Carney's ‘AI for All’ is a pro-worker plan that forgot the workers
There is a particular kind of confidence that comes from having run a central bank, and Mark Carney has it in abundance. When he stood up at the start of June to launch “AI for All”, Canada’s long-awaited national artificial-intelligence strategy, the framing was pure Carney: a productivity problem to be solved, a growth number to be hit, an adoption curve to be bent upward. More than $2bn of federal money, a promised 250,000 jobs by 2031, an extra $200bn of economic output, and a target of lifting business AI adoption from barely 12 per cent today to 60 per cent by 2034. It is a technocrat’s document, and on its own terms it is coherent.
The trouble is the label the government has chosen to put on it. Carney’s AI minister, Evan Solomon, called the plan “obviously a pro-worker plan”. That word — obviously — is doing an enormous amount of work.
Read the strategy closely and what you find is a supply-side adoption push: money for training, money for start-ups, money to coax Canadian firms into buying the technology faster. The “pro-worker” case rests almost entirely on jobs the government projects AI will create, not on any binding commitment to protect the workers it displaces along the way. Asked about those losses, Solomon reached for the reassurance that the Bank of Canada’s deputy governor has seen no evidence yet of widespread AI-driven unemployment — before conceding, in the same breath, that “that doesn’t mean this won’t happen”.
That is a remarkable thing for a minister to admit while insisting the plan is obviously pro-worker. It concedes the central risk and then declines to legislate against it.
Canada’s labour movement noticed immediately. The Canadian Union of Public Employees accused the government of “putting the profits of Big Tech billionaires ahead of workers and the public by soft-pedalling protections against the risks of AI”. Bea Bruske, president of the Canadian Labour Congress, said unions were “united in calling for stronger AI laws, independent oversight, protections against surveillance and discrimination and a greater role for unions in shaping how AI is used”.
Their asks are concrete, and they are revealing precisely because the strategy does not contain them: a legal duty on employers to retrain workers whose jobs are automated away, and an obligation to inform unions before AI systems are rolled into a workplace, not after. These are the unglamorous mechanisms of a genuine just transition — the difference between telling people they will be looked after and writing down what “looked after” legally means.
This is the gap that will define the strategy politically. The NDP’s Don Davies put it bluntly, warning that the plan “leaves Canadians dangerously exposed” and calling instead for a “humans-first AI strategy”. He is staking out the obvious opposition ground: a Liberal government deploying the vocabulary of social democracy — pro-worker, for all — while quietly declining its instruments.
It is worth being fair to Carney. Moving adoption from 12 to 60 per cent is not a trivial ambition, and a country that automates too slowly can hollow out its workers’ wages just as surely as one that automates recklessly. The productivity problem he is describing is real. But the Prime Minister has spent his political capital convincing Canadians that he is the safe pair of hands who understands markets and people. A strategy that treats AI adoption as an end in itself, and worker protection as something the Bank of Canada will let him know about later, tests that promise.
If the job losses Solomon admits “may happen” do happen — in call centres, in back offices, in the administrative middle of the Canadian economy — there will be no upstream to blame. Carney will own them, because he wrote the plan, chose the label, and decided what to leave out. “AI for All” is a confident bet on the upside. The question the unions are asking, and the one the government has not answered, is who carries the downside.