Just two weeks after securing its partnership with Adelaide University’s Australian Institute for Machine Learning (AIML), CommBank and the AIML built a potentially transformative “deep learning” capability expected to halve the speed for critical data processing functions within the bank.
While guarded on details of the newly patented capability, CommBank chief decision scientist Dan Jermyn revealed that the new deep learning model would process data for a range of “highly complicated” functions within the bank, including its fraud detection and customer servicing systems.
The capability would enable the bank, he said, to be “twice as fast, [and] twice as efficient with the most advanced stuff that we do”.
CBA in September announced its five-year strategic partnership with Adelaide University’s AIML, a reciprocal arrangement tasked with pursuing “foundational research” into machine learning (ML) whilst also exploring practical, real-world applications for AI/ML technologies within the finance sector and other industries.
Professor Anton Middelberg, deputy vice-chancellor and vice-president (research) at the University of Adelaide, said the collaboration would work to deliver “direct benefit to people’s lives by improving financial systems”, while AIML director Professor Simon Lucy said the Centre would “serve as a magnet for top-tier global AI talent”, with members to serve as hybrid appointments across both CBA and the AIML.
Among the key objectives of the CommBank Centre for Foundational AI include the generating of “high-impact” AI/ML research publications, the facilitation of industry-academia collaboration, and overall development of Australia’s AI capability and innovation ecosystem.
CommBank is already a recognised leader in AI/ML innovation space, recently announced as a global Top 10 bank in Evident AI’s Index for October 2024, a benchmark of AI adoption and maturity throughout the banking sector.
Jermyn, speaking at FST’s Future of Financial Services, Sydney 2024 conference on Friday, also revealed further details of CBA’s dedicated Generative AI-developed IT helpdesk capability, dubbed ‘ChatIT’, which was first unveiled in the bank’s Full Year Report in August.
The system, designed to support CBA staff with basic IT support, provides technical advice, quick troubleshooting, incident management, coding assistance, knowledge base search and a ‘vision capability’ (that is, the ability to identify and understand objects in images or videos).
Jermyn reports that the function has received a 4.5/5 user satisfaction rating among CBA staff, with 93 per cent of users reporting that its responses were “clear and understandable”.
ChatIT boasts average response times of 8.52 seconds, Jermyn confirmed. A problem that cannot be solved by ChatIT can quickly be escalated to human IT support staff (see below).
Example of a ChatIT interaction (presented by Dan Jermyn, FFS Sydney)
For a team often lumbered with simple but time-consuming requests, the real benefit of the ChatIT function has been in freeing this critical support team for more technically challenging tasks. As well, those requesting help can quickly self-resolve simple IT problems without significant delay.
“We’re able to unlock the incredible capability of our IT support people. More fundamentally, for [staff] who don’t want to be spending time trying to fix their computer, it’s giving them the time to think about, ‘How can we help customers?’.”
“[We’re not] trying to eradicate these functions. We’re creating time to free up our people to do the thing that they really came here to do – [to use] their creativity, to shape customer experiences.”
Escalating the issue (Dan Jermyn, FFS Sydney)
