Policy bazooka could fend off jobless AI world

The stock market is in a panic about the impact artificial intelligence will have on the jobs market. It’s reminiscent of the “lump of labour fallacy” taught in high school economics. The lesson goes that even if technology disrupts industries and creates unemployment, consumers will gain from cheaper goods. Over time, savings are spent elsewhere, which helps create new jobs. But even if the AI hysteria is overdone, there is a real risk that fiscal and monetary policy will be too slow to fix short-term disruptions.

There are early signs of moderate AI damage to the workforce. In the US, changes in joblessness since 2022 correlate with adoption of AI in the Real Time Population Survey data, with a third of the increase coming from industries with above average use of the technology. The “information” sector — telecom operators, broadcasters and publishers — report 70 percent adoption and a 75 percent rise in unemployed workers.

This uptake, however, has rattled investors who worry that the software industry’s business model will be threatened by Anthropic’s Claude Code. Concerns widened on Monday sparked by a widely circulated Substack post by Citrini Research which laid out an “intelligence crisis” in 2028. Citrini asked whether mass white collar displacement could trigger a collapse in spending that tanks profits across consumer sectors. The hit to GDP could be long lasting because, even if retailers also embrace AI and workers shift into gig roles, the economy would become more capital heavy and skew gains even more starkly towards the rich, who run down less of their wealth than middle class workers spend out of income.

This echoes past “underconsumptionist” theories, including that of Marxist economist Rosa Luxemburg. Yet historically, technology shocks have boosted GDP, not reduced it. Even if AI destroyed jobs at an unprecedented scale, that would imply steep falls in the cost of goods and services, raising real incomes. It could also make remaining jobs more stable, better paid and more prestigious, even those that are currently classified within the “gig economy.”

Crucially, money can be snapped into existence, unlike scientific advancement. During the pandemic, Western economies largely shut down, yet unemployment and output quickly rebounded because governments injected fiscal stimulus worth over 6 percent of GDP. The lesson should have been that deficient demand can be managed with effective policy.

In 2026 that thinking seems off the table. The post-2022 global inflation spike, mostly driven by supply shortages, has turned officials and voters against activist fiscal policy. Budget deficits remain high because governments lack the political capital to raise taxes. But a willingness to widen them again to counter a diffuse, slow moving threat — unlike the Covid 19 virus — will probably be limited.

Meanwhile, central bankers are keeping interest rates high, prioritising their need to avoid another inflation mistake while trying to pre-empt a worsening labour market. Even if an AI job apocalypse future is unlikely, it’s concerning that officials don’t seem more willing to stave one off.

Shares in computing firm IBM lost more than 13 percent on February 23, the largest single-day fall in 25 years. The sell-off coincided with news from artificial intelligence giant Anthropic which revealed on the same day that its Claude Code tool can help modernize software written in the COBOL programming language, which underpins many legacy systems used by banks, insurers and government agencies.