Kiwis increasingly turn to AI for health advice, but trust remains low

FEATURE - eHealthNews editor Rebecca McBeth

One in five New Zealanders now turns to artificial intelligence platforms like ChatGPT, Gemini, or Claude as their first port of call for health information, according to Stuff's inaugural Health of the Nation study.

The figure rises to 25 percent for those aged 18-44, but trust in AI-generated health advice remains low at just 28 percent.

The study of 1,480 New Zealanders reveals 63 percent of Kiwis turning to Google or similar search engines for health-related information. This is higher than those who go to an actual doctor (56 percent) or health websites hosted by organisations like the Ministry of Health or WebMD (35 percent).

Tania Moerenhout, a GP and senior lecturer at Otago University’s Bioethics Department, says the findings reflect broader international trends. 

"I am not surprised that people turn to it for health information, you see AI now entering all aspects of life" she says. 

Moerenhout says that AI has become increasingly difficult to avoid, as these days if you run a search on Google, the first thing that pops up is the AI response.

However, the low trust levels may indicate a “healthy scepticism”.

"I think that could be a good sign that people are not just taking it at face value, but really questioning it," she says. 

"People are aware that it comes with the risk of hallucinations: generative AI can tell you things that are wrong."

The human touch

Despite the digital shift, the survey shows medical professionals maintain overwhelming trust among New Zealanders. Doctors come out on top with 89 percent trusting them for health information, followed by nurses at 86 percent and pharmacists at 83 percent. Search engines achieve 52 percent trust, with social media at just 18 percent.

Chris Paton, associate professor at the Liggins Institute and the Department of Epidemiology and Biostatistics at Auckland University, says there is growing use of consumer AI chatbots by people looking for information about health, rather than directly visiting websites that have reliable information or talking with a healthcare professional.

"There are still considerable concerns with privacy, security, and inaccurate information produced by consumer AI chatbots," he says. 

"The low level of trust in AI shown by the survey demonstrates that most people have already experienced AI making errors so are rightly cautious."

Rosie Dobson, associate professor at Auckland University’s School of Population Health, says the results are not surprising and reflect what researchers are seeing in their work. 

“People are trying AI because it is accessible, but when it comes to their healthcare, they still want a human to be the one responsible for their care,” she says,

“Trust in health services and health professionals directly impacts patient engagement with health services and patient health outcomes. Current low levels of trust in AI further highlight the need for a careful approach to the use of AI in healthcare to ensure patient trust is not harmed, and the benefits of AI in healthcare can be realised.”

Enter Dr ChatGPT

Moerenhout says ‘Dr ChatGPT’ has already entered the clinic room. She recently had a very young patient who had clearly been interacting with an AI platform and came in with certain ideas about what was wrong.

She says this experience is similar to the earlier emergence of ‘Dr Google,’ but with additional complexities in terms of hallucinations and some people going down rabbit holes.

"The sycophancy problem is something we encounter as AI wants to please the user and so it is not going to tell you that you are wrong usually,” Moerenhout explains.

She says that for healthcare providers, these developments create both opportunities and challenges and she welcomes open discussion about AI-generated information.

"If you have got a shared decision making process going with your patient and a good relationship with your patient, I think you can have a conversation about that,” she tells eHealthNews.

"But the more we have this conversation, the more time it takes and we already work in a health system, especially in primary care, that is really time poor."

Current and future state

Paton says things may improve in the future as it is likely that soon free consumer-facing AI chatbots will be able to deliver reliable health information for the public which will help people find health information more easily.

"In the meantime, I would recommend that people continue to use the existing verifiable websites from the government and healthcare providers when trying to find information about healthcare conditions," he says. 

"If their concern is serious or urgent, then they should phone 111, Healthline, or book an appointment with a healthcare professional rather than using an AI chatbot."

If you have any questions re the above feature article, please contact the editor Rebecca McBeth.

Read more FEATURES

Return to  eHealthNews.nz home page