Ontario’s clinician survey suggests the technology is already in use; oversight should distinguish paperwork tools from decisions that can affect care.
The Canadian
Editorial Board
October 17, 2026
3 min read
The useful question about artificial intelligence in medicine is no longer whether it will arrive. For many Ontario practices, it is already present. An OntarioMD survey described by Canadian Healthcare Technology on October 14 found that 67 per cent of respondents used an AI tool in their practices. That is a reason to make oversight more concrete—not to accept every product’s claims or pretend a blanket rejection will settle the issue.
The survey included more than 1,600 physicians, nurse practitioners and clinic managers and was conducted in March 2026. Its scope is important: these are respondents from Ontario’s community-based care sector, not a representative count of every Canadian clinician. Nor do reported adoption rates prove that a tool improves health outcomes. They tell us about use and perceived needs, which are necessary starting points but not substitutes for evidence of safety and effectiveness.
The pattern of adoption argues against treating all clinical AI as one category. Fifty-eight per cent of respondents reported using AI scribes, compared with 18 per cent using clinical decision-support tools and seven per cent using inbox assistants. Drafting a visit note and helping inform a clinical decision create different risks, even when both products are sold under the same label. Assessment should follow the actual task and potential harm, rather than the excitement surrounding the technology.
OntarioMD’s account identifies concerns about privacy, liability, regulation, overreliance and the loss of the human element in care. Our view is that these concerns should become enforceable expectations: clarity about consent, where information is stored, who reviews an output and who remains accountable when it is wrong. Clinicians also need usable, trustworthy evaluations. Requiring a busy practice to independently investigate every vendor’s promises is a poor substitute for shared standards and transparent scrutiny.
Canada should support carefully governed adoption where evidence justifies it, while refusing to let administrative convenience become a shortcut around patient protection. A note drafted faster is valuable only if it is accurate, appropriately handled and checked before it becomes part of care. The survey shows appetite for tools and for guidance at the same time. Policy should respect both. The goal is not an AI-branded health system; it is a health system in which any new tool earns trust through demonstrable benefits and clear responsibility.
The Canadian · October 28, 2026
The Canadian · October 23, 2026
The Canadian · October 17, 2026