Why AI hasn't put you out of a job yet

Almost every week, the world takes another step towards artificial superintelligence.

The most powerful AI models today can perform a staggering range of tasks - from writing detailed reports to creating on-demand video content. So-called “hallucinations” (inaccuracies in content generation) are becoming less of a problem.

It’s no wonder that many people are increasingly afraid of being made redundant. Earlier this year, global Google searches for “AI unemployment” reached an all-time high. In cities like London and San Francisco, the question “How long do you think you have left?” has become a common topic of conversation.

However, the question is, is ChatGPT really taking away someone's job?

Many commentators argue that it is taking away. They often refer to recent research by Carl Benedikt Frey and Pedro Llanos Paredes of Oxford University, which points to a link between automation and a declining demand for translators. However, official US data at the same time shows that the number of people employed in interpreting, translation and similar occupations is as much as 7 percent higher than last year.

Others point to Klarna, a financial technology company that previously boasted about using artificial intelligence to automate customer support. But now it’s making a U-turn. “There will always be the option to talk to a human, if you want,” its CEO, Sebastian Šemjatkovski, said recently.

Others continue to scour macroeconomic data for signs of an AI-job apocalypse.

One popular measure is the ratio of the unemployment rate among recent college graduates to the overall American average. Young college graduates are now more likely to be unemployed than the average worker. The explanation is that they tend to hold entry-level jobs in knowledge-based industries—like paralegals or making presentations at consulting firms. These are the kinds of jobs that artificial intelligence can do well. So, maybe AI has eliminated those jobs?

Well, it doesn't. The data simply doesn't fit any logical assumption. The "relative unemployment" of young graduates started rising as early as 2009, long before generative AI technology came along. And their actual unemployment rate, which is around 4%, is still low.

If we go back to the measure we introduced in 2023, we analyze US employment data by occupation, focusing on the types of workers often believed to be at risk from AI. These are white-collar workers - people who work in back-end support, financial operations, sales and similar areas. Here too, a similar pattern emerges: we find no evidence of a negative impact of AI. In fact, the opposite is true - over the past year, the share of white-collar workers has increased slightly.

Overall, the unemployment rate in the US remains low at 4,2%. Wage growth remains quite strong, which is difficult to reconcile with the claim that AI is reducing the demand for labor. Trends outside the US point in a similar direction. Wage growth in Britain, the eurozone, and Japan is strong. In 2024, the employment rate in OECD countries - which measures the proportion of working-age people who are actually employed - reached a historical high.

There are two competing explanations for these trends. The first is that, despite the constant announcements about how companies are introducing artificial intelligence into every aspect of their business, very few are actually using AI for serious work. Official statistics show that less than 10% of American companies are using AI to produce goods and services.

Another explanation is that even when companies introduce AI, they don't lay off employees. AI may simply help workers do their jobs faster, rather than making them redundant.

Whatever explanation is correct, there is no reason to panic for now.

Translation: NB