Canada & Claude
Anthropic published a look at how Canada uses Claude this week, and the finding underneath it is interesting.
Adoption of any tool is gated by a few things stacking in sequence: whether people have access to it, whether they can afford the switching cost, and whether the tool actually fits the shape of the work in front of them.
One would expect, globally, across countries at wildly different income levels, the first two gates, access and affordability,, are still binding. A country's income is a decent proxy for whether those gates are open at all. So globally, income predicts adoption, because income is still the thing standing in the way.
But look inside one country. for example, Canada.
Every province in Canada has functioning internet, an educated workforce, and disposable income by global standards. The access gate and the affordability gate are already open everywhere. So they stop being the variable that discriminates anything, and the thing that's still binding, the new rate-limiting step, is whether your daily work even looks like something a language model is good at.
That friends is not wealth. That's task shape.
And the data shows just that: within Canada, income goes flat as a predictor, and industrial composition takes over. British Columbia and Ontario lead not because they're richer than Newfoundland and Labrador, but because they have more professional, scientific, and technical services work; the kind of work these models are already good at.
The gate that used to matter has closed behind almost everyone in a rich country. The one that matters now is whether your work was ever shaped like the kind of thinking these models do.
A language model works on symbols in, symbols out.
So the real question was never about income, or even industry; it's whether the actual centre of your job, not just its paperwork, is already made of language. Where it is, adoption follows. Where the real work happens with your hands, in a room, on a body, it doesn't, no matter how modern the industry looks on paper.
Full piece here.
K Baksh