How AI has transformed ways of working

AI has been seen as a game changer for organisations as it has created efficiencies in processes, analysed data quickly and provided thoughtful insights.

CSIRO-owned Data 61 is the data and digital specialist wing of the country’s science agency.

Petra Kuhnert, group leader statistical machine learning group at Data 61 and first runner-up Innovator of the Year at the Women in AI APAC awards spoke to Digital Nation on how AI has transformed the way the agency works.

She said some people think of AI as a robot where everything is automated, whereas for her, AI transforms the way she looks at ideas.

“When you unpack it, it contains all these kinds of methods to help transform the way that you look at things, for me instead of a robot, it's a computer and the algorithms and the methods that we come up with,” she explained.

“The methods I'm using and other people are using, particularly as we move more into this emulation space where we're trying to come up with surrogate versions of existing models, it's helping accelerate the way we do things. It's helping us make decisions a lot sooner and a lot faster.”

Kuhnert said these methods are giving her and her team more confidence in trusting certain predictions.

“We might need to invest more effort into understanding why we can't get good predictions in this space,” she said.

Kuhnert explains how AI allows her and her team to look at possible solutions and strategies faster than any other non-AI application would.  

“Part of what we're doing is not grabbing any sort of data and pushing it through a machine learning solution, but thinking through the kinds of data we need to make the solutions more viable and more meaningful,” she said.

“You can imagine there is this idea of AI can do anything and as long as we grab any sort of data, we'll be fine.”

She said if you are grabbing lots of data about the same thing, you’re not informing anything.

“Similarly, if you have somehow grabbed data that is very much biased towards a particular thing, particularly once you get into gender or things like that. Robo-debt is another good example because it was targeting certain kinds of people and that's why they were being sent [those debt notices],” Kuhnert said.

It is thinking about the data that's being collected or designed, Kuhnert said.

“Because we have to design a lot of experiments for these slow running models to be able to replicate them using a machine learning approach,” she noted.

“But once you’ve overcome that hurdle, you can accelerate a lot of the things that we're doing. We are looking within CSIRO to accelerate our processes and understanding a lot quicker, but if it's done properly, lead to solutions a lot quicker.”

Digital Nation Australia is a proud media partner of the Women in AI APAC awards.