The Automation Question

The Automation Question

Why the debate over machines and work is really a debate about choices

A Few economic questions provoke as much anxiety as the fear that machines are coming for our jobs. Each wave of new technology — the power loom, the assembly line, the microchip — has been met with predictions of mass, permanent unemployment, and each time the prophecy has failed to arrive. Today the same warning is issued about artificial intelligence and robotics. In my view, this recurring alarm is largely overstated: the historical record gives little support to the idea that automation destroys work in aggregate. Yet dismissing the anxiety entirely would be a mistake, because the more subtle disruptions automation brings are real, and our institutions have been slow to address them. The dislocations it causes fall unevenly, and the workers displaced are rarely the ones who enjoy the new opportunities that eventually appear.

B One of the great puzzles of recent decades is what economists call the productivity paradox: an era of extraordinary technological advance has coincided with historically sluggish growth in measured productivity. If computers and software are as transformative as their advocates claim, output per worker should be soaring. Instead, in most wealthy economies it has crawled. Explanations abound. Some argue the benefits simply take time to diffuse through the economy; others suspect that the gains are real but escape conventional statistics. This mismatch between the visible power of digital tools and the disappointing figures has become one of the central controversies of modern economics, and how one resolves it shapes almost everything one concludes about the future of work.

C The economist Marcus Renn draws heavily on the long historical record to argue that such fears are misplaced. Across two centuries, he notes, technologies that eliminated particular occupations also generated entirely new ones, often in numbers no one could have foreseen. Automating one task tends to raise the value of the tasks that remain, and cheaper goods leave consumers with money to spend elsewhere, creating demand and jobs in unrelated sectors. The telephone operator and the lift attendant vanished, yet employment kept rising, buoyed by occupations that would have been unimaginable to earlier generations. For Renn, the persistent error of the pessimists is to treat the quantity of work as fixed — a mistake sometimes called the 'lump of labour' fallacy.

D Not everyone finds this reassuring. Priya Anand accepts that automation need not reduce the total number of jobs, but insists that this is the wrong thing to measure. The decisive question, she contends, is who captures the gains. As machines take over routine work, the returns tend to flow to those who own the capital and to a shrinking group of highly skilled workers, while wages in the middle stagnate. Anand is also wary of drawing comfort from the past: previous transitions, she warns, unfolded over generations, whereas today's may arrive too quickly for workers and communities to adjust. A different objection comes from Tobias Fell, who focuses on measurement. Much of what modern technology delivers — free digital services, greater convenience, better-quality goods — is poorly captured by standard output figures, so the official statistics, he argues, understate the value that has actually been created.

E My own sympathies lie closer to Anand. The obsession with headline unemployment figures distracts from the genuine danger, which is not too few jobs but a widening gap between those who benefit from automation and those left behind. Whether automation proves a blessing or a curse will depend less on the technology itself than on two things: how the resulting gains are distributed, and whether workers are given real opportunities to acquire new skills. These are political and institutional choices, not technological inevitabilities, and it is a serious error to speak of them as though the outcome were already written.

F If that is right, the policy response matters enormously, and here I believe we have fallen short. Retraining programmes in many countries remain underfunded and poorly designed, reaching too few of the workers who need them most. Education systems still prepare students for a labour market that is disappearing rather than the one emerging. Some commentators propose a universal basic income — a regular, unconditional payment to every citizen — as a way of cushioning the disruption. Others favour wage subsidies, shorter working weeks, or stronger collective bargaining. Each proposal carries its own trade-offs, and the debate among them is far from settled.

G The lesson of history, then, is not that we should be complacent, but that we should be precise about what we fear. Machines are unlikely to leave us with nothing to do. They are, however, entirely capable of making our societies more unequal and more anxious if we allow them to. The future of work will be decided not in the laboratory but in the choices we make about education, taxation and the rules that govern the labour market. Those choices remain ours to make, and that, ultimately, is a reason for cautious optimism rather than despair.