Canada has a severe shortage of (AI) healthcare workers

Peter Phua

17 Jan 2026 — 2 min read

I recently came upon a really well done report on the state of AI in healthcare.

It's been prepared by an organization called "Arise-AI," and is available here.

I strongly recommend the report to anyone with an interest in the subject of AI applications. There's lots of interesting pearls here, including counterintuitive findings around how much time AI scribes are actually saving:

I think that the way most people are thinking about clinician augmentation in healthcare is still quite embryonic and it's not year clear where the biggest gains will be.

Will physician efficiency be improved most via:

Improved scribes?
EMR clinician "co-pilots" or agentic assistants?
In reducing medical admin burden so that clinics can service more patients with fewer admin staff?
Or with agents that can take a preliminary medical history before the patient sees their physician/NP/etc.?

My intuition is that the biggest gains will come from clinicians delegating components of the medical history to AI agents. Anyone who has done primary care will tell you that history-taking comprises a substantial fraction of time per visit in the outpatient setting.

Let's think about outpatient primary care in Ontario as an example. We are currently short family doctors/nurse practitioners on the order of several million patients. Marginal reallocations of existing funding or even increasing medical school slots simply isn't going to meet that demand, particularly as the population ages into greater and greater levels of medical complexity.

What will meet the demand - potentially - is augmenting existing clinicians with AI tools that meaningfully increase clinician efficiency.

If the same primary care office can service, say, 20% more patients because visits are conducted that much faster, then the apparent deficit of MD's/NP's simply disappears.

This is how we're going to solve the outpatient crisis in Ontario - by using AI healthcare workers to act as force multipliers for human clinicians.

Thankfully, there are a range of companies offering agentic workflows in this space.

More than anything, the province needs to accelerate the adoption of these tools in the same way that they've encouraged the adoption of scribes.

We have a healthcare AI diffusion problem in Canada

It's universally recognized that we're currently in a healthcare crisis. Most of the solutions come down to marginal reallocation of system resources - funding a new family medicine clinic in a specific rural community, marginally slightly increasing physician training slots, programs to onboard more international medical

Voice is the universal interface

I started working in Health Tech in the early aughts, with a small tech business I started with a friend from school (AudioNotch). At the time, the thinking was that app-based tools dedicated to specific disease states were the preferred thing to build - and you saw a lot of

Large Language Models are very good at differential diagnosis

This paper made the rounds recently, and not without good reason. It turns out Large Language Models are very good at formulating a differential diagnosis. On the other hand, some research indicates that the utility of these models might be more brittle than it appears: This paper recently went viral

What is Agentic Healthcare?

Agentic Healthcare is a blog about the application of AI agents to healthcare workflows. I'm Dr. Peter Phua, a Canadian-trained physician with an interest in AI for healthcare. A lot has been written about the healthcare crisis in Canada, a system which is currently in crisis at every

Read more

We have a healthcare AI diffusion problem in Canada

Voice is the universal interface

Large Language Models are very good at differential diagnosis

What is Agentic Healthcare?