New toy, old system

Trivial as these scenes seem, they point to a larger question of regulation once someone moves past harmless absurdity into using AI in a more sinister manner: unfortunately, no longer hypothetical.

It doesn't take a magnifying glass to see how AI has become woven into everyday work life, private or public: reports, presentations and visuals are now routinely produced through a simple prompt.

Supervisors know the tells by now: the same over-polished phrasing and suspiciously immaculate structure, appearing again and again; graphs arrive with misspellings, labels float awkwardly over visuals, and AI-generated ads produce faces that resemble real people while still looking like glossy caricatures of them.

What makes this odd is that almost everyone seems to know what is happening, yet very few people seem inclined to say much about it. The arrangement is convenient enough to keep everyone complicit.

Yet convenience disguises erosion: people gradually stop practising the skills they were hired for, because a prompt can reduce the work itself to something trifling: another passable document produced merely to survive the next soul-devouring meeting. That is where Pakistan's enthusiasm for AI begins to look premature.

The National AI Policy 2025 promises to train one million people and produce 10,000 trainers, while the Islamabad AI Declaration, adopted in February 2026, calls for explainable systems, human accountability and institutional oversight. These are not meaningless commitments, but the machinery required to enforce them is still being assembled.

By August, the Pakistan Digital Authority was advertising for an adviser to help establish the frameworks, standards and oversight mechanisms on which responsible AI use is meant to depend. Because let's face it, no one wants to be the kid outside the window while everyone else plays with the newest toy.

But there is something almost comical about sprinting towards advanced technology while the machinery beneath everyday public life remains painfully dated. Anyone who has stood in a long queue at Lahore's Allama Iqbal International Airport while an official wrestles with rotting hardware and outdated software will recognise the contradiction immediately.

A simple question about an e-visa can become a labyrinth of outdated sources, where even the latest visa details may not be readily available to travelers.

The same contradiction is visible across the public sector: notebooks, paper files and handwritten registers sit comfortably beside speeches about digital transformation, even as the Digital Nation Pakistan Act 2025 promises systemic public sector modernisation. And yet AI is already being treated like the next great leap forward.

No one is suggesting Pakistan should shut the laptop and wait for the rest of the world to move on without us; but it might help to finish setting up the desk before ordering the newest computer.

AI isn't a harmless software update: it can copy a voice, fabricate an image, produce convincing written work and create material that looks real enough to cause damage before anyone has time to question it.

Pakistan also cannot simply lift an AI framework from somewhere else and assume the job is done, because regulation has to make sense within the society where AI will coexist with the people most exposed to its consequences.

In Pakistan, where reputation, gender and social standing can dramatically shape how severely someone is punished by public opinion and institutions, AI misuse will inevitably land harder on some people than others. Otherwise, we risk doing what we so often do with technology: refashioning the skeleton of an archaic system and expecting it to run.