Fear that AI may take away jobs from young people seems misguided, writes Viet Vu. Jobs in Canada’s technology sector have been in decline since 2021.Dado Ruvic/Reuters
Viet Vu is the manager of economic research at the Dais, a think tank at Toronto Metropolitan University. Jake Hirsch-Allen is the director of partnership at the Dais.
“93% of investors say AI will destroy jobs, governments not prepared.” That sounds like a headline pulled from the pages of last week’s news, but Forbes published those words in 2016. Similar headlines appeared in the 1980s: “A robot is after your job.” The 1960s: “Counterrevolution to automation is feared.” The 1950s: “Uncle Sam finds uses for automation, Congress has it under investigation.” The 1940s: “Do machines displace men in the long run.” The 1930s: “Machines as ministers to men.” And the 1920s: “The machine age and its effects on the literary arts.”
It is not surprising, then, that this debate – which has captured attention for more than a century – has re-emerged with the advent of new automating technologies. Yet again, we’re told that this “time is different.”
In fairness, the specific target of the fear has shifted. In the 1990s, the conversation focused primarily on industrial sectors such as car manufacturing. In the 2010s, “middle managers” and specific occupations such as trucking were “most at risk.” Now, we are told tech is coming for white-collar jobs and early-career opportunities.
Much has been made about the rise in youth unemployment in Canada this year as a marker for the first impacts of AI. However, the data does not support this reason for the uptick. Youth unemployment has been on the rise, but not to unprecedented levels. And the trend more likely reflects a reversal to the mean from an especially tight labour market, which emerged from the pandemic, and weakening macroeconomic conditions owing to trade tensions with the U.S.
Earlier: AI adoption is upending the job market for entry-level workers
Observers also point to the most “AI-exposed sector,” the tech sector, noting the higher youth unemployment among computer science graduates. This has created a narrative around companies “replacing junior developers with AI.”
At least for the time being, this fear also seems misguided. Jobs in Canada’s technology sector have been in decline since 2021, when overinvestments during the pandemic were followed by necessary corrections during the recovery that followed. Increasing interest rates around the world to combat inflation also decreased access to cheap capital, as investors attracted to risky investing amid low interest rates left the table. The correction involved layoffs of the existing workforce, and a slowdown of capital has depressed new job postings and growth.
This reduction in demand for tech jobs came after historic government and higher education investments to encourage young people to study computer science, leading to an average annual increase of 11.9 per cent in program enrolment over 10 years, according to Statistics Canada. In sum, this created an even greater mismatch of supply and demand for computer science graduates.
It is true that some sectors are more exposed to generative AI than others; software development and customer support occupations top many lists. But we should differentiate between exposure to AI (being affected) from the imminent or dramatic impact of AI (the resulting consequences). For example, even in software development, the productivity impacts of incorporating generative AI tools continue to be debated where gold-standard RCT studies are producing ambiguous productivity results.
More broadly, enterprise adoption of AI seems to be slowing where AI adoption by large firms in the U.S. has stalled, and the vast majority of enterprise generative AI adoption efforts have failed, according to a report by MIT NANDA. Our own research, looking at Canadian firm-level productivity after AI adoption in the 2020–22 period, found no impact, at least in the short term.
Unlike Geoffrey Hinton and Dario Amodei’s predictions of imminent and cataclysmic job loss, prominent economists who study the impacts of AI on the labour market, including Daron Acemoglu, foresee only a modest impact on long-term productivity. Their conclusions, and ours, apply equally to Canada, the United States and much of the developed world.
Many companies are crediting AI with layoffs this year, but it is impossible to distinguish between AI-led productivity gains versus efforts to use AI announcements to shore up investor confidence. But when it comes to whether AI is affecting the broader job market for entry level-positions, the evidence overwhelmingly suggests it has had no measurable impact so far.
While there is uncertainty about how AI will shape the future of work, with luck, writers in 2035 will read 2025 headlines as signs we chose evidence over fear.