While there are many unknowns regarding the scope and potential of AI, there are at least two benefits have become crystal clear – that AI is well suited to solving complex challenges, and that it works best when it has access to large amounts of data.

Both truths bode well for Australia’s financial services sector, where the use cases for AI promise to resolve or automate challenges such as creating personalised service offerings, reducing fraud, and managing compliance obligations.

Seemingly all of Australia’s most prominent financial services organisations have made significant announcements regarding AI in the past 12 months, but some have been at it longer than others.

AI lives here

For the Commonwealth Bank, its AI journey stretches back close to a decade, during which time its chief AI officer and data scientist Dan Jermyn said it has been working hard to ensure the entire organisation can safely leverage AI.

“There are things that two years ago only our very best AI experts could achieve that now we have safely democratised to absolutely everybody in the bank, to be able to use, experiment with, and where it makes sense within our very robust risk management framework, actually deploy,” Jermyn said.

This work has included having large swathes of the workforce attend AI training sessions, from the CEO down, with more than 22,000 staff having viewed content from the bank’s AI For All learning program.

His team is also directly benefitting from this rapid adoption of AI in the technology group itself, with 30 percent of AI-generated lines of code now being accepted.

“That means they have more time to generate great new productivity to be able to spend more time thinking about what we can come up with next,” Jermyn said.

The bank’s decision to apply AI to its customer engagement engine early in its journey has ensured that AI has touched every single channel in the bank, but also ensured that fundamental requirements such as risk processes and governance, tooling, and training were considered from the beginning.

This has enabled CommBank to safely accelerate its efforts, such that the number of machine learning models it has in production today has exceeded 2000 – well up from a little over 1000 in 2022 – and enabled the bank to quickly benefit from newer developments such as generative AI, with the bank now processing more than 2 billion tokens (units of text data) each week.

“We have some very exotic things that we are working on that we think are going to be completely transformative,” Jermyn said.

“Most people who are implementing this are looking at assistance to coders or knowledge retrieval. We are doing all of that, but we are also looking at being able to serve customers with new insights and being able to provide a much more personalised and interactive experience for them.”

An example of this evolutionary capability exists within the bank’s methods for processing customer payslips, which initially used AI to manage communications and modelling, then expanded to include intelligent document processing. Now the bank is using multimodal-vision AI and foundational models to understand tasks at an even greater degree of accuracy.

“It builds on everything that we had had previously, so the guardrails and the governance processes that we have around that and our investments in these technologies are managed in a very coherent way,” Jermyn said.

Uplifting customer outcomes

Commonwealth Bank is not alone as a pioneer in AI, with examples present across the financial services industry.

The health insurer NIB Group for instance was an early adopter of AI chatbot capabilities, with the ‘nibby’ chatbot first launched in 2017 now managing around 65 percent of all inquiries.

According to NIB’s chief information officer Brendan Mills, exploration of AI’s capabilities has always sat well with the insurer’s corporate culture, which emphasises curiosity. This has translated into investments in AI literacy, culture, and education, and the supporting framework of ethics, safety, and responsibility.

In this regard Mills said NIB and other financial services may have benefitted from the strict regulatory environments within which they operate.

“It feels to me that some organisations are still trying to put those guardrails in, whereas I feel that ourselves and some of the other more heavily regulated organisations already had some of those programs in place,” Mills said.

“What we are doing is looking at how we adapt some of those programs and frameworks to make sure we are covering off all of our bases on the AI journey.”

Having these frameworks in place has enabled Mills to consider how his own teams are making use of AI in support of their activities, such as embracing copilot tools for developer efficiency and improving its cyber defence posture.

“We are a people and tech business,” Mills said.

“The opportunity is how do we do more with the same or more with less, and when I look through those opportunity domains there is quite a skew there to efficiency and how we driven greater value.”

The question of value is reflected in NIB’s exploration of how it can become a true health partner to its members by providing health and wellness services.

NIB’s AI-driven symptom checker service for instance aims to improve member experience and efficiency, particularly for international students and workers, by directing members to the right care setting, reducing pressure on hospital emergency departments. The service was accessed more than 13,000 times in the four months following its launch in February 2024, with 71 percent of users being from its international student and worker customer base.

“It’s had a massive take-up, where people are now interacting with it quite regularly to diagnose symptoms and get a recommendation of how they should proceed,” Mills said.

Making AI pay

Another financial services organisation with a long history of AI investment is the consolidated payment solution organisation Australian Payments Plus (AP+).

According to chief information officer May Lam, AP+ has been working with AI for some time as a tool to address fraud and scams, as part of its mission to make payments safe.

“Now we have a real time fraud engine that is built on AI, which is understanding the patterns and behaviour of the consume to define the scope of fraud at the point of sale,” Lam said.

"Using the AI, we can avoid the time we spend on the false positive and home in on what really matters."

The next stage of development will see that capability applied to real-time payments.

Critical to these initiatives has been the need to ensure that their data platforms are fit for purpose – a challenge that many organisations have faced on their AI journeys.

"Having the right data model and the right pipelines coming in, that is the critical learning for us,” Lam said.

"We are still on a journey, and I don't think anyone has cracked that. But combining different data points to make sense and elicit knowledge is really key."

AI also shows promise in fulfilling another of AP+s key objectives – reducing the cost of payments.

"I believe that broadly, the first use case is about making it simple internally, so driving the back office and driving the efficiencies of the customer interactions and leveraging the data to find efficiencies from a regulatory compliance perspective,” Lam said.

The long view

While the financial services sector has been one of the most bullish regarding AI investment, the full impact of the technology on the sector’s performance is still far from clear.

However, based on what he has seen so far, the Commonwealth Bank’s Jermyn said the returns and the rapid evolution of the technologies that unpin them gives him great confidence for the future.

“I’m very optimistic and excited about what AI means for our people – those in our technology teams and those serving our customers – because what we are seeing from AI is it is continually moving our people into areas where their real passion, skills, expertise, come to the fore,” Jermyn said.