The steady advance of artificial intelligence into one industry after another has turned its impact on employment into one of the defining economic questions of our time. While optimists foresee a net creation of jobs, pessimists brace for widespread unemployment. Having examined both cases, I am convinced that the eventual outcome will be dictated less by the technology than by the policies that accompany it.
The optimistic argument rests on the observation that AI, like earlier innovations, spawns as much work as it displaces. Entire professions have already emerged around building, training and governing these systems, and as automation lifts productivity, growing companies tend to expand their human workforce in domains machines handle poorly, from strategy to creative design. Freed from drudgery, employees can devote themselves to tasks that reward judgement and imagination.
Yet the pessimists raise a concern that history does not fully dispel. AI's reach is unusually broad, threatening routine roles in manufacturing, administration and customer service almost simultaneously, and the workers displaced are disproportionately those with the fewest qualifications and the least capacity to retrain rapidly. Compressed into a short period, such disruption could deepen inequality and leave a significant minority permanently behind.
My own conviction is that these outcomes are not predetermined but chosen. Where states and employers invest seriously in lifelong education, portable skills and social support, the transition can be absorbed and even harnessed. Neglected, however, the same forces could hollow out the middle of the labour market. AI, in short, is a powerful amplifier of whatever choices societies make, and mass unemployment is a risk to be managed, not a destiny to be accepted.