Bay Street in Toronto’s financial district. According to one study, fewer than 5% of layoffs in the U.S. were due to AI adoption last year.Mark Brinch/Globe and Mail
John Rapley is a contributing columnist for The Globe and Mail. He is a writer and scholar whose books include: why empires collapse and Twilight of the gods of money.
“Unemployment remains high at 6.8 per cent, youth unemployment is particularly high, and relatively few businesses say they plan to hire workers in the coming months,” Governor Tiff Macklem said after the Bank of Canada decided to keep interest rates on hold Wednesday.
It’s a tough job market, especially for new graduates. Anecdotal evidence places the blame on AI. Although large-scale worker replacement has not yet occurred, companies still claim that the need for future workers will be reduced. After Amazon announced last year that it would cut 30,000 jobs, blaming AI, a Harvard University research report found that companies implementing AI were cutting new hires while retaining existing staff. The company needed senior employees to continue quality control, but AI can produce similar results, so there was no need to hire junior employees.
However, you don’t have to dig deep into these reports to find something that’s poorly calculated. While surveys of employers do show that they plan to delay hiring due to AI, actual employment numbers do not yet show a link between AI adoption and job losses. For example, a study conducted in the US last year found that AI implementation resulted in fewer than 5% headcount reductions.
Amazon itself acknowledged that the layoffs were a reduction in overhiring during the pandemic, but Harvard University research data revealed that the trend of layoffs exists regardless of whether a company implements AI. This seems to suggest that the real cause of the employment freeze was macroeconomic. Recent studies and reports from Yale University’s Budget Institute The Financial Times similarly says that AI has had little disruptive impact so far. Most of last year’s layoffs were instead due to the cyclical slowdown in the economy.
But while AI may not be the culprit behind today’s poor job market, the AI bubble may be. U.S. GDP growth remains strong, largely because so much capital is flowing into the AI sector, driving up stock prices and sustaining consumption among the wealthy. But in other parts of the economy, hiring has stalled, wage growth has slowed, prices have risen and consumer confidence has fallen to record lows.
Add to this the instability of the Trump administration’s policy decisions and the fact that employers in most economies are putting hiring and expansion plans on hold. This, in turn, has implications for the United States’ major trading partners, including Canada.
What you need to know about the AI bubble – and how it will burst
But just because the employment slowdown is cyclical and will probably end once the economy recovers doesn’t mean it’s not a serious problem. When new graduates struggle to find work, it can impact their lifetime earnings and productivity. As long as they are unemployed, not only will the skills they have acquired through training become obsolete, but they will also be unable to acquire new skills on the job.
Governments can therefore play an important role in smoothing the employment cycle and enabling young people to enter the workforce and increase productivity by providing incentives for employers to create internships and bridge employment. However, there is another, more important reason why the government should do more to support the employment of new graduates.
So far, the evidence is that AI is making employees less productive, not more. A recent study published in Harvard Business Review found that employees are using AI to create “workslop,” content that “masquerades as good work but lacks the substance to meaningfully advance a given task.” As a result, colleagues had more work to do to organize and make sense of the artifacts.
What I learned from teaching using AI
This is echoed in a recent survey of workers who reported that AI has increased productivity, but reduced quality and required more cleaning. Emphasizing quantity over quality is not the secret to financial success.
But AI’s unfortunate impact may not stem from any inherent shortcomings in the technology. Rather, it may be due to bad policy. To find out how, let’s look at two different approaches to AI.
Currently, the United States and China are both dominating the AI revolution. The United States has adopted a deregulatory approach, allowing large corporations to expend vast amounts of capital and energy in the pursuit of artificial superintelligence (AI that could one day replace humans). In contrast, the Chinese have focused on developing “good enough” AI, incorporating it into machines and products, and have proven to be highly innovative. To achieve this goal of widespread adoption, education systems are preparing students to use AI more effectively for beneficial outcomes.
So if Canada really wants to prepare to emerge on the other side of cyclical recession, it should similarly encourage employers to let young people experiment with AI in ways that harness its potential, and to learn how to better harness AI rather than fear it. And once the U.S. AI bubble bursts, this technology may be put to practical use.
