The capabilities of machine learning quickly became more obvious and mainstream in 2023 as powerful language model ChatGPT became widely used and Google’s Bard launched in January.
It sparked a new wave of AI understanding and more deliberate usage, given we were already relying on machine learning through our conversations with Siri, Google Assistant and Alexa, facial recognition and fingerprint scanning technologies – to name just a few.
Also in 2023, US Walmart stores introduced robots to scan shelves for inventory, Ford introduced it to optimise factory systems and detect failures, and Air India became the first airline to introduce a robot virtual agent, Maharaja, to answer more than 6000 customer questions across four languages.
Pardon the pun, but AI is literally taking off…
So how quickly is machine learning evolving and what part will it play by 2025?
Dr Tiongliang Liu from the University of Sydney is a machine learning specialist. He is also the Director of the Sydney Artificial Intelligence Centre. He agrees it’s been a big year for machine learning.
“There’s a new wave of AI, and especially the development of foundation models, they have very good performance,” he said.
“For many tasks, they have comparable performance with human beings – that’s why they are very influential and they can be used for many sectors to improve our lives. This is really an exciting moment.”
Just as humans draw on past learnings and experiences to solve problems, machines could use the accumulation of learnings, patterns of data and experience to do the same.
“The philosophy of AI is to analyse the data, find the patterns by using the machine learning algorithms that can be fast, very efficient and very accurate, so it can save a lot of human labour,” Dr Liu said.
Dr Liu said AI was being used in many sectors, including health, education, sustainable energy development and marketing.
But areas of machine learning that Dr Liu expects will improve by 2025 include weather forecasting and predictions, agricultural and environmental applications and the accuracy of medical imaging-led diagnosis.
“With AI technology we can detect which decisions are more reliable and which decisions may be incorrect and then we can invite more experts to investigate those not very reliable cases and help them to further improvement … this actually would be very helpful and could be used to save lives for very complicated cases,” he said.
Australia’s part in a US-centric industry
Dr Liu said there were significant improvements in creating digital people, which were now looking very natural. That’s something ASX-listed company Unith (ASX:UNT) for example, has been developing.
Other ASX-listed plays are also building their strategies on machine learning, tapping into emerging technologies. Examples include Appen (ASX:APX) which provides AI training data and annotation services and Dubber Corporation (ASX:DUB) which offers AI-powered speech recognition and capturing technology.
Brainchip Holdings (ASX:BRN) claims to be developing spiking neutral processors for AI applications, Volpara Health Technologies (ASX:VHT) is using AI for screening for breast cancer and Telix Pharmaceuticals (ASX:TLX) is using AI to progress drug discovery and developments.
“Australia has a very good history of developing machine learning and AI, we have many very outstanding researchers who’re now, for example, overseas,” Dr Liu said.
“I know there’s a lot of discussions about how the Government, how many different sectors can be involved with developing AI.
“AI is very hot and now many researchers are working on this.”
AI’s developing quickly thanks to the experts committed to the cause, along with hardware improvements.
Concerns machines could overtake humans
AI is a tool that could be developed and used for good and bad outcomes.
“This (AI) is a tool which could be used for something negative, we should be very careful about this,” he said.
He also warned that decisions made by AI were not always correct, even though there was a high probability of accuracy.
Other risks exist too, especially around privacy and discrimination.
“If we do more research and put more regulations, I believe those concerns can be very well addressed,” he said.
“AI can also greatly improve the quality of human life (…) combined with human jobs we can do better things in the future.”