A new poll on AI and health care suggests that while some Canadians may be turning to chatbots for medical advice, they’re far less comfortable with the technology being used to make medical decisions or replace a visit with a doctor.

The Liaison Strategies poll of just over 1,500 Canadians in May found that 46 per cent of respondents had asked an AI chatbot for medical advice in the past year.

But 68 per cent of respondents said they would rather wait two weeks to see a real doctor than get an AI-generated diagnosis immediately, and only 13 per cent said they were comfortable with AI diagnosing and prescribing medication without a doctor.

The poll also found that respondents were worried about how artificial intelligence would impact the health-care system and about potential privacy breaches.

Asked if they believe AI will make Canadian health care better, worse or have no impact in 10 years, 40 per cent said they believed it will make it worse, while 30 per cent said they believed it would improve health care.

Seventy-eight per cent of respondents said they were worried AI will make medical visits feel cold and impersonal, while 83 per cent said they were concerned private tech companies will profit too much from the public health system.

On the issue of privacy, 52 per cent said they were very concerned that hospital AI will make records more vulnerable to hackers, and 47 per cent said they opposed sharing anonymous records with researchers to improve AI tools.

WATCH | More Canadians turning to AI chatbots for health advice:

Are AI chatbots giving good medical advice to Canadians?

More Canadians are turning to artificial intelligence with questions about their health, but research shows the answers they’re getting could be incorrect or misleading. As CBC’s Jo Horwood explains, while some AI tools are being explored for use in the health industry, researchers are hesitant to put them in the hands of consumers.

Resistance to AI in health care

The poll’s findings indicate there is a lot of resistance to artificial intelligence being used in the health-care sector, said David Valentin, a principal with Liaison Strategies, a Toronto-based national public opinion research firm.

Valentin said he was struck by how many respondents who indicated they had poor access to care would rather wait for a human doctor than get an AI diagnosis.

“I think where the real worry is is that these tools might make it more difficult to see a doctor because rather than being seen by a doctor, you might just be screened by AI,” he said.

Valentin said he hopes policymakers will see these findings as an indication that they need to be more transparent about how AI is being used in health care and why. 

“And we need to do more explaining about what measures we’re taking to safeguard people, to safeguard people’s personal information and how that is taking place.”

Dr. Amol Verma, a clinician scientist at St. Michael’s Hospital in Toronto, said he sees opportunities for AI to be used for some administrative duties like scheduling staff or deliveries that would make the health-care system more efficient.

“Right now, a lot of those processes are very manual, and we could reasonably use AI with some humans helping to improve the efficiency of the delivery of our health-care system — and that’s an extraordinary opportunity today,” he said.

WATCH | Study on AI in emergency rooms:

AI shows promise in emergency room diagnosis

New research shows that not only can AI reason like human doctors when making correct diagnoses, but it outperformed them in five studies. But experts say human judgment is still key to ensuring safety and patient outcomes.

The latest science shows AI is not reliable enough to replace human judgment, said Verma, who is also a professor of AI in medicine at the University of Toronto.

“But could it help? Could it complement that, bringing information and intelligence to the surface?” he said. “That’s what we’re trying to do in the health-care system, but it’s going to take some real work.”

Verma said the poll results show Canadians are treating AI “with a healthy degree of caution.”

“We want to see the proof points first, right? And we do value that human relationship, that human connection,” he said. “And so I think when we think … how do we incorporate these new technologies, we should take that into account.”