Today, experts are fairly divided on how AI will affect the labour market. One camp argues that AI will take over many occupations, leaving large groups of people unemployed. The other believes that AI will primarily make us more productive. Just as in previous technological shifts, the jobs that disappear will be replaced by even more jobs in entirely new occupations.
One thing almost everyone agrees on, however, is that AI is developing incredibly fast. It is now almost exactly four years since ChatGPT was launched publicly, and the improvements over that relatively short period have been dramatic. Despite this, there is still no clear evidence that AI is making large groups of people unemployed.
Admittedly, there are increasingly frequent reports of companies planning to replace employees with AI. But in macroeconomic indicators such as employment and unemployment, there are still no clear effects to be seen.
One possible effect of AI that has been observed in several countries, however, is that young people appear to be finding it increasingly difficult to enter the labour market. In the US, the situation has been described as “low hire, low fire”: not many people are being hired, but not many are being laid off either.
In Sweden, too, a recent study by the Swedish Agency for Growth Policy Analysis shows that labour mobility is low. While there are still no signs of a broad structural transformation of the economy, there are clear indications that young people are being affected. Among recent graduates from programmes with high exposure to AI, employment income has grown significantly more slowly since 2023, while a declining share are finding jobs at the companies most exposed to AI.
Built-in inertia is holding things back
There are undoubtedly many reasons why AI has not yet had a major impact on the labour market. One could be that many companies are already in a position where they could reduce their workforce with the help of AI, but built-in inertia is preventing this from happening on any significant scale.
As long as a company is doing well and making healthy profits, everyone is fairly content. Shareholders get their returns, employees are happy, and management faces little external pressure to start cutting jobs. Those kinds of uncomfortable decisions are normally made only when business starts to deteriorate. At the same time, there is little need for major new recruitment, since AI allows companies to produce more with their existing workforce.
Risk of a sudden tipping point
Personally, I belong to the group of economists who believe that AI will ultimately have a significant impact on the labour market. It is difficult to escape the fact that AI is rapidly getting better at almost everything. AI will be able to perform many jobs better and at lower cost than humans, suggesting that many people will eventually be outcompeted.
But if such a profound transformation also happens rapidly, there is a risk that the economy and society will not have time to adapt, causing widespread discontent and social tensions that may prove difficult to manage.
As long as AI is not sufficiently general and reliable to do everything humans can do, most jobs will remain safe. But once AI becomes good enough to cross that threshold, it could suddenly become capable of performing a vast range of jobs. In particular, the rapid development of AI agents that can act autonomously and carry out tasks offers a glimpse of where we may be heading. The risk is that we could reach a tipping point, with the impact on the labour market suddenly accelerating in the years ahead. So how can we manage such a scenario?
The key to an orderly transition in such a situation is for new jobs to be created at a pace that gives both the economy and people time to adapt. But where will those jobs come from?
More bureaucrats?
Politicians will have a key role to play in ensuring that the transition takes place at a pace society can cope with. As part of that process, it is quite possible that many new jobs will be created through political decisions.
Perhaps becoming a bureaucrat is a future-proof career choice. The word has negative connotations today, but that too could change. In practice, there are many situations in which we actually want a human being involved in decisions that affect us, whether it concerns a municipal building permit or a loan application at a bank.
There is, of course, a cost. If tasks that could technically be automated become more expensive and slower because a human has to be involved, someone will have to pay for that through taxes or higher prices. But if the alternative is mass unemployment and growing social tensions, that may still be a price well worth paying.
So do not be surprised if the number of “bureaucrats” increases as AI begins to make serious inroads into the labour market. Nor should this necessarily be seen as a failure, but rather as part of the process of helping society adjust to a new era.