Guardian: AI is already cutting jobs, and the energy crisis could speed up layoffs

Artificial intelligence may hit the labor market considerably faster than earlier waves of technology, and the new energy crisis only heightens that risk. When economic growth slows, raw materials and energy get more expensive, and automation tools become ever more accessible, choosing machines becomes almost mechanical for business. In this combination, the problem is no longer a debate about the distant future: the question is how quickly governments can cushion the blow for people whose tasks AI can take over right now.

In his column, Larry Elliott frames this moment through the old idea of "creative destruction." Capitalism constantly renews itself, pushing out outdated ways of working in favor of new ones, and every such transition is painful for some workers. The difference is that technology used to automate mainly physical labor, while the current wave of AI is moving into cognitive tasks. It is no longer only about the assembly line, the warehouse or the checkout, but also about document analysis, customer support, text drafting, basic software development, accounting operations and other office work that until recently was considered relatively safe from automation.

Such transitions usually go more smoothly when the economy is growing and the labor market can quickly absorb people who have lost their previous positions. Then states have time for retraining, companies have time to restructure processes gradually, and employees have a chance to move into new roles without a long gap. But in the author's view, the present backdrop is almost the opposite.

Even before the latest escalation of the conflict in the Middle East, global growth looked weak and hiring prospects unstable. After the spike in geopolitical tension, expensive energy, raw-material disruptions and gloomier economic expectations were added to the mix. The IMF has already lowered its growth forecasts, which means it is even harder for businesses to keep their staff and even easier to justify accelerated automation.

This is where the energy crisis and AI begin to reinforce each other. When companies face rising costs for electricity, logistics and materials, they cut costs more aggressively. If technology is available at that moment that can replace part of office work faster and more cheaply than before, management has a strong incentive to remove people from processes.

Optimists respond that history has many times frightened society with "machines that will take away jobs," yet in the end new technologies created more employment than they destroyed. Elliott believes there are at least two reasons not to be complacent this time: AI may turn out to be a far more general-purpose technology than earlier waves of automation, and new jobs, even if they appear, are by no means obliged to be as well paid as the ones that disappeared.

This leads to the scenario that worries the author most. If automation starts by knocking out well-paid white-collar positions, it will hit consumer demand. Machines really can work around the clock, take no vacations and never get sick, but they do not buy cars, rent homes, or spend money on restaurants and household services.

The research firm Citrini, cited in the column, described a possible 2028 crisis in exactly these terms: companies adopt AI en masse for efficiency, lay people off, demand in the economy falls, revenue sags, and businesses respond with a new round of cuts and even more automation. Such a vicious circle could hurt not only employment but also the stock market. The paradox is that the crisis would then arise not from AI failing, but from AI being deployed too successfully at the level of individual companies.

The main conclusion of this logic is highly practical: it is no longer enough for governments to simply welcome innovation and hope the market sorts everything out by itself. Fast, large-scale measures are needed: retraining, a new industrial policy and a redistribution of the gains from productivity growth. Otherwise most of the benefit will go to a narrow circle of tech companies and investors, while society gets prolonged pressure on employment, weaker demand, and an economy in which automation raises business efficiency but at the same time undermines the foundation of mass consumption.