By the close of this decade, artificial intelligence will have quietly rewritten the rules of nearly every industry on earth. The numbers are staggering: McKinsey estimates AI could contribute up to $13 trillion to global economic output by 2030. The World Economic Forum projects that automation will displace 85 million jobs by 2025 while simultaneously creating 97 million new ones. The net arithmetic may look promising. But here is the uncomfortable truth that rarely makes headlines: those new jobs will overwhelmingly materialise in economies that have invested in human capital, digital infrastructure and ethical foresight. For countries that simply consume AI rather than shape it, the revolution could look far less like liberation and far more like dependency.
Pakistan must decide, urgently, which side of that divide it intends to occupy.
The global AI race is not just a technological contest — it is a civilisational test. For developing nations, the stakes could not be higher
The global AI landscape today is dominated by a narrow axis of power. The United States and China together account for nearly 70 percent of all AI research output and private investment. Europe is racing to regulate what it could not lead to. The rest of the world, home to the majority of humanity, is largely receiving technology designed elsewhere, trained on data that does not represent them and governed by values that were not theirs to choose. This is not merely a technical imbalance. It is a structural inequality with consequences as profound as those of colonialism once were.
For developing nations, the peril is specific. Automation threatens to eliminate the low-skill, labour-intensive manufacturing base that historically allowed countries to climb the development ladder. If that ladder is pulled up before Pakistan has climbed it, the results will be devastating for a country where 64 per cent of the population is under 30 and the formal economy already cannot absorb its youth. A generation arriving at the gates of the workforce may find those gates automated shut.
And yet, despair is not the only available response. The very scale of AI’s disruption creates openings for nations agile enough to seize them.
Pakistan’s path forward must be grounded in a clear-eyed distinction: the difference between using AI and understanding it. To use AI is to become a consumer in someone else’s value chain. To understand it – to develop scientific literacy, regulatory capacity, ethical vocabulary and eventually the indigenous capability to build and govern it – is to participate in the future on one’s own terms.
This begins, inescapably, with education. Pakistani universities must go far beyond introducing elective courses in data science. The deeper transformation required is pedagogical: a culture of inquiry, evidence and critical thinking that equips young people not merely to operate AI tools, but to interrogate them. Who trained this model? On whose data? Toward whose ends? These are not technical questions alone; they are civic ones. A student who cannot ask them is not prepared for the world being built around her.
The government, meanwhile, cannot afford to treat AI policy as a boutique concern of the technology ministry. Questions of algorithmic accountability, data sovereignty and automation’s labour-market effects belong in the mainstream of economic planning. Pakistan needs a national AI strategy – not a glossy document for international conferences, but a living framework with teeth: investment in public digital infrastructure, protections for workers in vulnerable sectors and meaningful participation in the global norm-setting bodies where the rules of AI are being written right now.
There is also a cultural argument to be made. Pakistan possesses something that no algorithm has yet replicated: the depth of its civilisational heritage, the richness of its linguistic diversity and the creativity of its people. These are not soft consolations. They are, in an age of homogenised machine intelligence, genuine competitive assets. AI systems trained overwhelmingly on English-language, Western-produced data are poorly equipped to serve the needs of Urdu speakers, of rural communities, of contexts shaped by traditions that Silicon Valley has never encountered. The opportunity to build AI that actually works for Pakistan and, by extension, for the broader Muslim world and the Global South is real. But it requires Pakistani talent choosing to pursue it.
The future of AI will not be determined by machines. It will be determined by the choices societies make about what they value, what they protect and what kind of future they are willing to build. Pakistan is not too late. But the window for deliberate, sovereign participation is narrowing. The question is whether the country will think for itself or simply be thought for.
The writer is a Lahore-based public policy analyst and can be reached at [email protected].