A complement, not a substitute: Generative AI’s role in Canadian healthcare in 2026

Canada’s healthcare system is under sustained strain that incremental reform has failed to resolve. Workforce shortages, clinician burnout, fixed capacity, outdated public-sector technologies and persistent access and equity gaps continue to constrain care. Six million Canadians lack a regular medical home. Wait times are rising. Demographic change will intensify pressure for at least the next two decades. In this context, inaction carries significant risk.In a Canadian Standards Association (CSA) Public Policy Centre report released recently, I argue that accelerated, supervised adoption of generative AI in healthcare is both necessary and defensible. The balance of risk now favours a fast-paced but thoughtful bias toward implementation.Physicians are not waiting. The January 2026 CMA-CFIB report on administrative burden found that 28% of Canadian physicians use generative AI tools daily in practice. Ambient documentation is the leading complementary use case. Interview evidence puts primary care scribe adoption at 25% to 35%. OntarioMD and Canada Health Infoway safe-harbour lists have removed much of the procurement and privacy friction. Once scribes are in, use typically expands into pre-visit summaries, structured assessments, form completion, translation and patient explanations. Clinicians keep accountability. They also recover hours of cognitive load and after-hours documentation.Second-screen clinical decision support is following the same curve. These tools work outside the EMR, do not touch protected health information and replace the informal Google-and-UpToDate pattern with auditable, evidence-aligned retrieval. The 2025 Offcall Physicians AI Report found 67% of U.S. physicians using AI daily and nearly 90% weekly, with OpenEvidence-style decision support leading at roughly 45%. Canadian uptake is close behind.Patients are moving faster than the system that serves them. Internet health-search traffic fell roughly 30% globally between mid-2024 and mid-2025 as users shifted to LLMs. OpenAI's January 2026 AI as a Healthcare Ally report estimates more than 40 million people use ChatGPT regularly for health questions, with about 70% of those interactions happening outside clinic hours. They are filling a vacuum the system has left open.The report works through structural pressures and where gen AI helps. It sets out the complement-versus-substitute distinction underpinning safe adoption. It draws on early Canadian implementation lessons across primary care, hospitals, long-term care, pharmacy and public health. And it addresses the policy decisions that follow on regulation, data, competition and digital sovereignty. The report draws an explicit line between supervised complementary tools, which scale because they fit existing scopes of practice and liability and substitutive systems, which require governance Canada has not yet built. The paper explicitly argues that “replacing doctors” is not going to happen any time soon.The report also argues that well-intentioned attempts to slow clinician or patient adoption in the name of caution are unlikely to improve safety. They will widen the gap between what is in use and what is supervised, and leave the most exposed Canadians worse off, not safer.Physicians who lead this transition will shape it. Those who wait will inherit it.Will Falk is a senior fellow at the CD Howe Institute and an executive fellow at the Rotman School of Management. He's also a policy fellow at the CSA Public Policy Centre.

A complement, not a substitute: Generative AI’s role in Canadian healthcare in 2026

Register for free access or login.

Related Topics