A study found that nearly three-quarters of the country’s 1.1 million public sector workers are in roles that are highly exposed to AI.Cole Burston/The Globe and Mail
A greater share of jobs in the public sector are vulnerable to automation by artificial intelligence tools than the Canadian work force as a whole, according to a new study from The Dais think tank at Toronto Metropolitan University.
The study released Monday found that nearly three-quarters of the country’s 1.1 million public sector workers are in roles that are highly exposed to AI. Some of those jobs, about 25 per cent, are complementary with AI, meaning the technology can assist or augment the work.
But 49 per cent of public sector jobs are made up of tasks that “current AI technologies are well positioned to substitute or replace.” The same is true for about 29 per cent of the overall Canadian work force. (The study was limited to direct government jobs, and eliminated some, such as education and health care workers, from its tally.)
The federal public service has the highest share of such workers, at 58 per cent, followed by the provinces. Both public sectors are more heavily stacked with people in business, finance and administrative roles, such as human resources professionals, auditors and accountants. These roles can involve repetitive tasks suited for AI. Municipalities employ more front-line workers, such as firefighters and landscapers, according to the study, which are less affected by generative AI.
AI adoption is upending the job market for entry-level workers
“The potential downside risk is as large as the opportunity,” said Viet Vu, manager of economic research at The Dais. “It all depends on how you end up deploying the technology.”
There are plenty of ways to use generative AI tools powered by large language models (LLMs) in the public sector to improve productivity. That includes transcription and translation, assisting with writing and editing reports and briefs, and even analyzing geospatial data for traffic and land use.
There is also the risk that roles are replaced outright by AI tools. But Mr. Vu said the reality of some jobs is more complicated than that. He gave the example of using AI to summarize policy documents. “One potential instinct is the government to say, ‘We don’t need any policy analysts any more,’” Mr. Vu said. But that task is just one of many performed by policy analysts.
Still, he could foresee the government deploying AI tools to help policy analysts improve productivity. “LLMs will allow higher efficiencies to be gained, which may end up with fewer people in specific roles,” he said.
That risk of displacement is one reason why the report recommends the public sector think seriously about long-term work-force planning, including efforts to provide training and education to help ensure those most vulnerable to displacement can transition to other jobs. Labour unions also need to be involved in these discussions, Mr. Vu said.
Another of the report’s recommendations is for public sector agencies to create an environment to experiment with AI and track successes and failures, including for low-risk repetitive tasks where the technology can augment work.
Prime Minister Mark Carney has said the federal public service needs to become more productive by “deploying AI at scale,” which can also help improve service quality for Canadians. Last week, Ottawa said it had signed a non-binding agreement with Cohere Inc. to look for ways to deploy AI within the public service.
Mr. Carney campaigned on a promise to cut back on operational spending, and ministers have been directed to find ways to do so in their departments. Joël Lightbound, the Minister of Government Transformation, Public Works and Procurement, has said the government plans to reduce the size of the public service through attrition, while looking for ways to boost efficiency with technology.
In the private sector, some chief executive officers are musing openly about slowing hiring or even reducing work forces because of AI. Canadian companies Klue Labs Inc. and Scinapsis Analytics Inc. have already made job cuts related to AI.
Dario Amodei, the CEO of AI developer Anthropic, made headlines earlier this year with his pronouncement that AI could eliminate half of entry-level white collar jobs.
But that opinion strikes others in the industry as farcical. “Dario was also saying a couple years ago that AI was going to kill us all,” Cohere CEO Aidan Gomez told The Globe and Mail earlier this month, referring to concerns about the existential risks of AI. “That narrative has changed as well, now that we’re not all dead. AI will be an augmentation and not a replacement of the work force.”