The Questions We Ask About AI Reveal What Matters to Us
Countries everywhere are trying to decide what to do about artificial intelligence. That does not mean they are all having the same conversation.
In Canada, the discussion is becoming more serious. Parliament is asking about rights, privacy, commercialization, culture, and sovereignty. Provinces are building different kinds of responses: some focused on public administration, some on governance, some on infrastructure and economic growth. Regulators are asking whether trust, transparency, and accountability are strong enough for what comes next.
These are good questions. But they are not the only ones.
Most AI governance still assumes that the systems we need to manage are the ones doing things to us: assessing us, deciding about us, processing our data, shaping our information environment, affecting our jobs, influencing public life, and perhaps even shaping how we understand reality. That is where most of the current framework is pointed.
What it is less prepared for is a different category of system: AI that we choose to let into our personal lives.
These are systems people return to by choice. They are designed to listen, remember, respond, and adapt over time. Companion tools. Emotional support interfaces. Relationship-like products that are technically sophisticated but experientially soft — designed to feel responsive, familiar, and increasingly human-adjacent.
This is not a future problem. It is already a market.
And it raises a different set of questions. We know that AI systems can draw people into highly personal territory very quickly. We know that regulators are concerned about how personal information is collected, used, and explained. We also know that companies are sometimes making difficult judgment calls internally — what to flag, what to escalate, what to disclose, and what to keep inside the firm.
That is not just a technical issue. It is a governance issue. And it becomes more serious when the product is not merely useful, but emotionally sticky.
The hardest questions here are not about model architecture. They are about people.
What happens when connection becomes a designed experience? What obligations does a service have when it knows a user is vulnerable? Who is responsible when harm does not come from a malfunction, but from a system working as intended? How should law treat data that is not simply personal, but intimate — grief, loneliness, anxiety, attachment? And what does consent really mean when the system is built to feel like it cares?
These questions are still slightly outside the centre of today’s AI debate. Not completely outside — but still at the edges.
That matters, because the sovereignty question here is different too. It is not only about compute, infrastructure, or who owns the data centres. It is also about who shapes the experience of human connection, and on whose terms. It is about whether a private company’s internal judgment about when to act, when to disclose, and when to stay silent is enough governance for systems designed to become part of people’s emotional lives.
Canada is asking serious questions about AI. We just need to make sure we are asking the full set.