panels, reputations, slogans and social media applause. Too little is built on sustained reading, honest disagreement, peer challenge and the slow work of developing ideas. The pseudo-intellectual thrives where institutions do not demand depth. AI will make this worse unless universities and serious thinkers raise the standard, because a shallow argument can now be dressed up in excellent grammar. The costume improves; the mind underneath may still be wearing slippers.

Pakistan’s reform agenda must therefore connect three things that are usually discussed separately: basic education, university governance and lifelong certification. Basic education must bring excluded citizens into learning and give them foundations. Universities must become merit-led institutions capable of independent thought. Certification must become modular and renewable so workers and professionals can adapt as tasks change. These are not separate boxes. They are one system.

The state has a role, but it should not be confused with control. Government should set national learning goals, finance access, protect standards, publish outcomes, support teacher training, fund research and create pathways for excluded learners. But it should not suffocate universities through bureaucratic appointment systems, compliance rituals and political interference. A society cannot command creativity through file work. It has to build conditions in which talent can breathe.

None of this argues against specialization. Pakistan will still need doctors, engineers, economists, teachers, lawyers, data scientists, technicians and skilled workers. The argument is against treating specialization or institutional status as something acquired once and never tested again. The country needs specialists who understand systems, policymakers who understand markets, teachers who reward curiosity over obedience, and business leaders who understand society.

AI will not automatically share its gains. Institutions decide whether new capability is broadly distributed or narrowly captured. In Pakistan’s case, the decisive institutions are not only ministries, regulators and markets. They are schools, colleges, universities and certification systems. If these remain rigid, underfunded, bureaucratic and status-driven, AI will not rescue Pakistan. It will simply expose the weakness faster.

The opportunity is still real. Pakistan can use AI to extend good teaching to children who have been left out, help teachers prepare better lessons, support students in local languages, build research networks across universities, modernize public policy analysis and create a lifelong learning architecture. But this will require deliberate reform, serious money, institutional autonomy and a national obsession with learning. There is no shortcut around that. AI may help Pakistan leapfrog stages of education, but it cannot leapfrog the need for seriousness.

The choice is therefore clear. Pakistan can use AI as a decorative tool on top of an unreformed system, producing better-looking mediocrity. Or, it can use AI as the trigger for the most important national reform of all: rebuilding education around curiosity, judgment, experience, autonomy and continuous capability. The first path is easy and useless. The second is difficult and necessary. The future will not wait for Pakistan to finish its paperwork.

Copyright Business Recorder, 2026