You probably used artificial intelligence before breakfast. Not the version that writes essays. The version that has been embedded in ordinary software for years without anybody putting a label on it.
Maps predicts your traffic and reroutes you with machine learning. Your phone unlocks by running your face through a neural network. Every social feed you scrolled this morning was ordered by a model deciding what holds your attention. None of that got announced. It just showed up in the update.
The number for Alberta business
Statistics Canada surveyed businesses in April and May and found that 18.4% of Alberta firms had used artificial intelligence to produce goods or deliver services in the previous twelve months. The national figure was 19.2%.
Two years earlier the national number was 6.1%. Adoption has more than tripled inside 24 months, which is a faster curve than almost any business technology of the last thirty years managed.
The other side of that same statistic deserves equal billing. More than 80% of Canadian businesses still have not put this technology into their operations at all. Both things are true. The adoption curve is steep and the base is still small, which is exactly the moment when the gap between early and late gets expensive.
The spread between industries is enormous
Information and culture came in at 42.3%. Finance and insurance at 40.4%. Primary agriculture at 4.5%.
That is close to a tenfold difference between the top and the bottom, inside the same economy, in the same year. Some of it is genuine. A model that reads documents is more obviously useful to an insurance underwriter than to a cow calf operation in August.
But some of it is a measurement problem. Ask a grain farmer whether they use artificial intelligence and most will say no. Ask whether their sprayer distinguishes weeds from crop, whether their grain dryer optimizes itself, or whether their agronomy software recommends variable rate application, and the answers change. The technology arrives inside equipment long before anyone calls it by name.
It is already in the patch
The clearest Alberta example is in oil and gas, where models have been reading seismic data and improving drilling decisions for years. Alberta’s energy industry did not adopt this because it was fashionable. It adopted it because a better read on a subsurface formation is worth millions of dollars per well, and the return was calculable before anybody committed a budget.
That is the pattern worth exporting to the rest of the provincial economy. Not enthusiasm. Arithmetic. Find the decision that is expensive to get wrong, check whether a model improves the odds, and measure the result.
The questions that deserve answers
There are fair concerns here and pretending otherwise helps nobody. Privacy, because these systems run on data about people. Jobs, because some tasks genuinely go away. Electricity, because the compute has to be powered somewhere and Alberta is increasingly where. Accountability, because when a model gets a decision wrong somebody has to own it.
Those conversations should happen and they should happen with real numbers rather than slogans on either side. What does not help is the framing that treats adoption as a decision still pending. It is not pending. It is at 18.4% and climbing, in a province that has never been last to a productivity tool in its history.
The useful Alberta conversation is narrower and more practical. Where does this actually help, where does it not, and who stays accountable when a business puts it to work on a customer. Nobody voted on the horse and buggy either, and the province still got a say in how well it handled what came next.
The technology is already inside an ordinary Tuesday. The only open question is whether Alberta businesses are using it deliberately or by accident.
Is your business in the 18.4% or the 80% that has not started yet?
