The AI Jobs Apocalypse: A Myth Whose Time Has Passed?

Recent reports from Anthropic and OpenAI have sparked a reevaluation of the impact of artificial intelligence on employment. For years, tech industry voices warned about an impending “jobs apocalypse,” with some predicting that AI would wipe out up to half of all entry-level jobs within a few short years.

These dire predictions were based on AI’s supposed ability to take over nearly every task, from menial labor to complex problem-solving. However, Anthropic’s latest analysis suggests this narrative has been greatly exaggerated or misinformed. According to their report, there has been no systematic increase in unemployment for highly exposed workers since late 2022, and deployment of AI technology remains a fraction of what’s feasible.

One explanation for the discrepancy between predictions and reality is the “O-ring argument.” This concept draws on the space shuttle Challenger disaster, suggesting that even if AI cannot perform every task perfectly, it can still increase the value of remaining tasks. In other words, as long as AI takes over some tasks, humans are free to focus on higher-value work.

However, this line of thinking glosses over a crucial point: the actual impact of AI on employment is complex and nuanced. Despite strong substitution at the task level, overall employment effects are modest – reduced demand in exposed occupations is offset by productivity-driven increases in labor demand at AI-adopting firms.

The politics surrounding AI have taken an unexpected turn. As more Americans become aware of AI’s potential downsides – from energy consumption to job displacement – public opinion has shifted decidedly against the project. Seven in 10 Americans now oppose building AI datacenters in their area, indicating that the “jobs apocalypse” narrative has damaged the public’s perception of AI.

Research on the labor market impact of artificial intelligence is still in its infancy. Economists like David Autor argue that AI is improving rapidly and shows no sign of hitting a ceiling anytime soon. However, insiders are acknowledging that something has gone amiss – or perhaps it’s simply recognizing that the dystopian AI future remains very much on the cards.

It’s worth recalling Robert Solow’s quip from the early years of the computer revolution: “You can see the computer age everywhere but in the productivity statistics.” It took another decade for computers to show up in the stats, and it may take just as long for AI to deliver on its promise of vast economic opportunity – at a price that humanity is willing to pay.

The economics look fragile. How much of GDP are we willing to invest in AI datacenters? According to some estimates, we’re headed toward 20%, 30%, or even 40% – and the International Energy Agency warns that power demand from datacenters will more than double by 2030.

As we reexamine our assumptions about the impact of AI on employment, it’s time to consider whether the “jobs apocalypse” was always a myth whose time has passed – or if it simply requires a more nuanced understanding of what AI can and cannot do.