OPINION: Canada is training its citizens in AI—but Google is writing the code

$5 million for the Alberta Machine Intelligence Institute

The op-ed celebrates the work of Amii, which is building a consortium of 40 postsecondary institutions to teach AI to 125,000 students. What the op-ed barely mentions—and what the press release reveals—is that Google Canada has funded this initiative to the tune of $5 million.

This isn’t philanthropy. It’s market shaping. Every student trained in Google tools, every professor who integrates Google Workspace or Gemini into their course, every librarian who uses Google’s educational resources—all become agents driving the adoption of a specific commercial ecosystem.

The University of Waterloo’s model: “learners building for learners”

Waterloo’s Futures Lab, also supported by Google, has produced prototypes such as a virtual science lab and an app for understanding mathematics. These are wonderful projects. But they are projects born within a framework defined by Google, using tools provided by Google, as part of a partnership structured by Google.

And yet, no one is asking the fundamental question: what happens when a single company simultaneously funds the research, education, tools, and public discourse surrounding a technology it sells? That’s called a conflict of interest. In any other field—pharmaceuticals, the military, energy—this level of influence over public education would cause a scandal.

Google isn’t educating Canadians out of love for the flag

The op-ed is built on a specific emotional framework. Sabrina Geremia begins by mentioning her three sisters, who are educators, and her “deep belief in the power of learning.” It’s touching. It’s also a deliberate framing that transforms a business strategy into a family story.

Here’s what the op-ed never says: Google needs millions of Canadians to use its AI tools. Every trained user is a potential customer. Every company that adopts AI through Google products generates recurring revenue—cloud licenses, Workspace subscriptions, API credits. Training isn’t the goal. Training is the funnel.

$13 million for the “AI Opportunity Fund”—an investment, not a donation

The op-ed mentions a $13 million fund for organizations like Skills for Change, the First Nations Technology Council, and the Toronto Public Library. These are admirable organizations doing essential work. But $13 million, for Google, is less than an hour’s worth of revenue. The company generated 350 billion U.S. dollars in 2024.

This isn’t a sacrifice. It’s a remarkably low customer acquisition cost.

Prompt engineering isn’t a skill—it’s a temporary interface

The op-ed emphasizes prompt engineering as a fundamental skill. However, anyone who actually works in the field knows that tomorrow’s interfaces will render prompt engineering as obsolete as the DOS command line. Models are becoming more intuitive every quarter. Teaching prompt engineering in 2026 as a lasting skill is like teaching HTML in 1998 and promising that it was the key to the digital economy.

The truly enduring skills in the age of AI are the ones the op-ed barely mentions: critical thinking, the ability to assess the reliability of a source, an understanding of algorithmic biases, and non-replicable creativity. These are skills that, by definition, do not depend on any proprietary tool.

The Difference Between “Using AI” and “Understanding AI”

Training someone to use ChatGPT or Gemini is training a consumer. Training someone to understand how these systems work—their limitations, their biases, their economic and social implications—is training a citizen. The op-ed discusses only the former. And yet, it is the latter that will determine whether Canada will be a producer or merely a consumer of this technological revolution.

Geoffrey Hinton, the Canadian godfather of AI, did not spend his career teaching prompt engineering. He laid the theoretical foundations upon which Google, OpenAI, and all the others have built their empires. The difference between these two levels of expertise is the difference between waiting tables and owning the restaurant.

An appealing model with unanswered questions

The example of the Ottawa Catholic School Board is presented as a model. Students ask the AI to explain a scientific concept “in soccer terms” or to create a quiz based on their favorite TV show. It’s engaging. It’s pedagogically creative.

But the tough questions are missing. What data are these students sharing with the AI systems they use? Who has access to their queries, their mistakes, and their interests? When a 12-year-old tells an AI that their favorite show is Stranger Things and that they’re struggling with fractions, that information has commercial value. Who owns it?

AI as a “critical thinking partner”—really?

The op-ed asserts that students are learning to view AI as a “critical thinking partner, not a magic button.” That is the stated intention. The measured reality is different. The KPMG study cited in the op-ed itself reveals that more than 70% of students use generative AI for their schoolwork. The key word is “for”—not “with.”

And yet, the op-ed treats this figure as evidence of demand rather than a warning sign. When 70% of students are using a tool to do their homework, the question isn’t “how can we give them more of it?”—it’s “what are they actually learning?”

From Hinton to Hassabis: The Canadian Brain Drain

Geoffrey Hinton conducted his research in Toronto. Demis Hassabis, co-founder of DeepMind, worked with Canadian researchers. Yoshua Bengio is still in Montreal. Canada produced the minds that made AI possible. But the companies that are monetizing this research—Google, Meta, OpenAI, Microsoft—are all American.

Geremia’s op-ed celebrates the fact that Canada is “entering the adoption phase.” But adoption is not innovation. Adopting Google’s AI is like adopting Saudi Arabia’s oil: we use it, we depend on it, but we control neither the price nor the supply.

The Real Missing Investment

Canada is investing $2.4 billion in its Pan-Canadian AI Strategy. That’s significant. But the United States, through the private sector alone, invests more than $200 billion a year. China, according to some estimates, exceeds that figure. Training AI users is necessary. Training AI creators is essential.

And yet, the op-ed never mentions basic research. Not a word about funding for university labs. Not a word about retaining researchers. Not a word about technological sovereignty. Just courses, certificates, and micro-courses. Consumer skills, not creator skills.

When “transform” Is a Euphemism for “Eliminate”

The op-ed claims that AI “transforms jobs rather than replacing them.” This is the most dangerous sentence in the text. Not because it’s false—it’s partially true—but because it glosses over the pain of the transition.

In January 2026, Shopify—a Canadian company—announced that its teams could no longer request new hires without first proving that AI couldn’t do the job. Bell Media eliminated hundreds of editorial positions. Canadian call centers are losing jobs every month to chatbots.

These people don’t need a course in prompt engineering. They need a job.

The Paradox of Training as the Sole Solution

The op-ed proposes training as the sole solution. But training without industrial policy, without social safety nets, without regulating the pace of adoption—that’s like telling people to swim faster while we turn up the current. And yet, that’s exactly what Google is doing: accelerating the adoption of AI while telling displaced workers to “get trained.”

It’s the same rhetoric as that of the tobacco industry, which funded cancer research while selling cigarettes. Except that in this case, the product isn’t harmful in and of itself—it’s transformative. What’s harmful is pretending that the transformation will be painless if everyone takes an online course.

When the Industry Defines What “Competent” Means

A century ago, mining companies defined what a miner needed to know. Fifty years ago, automakers defined the skills required of assembly-line workers. Today, tech giants define what it means to “be competent in AI.” The pattern is the same. And the result is always the same: workers trained to meet the needs of the industry, not their own.

When Google funds Amii to create “AI literacy” programs at 40 universities, what tools are taught? What ethical frameworks? What space is given to open-source alternatives? To critiques of the concentration of technological power? To teaching resistance as much as adoption?

The Microsoft Precedent in Schools

In the 1990s, Microsoft offered free licenses to schools around the world. An entire generation grew up knowing only Word, Excel, and PowerPoint. When they entered the job market, guess which ecosystem companies adopted? Microsoft’s “generosity” created a de facto monopoly that lasted two decades.

Google is following exactly the same model. Chromebooks in schools. Google Classroom for teachers. Gemini for students. And now, Google-funded training programs, taught using Google tools, at Google’s partner universities. This isn’t education. It’s cognitive colonization.

Five Measures Google Will Never Propose

First: Create a national AI training program that is independent of industry, publicly funded, and with a curriculum developed by academics and civil society representatives—not by the companies that sell the tools.

Second: Invest heavily in basic research, not just in adoption. Canada has the talent. It needs the labs, the salaries, and the conditions to retain them.

Third: demand full transparency regarding the data collected by AI tools used in schools. Every query from a 12-year-old is a data point. Who owns it? Who analyzes it? Who profits from it?

The two most urgent measures

Fourth: Establish transition safety nets—not courses, but safety nets—for workers whose jobs are eliminated as the “transformation” takes place. A transition income, genuine support, and retraining programs funded by the companies that are automating.

Fifth: regulate the pace of adoption in critical sectors. Not to stifle innovation, but to ensure that society has time to absorb the changes. Europe is doing this with the AI Act. Canada should do better, not less.

And yet, none of these measures appear in Google’s op-ed. Because none of these measures serve Google’s interests.

When Criticizing Google Seems Like a Losing Battle

Here’s the genius of this strategy: how can you criticize a company that invests in education? That funds training for Indigenous peoples? That supports public libraries? That wants “every Canadian” to benefit from AI?

That’s precisely the trap. Strategic generosity creates a moral debt that makes criticism socially costly. Criticizing Google after they’ve donated $13 million makes you look ungrateful. And Google knows it.

Philanthropy as a Political Shield

This is nothing new. Carnegie built libraries while his steel mills crushed workers. Rockefeller funded universities while Standard Oil crushed the competition. Strategic philanthropy has always been the price of impunity.

Google is investing $13 million in Canada while avoiding billions in taxes thanks to tax structures that the op-ed, of course, fails to mention. If Google paid its fair share of taxes in Canada, the government could fund ten times as many of these training programs—without Google having its logo on the curriculum.

Three sentences that sum it all up

Read with a fresh perspective, the op-ed’s message boils down to three sentences. AI is inevitable—don’t resist it. Training is the solution—not regulation. Google is your partner—trust us.

It’s a message of power disguised as a message of service. And it’s effective precisely because it contains a grain of truth. Yes, AI is transformative. Yes, training is essential. Yes, Google has useful resources. But the conclusion—that Google should be the preferred partner in this transformation—does not follow from the premises. It is surreptitiously inserted, like a Trojan horse in a discourse on education.

What Canada Deserves

Canada deserves better than a training strategy dictated by its suppliers. It deserves a sovereign public policy on AI—one that is democratically developed, tax-funded, and independent of commercial interests. It deserves researchers who stay, Canadian companies that compete on a global scale, and workers who are protected during the transition.

Above all, it deserves an honest debate about who controls the technology that will redefine its economy. And that debate cannot be led by those who sell the technology in question.

Education without sovereignty is educated servitude

Training millions of Canadians to use the tools of a handful of American companies without investing in the creation of Canadian alternatives is to educate a nation into its own dependence. It’s the classic colonial model, reimagined for the digital age: we export the raw materials (brains, data, basic research) and import the finished product (tools, platforms, training programs).

And yet, Sabrina Geremia’s op-ed will be read, shared, and applauded. Because she says what everyone wants to hear. Because she promises a bright future. Because she flies the Canadian flag over a strategy that primarily serves Mountain View, California.

Canada invented artificial intelligence. It would be tragic if it became the first country to train itself in technological submission—with a smile, a Google certificate in hand, and the pride of having been so well trained by its masters.

Signed, Jacques PJ Provost