Canada Needs an AI ‘Moonshot’

Canada creates remarkable talent in artificial intelligence. Our universities produce world-class research. Our startups attract global customers and investment. Governments craft AI strategies and fund initiatives.

 And yet, when it comes to using AI to solve the problems Canadians care most about, progress has been limited.

The issue isn’t technology. It’s ambition.

Too often, AI policy focuses on tools rather than outcomes: new models, platforms or compute infrastructure. What’s missing is a clear answer to a more basic question: What do we want AI to accomplish for Canadians? Until we articulate our national mission, we’ll continue using AI for incremental gains. We’re missing the opportunity for transformational change.

One way to reframe the conversation is through “moonshots” – large, concrete goals that are easy to understand, difficult to achieve and significant enough that they require the mobilization of multiple sectors. The idea isn’t new. The original moonshot wasn’t about building better rockets; it was about landing a human on the moon. The technology followed the goal, not the other way around.

AI policy must work the same way.

Consider health-care wait times. Canadians consistently rank access to timely care as a top concern. Decades of reforms have produced limited results, in part because health-care systems are complex and fragmented. Deployed thoughtfully, AI could become a key piece of a system to co-ordinate patient flows, predict bottlenecks, optimize staffing and reduce administrative burden. By enhancing productivity, AI can increase the value of each hour of health-care labor.

But this won’t happen through isolated tools adopted hospital by hospital. It requires a national strategy with a well-specified goal – for example, cutting average wait times to one-tenth of their current level – and then aligning incentives, data and governance to achieve that outcome.

Or take early literacy. Reading proficiency at age eight is one of the strongest predictors of lifelong outcomes, from educational attainment to health. Canada already collects extensive data through schools, libraries and social services. AI tools can be used to identify at-risk children earlier, personalize learning supports, and determine which interventions are most effective. But this only works if the objective is explicit and shared: dramatically improving early literacy rates, not simply “using AI in education.”

Natural resource management offers another opportunity. Canada’s economy depends on agriculture, forestry, mining, and energy – all of which are increasingly vulnerable to climate volatility. AI systems that integrate satellite data, information collected by a variety of other sensors and historical records could help predict wildfires, optimize water use and reduce environmental harm from events such as oil spills. The payoff isn’t a better algorithm; it’s more resilient communities and ecosystems.

Other moonshots are equally tangible: improving the detection of national security threats or reducing chronic homelessness by better coordinating housing, health and social supports. In each case, AI is not the hero of the story. It’s only one piece of the puzzle. It’s a necessary but not sufficient part of the solution.

This distinction matters because large-scale AI adoption is not primarily a technical challenge – it’s an organizational one. The hardest work is no longer training the neural net, but rather changing processes, sharing data across institutional boundaries and redesigning how decisions are made. These changes are costly, risky and often politically uncomfortable. Without a compelling public goal, they are unlikely to happen.

Crucially, many of these moonshots will not be taken on by the private sector alone – not because they lack value, but because their returns are social. Shorter hospital wait times, for example, benefit society broadly rather than generating profits for private firms. Venture capital rewards speed, scale and the ability to keep competitors at bay; public-interest AI moonshots demand patience and coordination across institutions – and yield benefits that accrue broadly rather than to a single firm. Governments must step in where the market will not, creating the conditions in which private innovation can later flourish.

Canada’s “AI strategy” should be judged not by how widely AI is deployed, but by whether it measurably improves outcomes for Canadians at a scale commensurate with the technology’s power.

This outcome-first approach also brings into focus an increasingly urgent question: AI sovereignty. If Canada’s most important public systems come to rely on AI developed, governed and controlled elsewhere, do we risk losing not just economic value but strategic autonomy? AI sovereignty does not mean isolation or protectionism; it means ensuring that Canadians retain meaningful control over the data, infrastructure, and decision-making systems that underpin essential services. Achieving that requires public leadership – in data stewardship, procurement, standards and long-term investment – so that Canadian values (judgment!) are embedded in the systems we come to depend on.

This strategy also helps address concerns about jobs. If AI is used merely to automate parts of an already strained system, we may lose many jobs despite gains in productivity. But if we aim higher – using AI to deliver equal or better care while reducing wait times by 90 per cent, rather than simply maintaining today’s long waits with fewer staff – we’ll need more skilled people, not fewer, to design, manage and improve this better-performing system. We will have used AI to generate much higher returns on human labour. In such a system, each professional becomes far more productive – and public investment in talent yields correspondingly greater returns.

Framing AI around moonshots emphasizes productivity and impact, not automation for its own sake.

Canada is well positioned to lead this shift. We have strong public institutions and deep expertise in AI research. We now also have a federal Minister of Artificial Intelligence and Digital Innovation, who is highly energized to build for Canada and compete globally. What we need is the discipline to choose a small number of bold goals and pursue them relentlessly.

That means resisting the temptation to fund dozens of disconnected pilot projects. It means accepting that real progress will require coordination across governments, sectors and disciplines. And it means judging AI investments by their results, not their novelty.

If we want AI to matter, we must stop debating what we think the technology might do and start putting it to work with the vision and ambition Canadians need to continue to be a free and prosperous nation in this new technological age.

Ajay Agrawal is a professor of strategic management at the Rotman School of Management, the Geoffrey Taber Chair in Entrepreneurship and Innovation and founder of the Creative Destruction Lab. In this column, he draws on an AI Strategy proposal he prepared for a federal task force led by Evan Solomon, minister of Artificial Intelligence and Digital Innovation.