Login
How AI can capture the ‘best possible version’ of different professions
The state of affairs in the artificial intelligence space across the country, as well as how AI can be used to positively impact different professions, were revealed in this panel discussion.
Lauren Croft
As part of the King & Wood Mallesons 2022 Digital Future Summit, the Data and AI for Good panel discussion explored the power of AI, visions for the role of AI in the future and the complexity of the AI journey.
Lee Hickin, who is a committee member of the NSW AI Advisory and chief technology officer for Microsoft, said that, additionally, there are numerous AI technologies that have been adopted in Australia — and some that are yet to make it to our shores.
“Something I see in Australia, specifically, us having a really strong foothold on AI-led robotics. We have a really strong investment here in Australia on the west coast, obviously around sort of mining and resources on the east coast around agriculture and agriculture in the land management.
“We often see a lot of thought around the investment in AI from a sort of education, skilling and research point of view, and these are important, but actually applying AI into fields that matter to us as a nation, economically, as well as potentially for the future of how we operate as a planet. That’s where I see a lot of investment going,” she said.
“Something else that we’re seeing quite a good strong view of here in Australia is how are we going to work with AI and robotics and the machinery that goes with it? How are we going to make it part of what we do? So that it takes us as human beings out of these environments; the dirty, the dangerous and demeaning? How do we make sure humans aren’t in those loops, but we’re not replacing humans with AI.”
To understand these technologies, it is first important to understand that AI is both philosophical and creative, as it is technological, quipped National Artificial Intelligence Centre director Stela Solar.
“I find it quite useful to refer to the three stages of AI, the three theoretical stages of AI. The very first one is narrow AI. It’s when AI can be trained to do a specific thing, and he just does that thing. So, for example, it might be really good at forecasting supply and demand. Or it might be really good at translating a language, or it might be really good at spotting anomalies on a production line. But one model can’t do what another model does. So, that is narrow AI,” she explained.
“Then the second stage is general AI. And this is the theory here is that AI is the same level of intelligence as a human. And the way that’s defined is that you can transfer knowledge from one domain into another domain. So, you start doing this lateral thinking. And then the third theoretical stage is super AI, which is where AI is more intelligent than a human. And this is the stuff we see in movies. And these are the stories that we are surrounded by. The reality is that the only thing possible today when it comes to AI technology is narrow AI.”
Narrow AI, however, can be trained, in theory, to get certain tasks right where humans may make mistakes. This notion is something that Telstra cloud, data and shared technology executive Dayle Stevens OAM said was entirely situational.
“If you’re working in medical imaging as a doctor that has founded this company, then you’re in the moment, you’ve got 7,000 images you’re looking at every moment, you’re working endless hours, and you’re tired and, and the human situation means that you can make mistakes. And so, when [the doctor took] himself out of that context into ‘I am training, I am training [the AI models] machine learning right now’, then he’s removing all of those other factors,” she explained.
“And I think that, like in that situation, that was where having the confidence in how you were training the AI was so that you’re training it for the best possible version of you in that moment with what you’re using to train that AI.”
Additionally, despite a lot of the current data coming from the past, Ms Solar said that AI technologies are paving the way for an increasingly positive future.
“Our past hasn’t been just, it has had a lot of inequity that has been captured in the data, and we don’t even realise it. And so, what we really risk is that our models end up propagating that bias into the future. But I’m really impressed by some of the AI innovations that are able to actually find the bias and counter it,” she said.
“And I’m reminded of a tool called Fair Learn. And Fair Learn can actually identify bias in data, such as gender-based bias; I think age is another one. But essentially, AI is able to notice the bias that we ourselves cannot. And then this AI toolkit can actually reverse that bias. And I think that that is tremendous to lean into, for how AI can actually help us do better as humanity.”
