Artificial intelligence may have blown up with the launch of ChatGPT but for years, scientists have been pondering how the technology might improve cancer care.

It's already being used in some hospitals around the world but experts say patients shouldn't be concerned about it replacing medical processionals.

The biggest opportunity, they say, is analysing information health staff already collect.

"AI is coming like a tsunami", said David Speakman, the outgoing chief medical officer of Melbourne's Peter MacCallum Cancer Centre.

He tips that within three years, AI tools will be used to monitor vital signs like blood pressure and oxygen levels to predict a patient's decline before they show signs of trouble.

"There's no doubt that it can do it much faster than currently humans can," Dr Speakman said.

"If you said to me, 'what's the biggest thing AI can do in medicine in the near future', it will be that type of stuff because the ability to do that will change the game for so many patients across the board so quickly."

Trials are underway to see if AI programs can work out what type of cancer a patient has and where it would spread based off a biopsy on a slide but that's still a while away.

Knowing if certain treatments would be successful is the ultimate goal.

From 2024, radiologist Helen Frazer will lead a trial using AI to analyse mammograms.

As clinical director of BreastScreen Victoria and St Vincent's BreastScreen, she knows many patients are stressed by the experience and is determined to change that.

"We call back many women to assessment each year with an indication for cancer that are subsequently determined not to have a cancer," Dr Frazer explained.

"In the interim they have had extra imaging, they've had anxiety, sometimes they've had a needle test."

Interpreting scans is complex. Each is examined by at least two doctors which takes about two weeks.

The trial results won't be known for about five years but if successful, it's hoped AI could take on the role of at least one doctor in screening.

"If we can actually reduce those repetitive, more mundane tasks and leverage the AI capability to do that, it's going to enable us to spend more time with patients, more time reading those more complex mammograms," Dr Frazer said.

The next step is predicting whether a patient may develop cancer between routine screenings.

"If we can personalise screening to detect cancer earlier before it spreads, we move closer towards breast cancer being a curable disease," she said.

RMIT University's dean of computing Karin Verspoor says many people think of AI as robots or intelligent machines but it's more like simulating intelligence, and the best use will be finding patterns in big data sets.

"I'm spending a lot of time just explaining AI because there's a lot of hype out there and a lot of both mystical thinking and concern," Dr Verspoor said at the Victorian Comprehensive Cancer Centre Alliance research conference.

She said language processing programs similar to ChatGPT are far less mature than image processing but that could change.

Within ten years, she expects AI devices that monitor and take samples from patients will be available in remote locations and build on telehealth systems, reducing travel time and breaking down barriers to healthcare.

"Now that there's a lot of attention on these technologies, we're going to see more acceleration of their use ... because people can see the value," she said.

For Dr Speakman, it's important to reassure patients that nothing would be introduced without stringent testing and it could never replace the role of a clinician.

Nor could it replicate their determination to find breakthroughs.

"I don't see yet that AI is going to have that same motivation ... the motivation of clinicians and patients to find solutions to problems is massive," he said.

"If we lost that, we'll lose out on the potential of AI."