Will AI kill your job?
ECONOMIC progress does not exhibit a linear trajectory. Episodes of abundance and prosperity sit alongside instances of pessimism and hardship. Today’s world is no different. Aside from the glumness unleashed by the recent US-Iran war and its repercussions, perpetual anxiety seems to have enveloped the economic landscape since at least a decade or more. And it has to do with how artificial intelligence (AI) will shape the economy and job market. Some of the predictions paint such a frightening picture that it would make Dr Doom himself blush.
But is the probable future really so terrifying? As with most such queries in economics, let us turn to economic history and see what it divulges, mainly because this fear has resonated persistently across history — at least since clergymen William Lee invented the ‘stocking frame’ in the late 16th century (Elizabeth I had refused to grant him patent, fearing joblessness for hand-knitters).
In 1927, US labour secretary James J. Davis, who served under three presidents, dwelt upon this issue. Here are a few excerpts from his speech: “Every day sees the perfection of some new mechanical miracle that enables one man to do better and more quickly what many men used to do. … But what is all this machinery doing for us? What is it doing to us? … …one man now does what 41 men formerly did. What are we doing with the men displaced? … If you take the long view, there is nothing in sight to give us grave concern. I am no more concerned over the men once needed to blow bottles than I am over the seamstresses that we once were afraid would starve when the sewing machine came in. We know that thousands more seamstresses than before earn a living that would be impossible without the sewing machine. In the end, every device that lightens human toil and increases production is a boon to humanity.”
That was 1927 — Davis asserting that technological progress brings more opportunities and progress than problems and misery.
Technology tends to alter the nature of specific work rather than completely replace it.
But later years kept bringing up similar fears. The most prominent voice amongst them was that of Dr Norbert Wiener, widely regarded as a pioneer of modern AI (which he called ‘cybernetics’). In 1948, he warned of mass unemployment and an industrial revolution of “unmitigated cruelty”. His 1960 paper (‘Some Moral and Technical Consequences of Automation’) insisted that it was different this time round as advanced machines would now out-evolve and rule over men.
Forward to 1982. Economist and Nobel laureate Wassily Leontief penned his fears of how automation would usher in large-scale unemployment in the coming years due to thinking machines substituting human intellect, just like manual machines had been replacing human muscle. He cited job losses due to ATMs as an example. (Interestingly, former president Barack Obama cited the same in 2011).
Now to the present day — 2026. A decade back, as new AI software began performing the tasks of radiologists, no less than the 2024 Nobel laureate Geoffrey Hinton (‘godfather of AI’) opined that we should stop training radiologists as they would soon become redundant.
How do all these predictions stack up against reality? Let’s begin with radiologists. It’s 2026 and more than 700 approved radiology software progammes are in operation, yet human radiologists are more in demand than before and their average salary has jumped substantially.
How about job losses due to ATMs (the first one was installed in 1969) and other inventions in the financial sector? James Bessen gathered the data and shared his findings in 2015 (‘Toil and Technology’), which showed that the number of bank tellers actually rose with the increase in ATMs.
There are countless other examples. And despite all the doomsday warnings, global per capita income has tripled since 1950, global trade has grown 45-fold in volume, and net job creation (jobs destroyed minus jobs created) during 1977-2023 in the US, the most technologically advanced economy at this time, remained positive at approximately 70 million.
What this demonstrates is that give and take all (‘net’ in econ lingo), technological growth has expanded the economic envelope rather than shrinking it. Within this expanded envelope, job creation in the long run is usually higher than job destruction. Certain industries or services may vanish, but new ones usually arise, and many of same experience growth in related work.
The World Trade Report 2025 also predicts something similar, concluding that AI would lead to a significant increase in real income and global trade.
This, however, does not imply that we take our eyes off this matter. As James Davis noted, the probability of short-term displacements is real; specific sectors and jobs may see severe contraction or even redundancy. In such episodes, it’s the worker that is usually at the receiving end. It’s here that an active public policy (fiscal and monetary components) has a fairly critical role to play.
Mostly though, technology tends to alter the nature of specific work rather than completely replace it. Part of the reason is that automation is not without cost, and partly because humans and machines don’t happen to be perfect substitutes. Or as in the case of ATMs, automation made it cheap enough to substantially expand the number of branches, requiring even more tellers whose job nature now bent towards customer services.
The march of technology is relentless; its spillovers are real. It throws up challenges, but also opportunities that the most dynamic and entrepreneurial economies are best placed to optimise, with the decline of some sectors and services usually accompanied by the rise of others. Any possibility of the most pessimistic scenarios becoming a reality would mean the complete upending of what history has witnessed till now in terms of technological change and its resulting economic impact, ushering in something so unexpected that the world has never witnessed.
I believe we can discard such a possibility as this era of AI will probably be no different. For the moment, cast aside the doomsday scenarios.
The writer is an economist. His current research focuses on economic reforms, the history of economic thought and the long-term analysis of the issues plaguing Pakistan’s economy.
Published in Dawn, May 22nd, 2026
