After the Bank of Canada voted to keep interest rates steady on Wednesday, Governor Tiff Macklem noted that “the unemployment rate remains high at 6.8 per cent, youth unemployment is particularly elevated and relatively few businesses say they plan to hire more workers in coming months.”
It’s a tough job market, alright, not least for new graduates. Anecdotal evidence lays the blame on AI. Although it isn’t yet replacing workers on a large scale, firms are nonetheless saying it’ll reduce their future need for workers. After Amazon announced it would eliminate 30,000 jobs last year, blaming AI, a working paper out of Harvard revealed that AI-adopting firms were cutting new hires, all while keeping existing staff: They needed senior employees to continue doing quality-control, but didn’t require junior hires since AI could produce similar outputs.
But you don’t have to drill deep into these reports to find something which doesn’t quite compute. Although surveys of employers indeed find them planning to slow hiring owing to AI, the actual job numbers don’t yet show a link between AI adoption and reduced employment. Last year in the U.S., for instance, one study found that fewer than 5 per cent of layoffs resulted from AI adoption.
Amazon itself admitted that its retrenchment was a rollback of the excessive recruitment during the pandemic, while the Harvard study’s data revealed the trend toward layoffs existed whether or not firms were adopting AI. That would seem to suggest that the real cause of the hiring freeze was macroeconomic. Recent studies by Yale’s Budget Lab and reporting fromthe Financial Times similarly found little disruptive impact so far from AI. Most of last year’s layoffs resulted instead from a cyclical slowdown in the economy.
But while AI may not be the culprit in today’s bad job market, the AI bubble might. Although U.S. GDP growth remains strong, it’s primarily because so much capital is flowing into the AI sector, driving up share prices and sustaining consumption among the rich. But in the rest of the economy, hiring has stopped, wage growth is slowing, prices are rising and consumer confidence is plumbing record lows.
Add in the volatility of the Trump administration’s policy making and the fact that employers in most of the economy are holding off hiring or expansion plans. That, in turn, is affecting the U.S.’s major trading partners, not least Canada.
However, just because the slowdown in hiring is cyclical – and will presumably end when the economy picks back up – that doesn’t mean it isn’t a serious problem. When recent graduates struggle to find work, it can affect their lifetime earnings and productivity: For as long as they remain unemployed, not only do the skills they acquired in training grow dated but they fail to acquire new ones on the job.
The government can thus play an important role smoothing out the job cycle by providing incentives to employers to create internships or bridging appointments, so that young people can enter the work force and become productive. But there’s another, even more important reason the country should be doing more to help new graduates find work.
So far, there’s evidence that AI is making workers less, not more, productive. One recent study in the Harvard Business Review found that workers were using AI to produce “workslop” – content that “masquerades as good work, but lacks the substance to meaningfully advance a given task.” The result was more work for colleagues tidying up or making sense of the outputs.
This was echoed in a recent survey of workers, who reported AI helped them speed up their output but reduced its quality, forcing a lot of clean-up work. Quantity over quality is not a great recipe for economic success.
But the disappointing impact of AI may not result from the technology’s inherent shortcomings. It may rather be a function of poor policy. To see how, we can look at two distinct approaches to AI.
Together, the U.S. and China currently dominate the AI revolution. The U.S. is using a deregulated approach, allowing large companies to consume vast amounts of capital and energy in the pursuit of artificial superintelligence – the AI that will supposedly one day replace humans. The Chinese, in contrast, have focused more on developing “good enough” AI that they then embody in machines and products, which are proving highly innovative. To meet this goal of widespread adoption, the education system is preparing students to use AI more effectively, with beneficial results.
In short, if Canada really wants to be ready to take off on the other side of a cyclical downturn, it should be incentivizing employers to similarly get young people to experiment with AI in ways that exploits its possibilities – to learn how to better use it, not fear it. Then, when the U.S. AI bubble bursts, the technology may really find its use.