ANZ is exploring ways to use generative AI to answer customers’ questions faster, in preparation for changes it sees in the way banking customers will engage with mobile and web-based channels.
Head of wholesale digital Leigh Mahoney predicted an impending change to the “the way customers interact with channels – web and mobile – in the future”, based on generative AI’s trajectory.
“At the moment it’s a lot of point and click to extrapolate information,” Mahoney told the recent Microsoft AI Summit in Melbourne.
“I think in the future, channels will be a lot more conversational, and so customers will ask your service what to do – how do I make a cross border payment? How do I get this efficiency? How do I do these things rather than trying to sift through help documents and reams of information that somebody would typically go through if they were asking for help?”
Mahoney said that the answer to a question today might require reading many documents.
“Being able to summarise that quickly and get your customer to an answer faster so that you improve customer satisfaction, I think that’s really important and that’s something we’re exploring right now,” he said.
Summarisation – of extensive documents, meeting minutes or knowledge stores – is a common early use case for generative AI.
“There’s a whole heap of use cases in knowledge management, from analysing documents and providing summaries right through to [document] comparisons,” Mahoney said.
“Content authoring is also interesting but simple – I think we need to explore this a little bit more, with inspection of the output.”
Mahoney also noted the additional risk management and governance required for customer-facing use cases of the technology, such as to power the evolution of customer service channels.
“One of the things we need to be careful about is how we deploy GenAI internally and how we deploy it externally,” he said.
“How you deploy this internally will be a different approach to how you employ that externally with customers, especially when you need to be making sure you’ve got a lot of inspection across what the output is. How are you making sure that the output is what you were looking for and is actually solving problems?
“A lot of this is underpinned by strong risk management and governance frameworks.”
ANZ, like a number of other Australian banks, has been vocal in its explorations of AI technology.
Last year, its Institutional division incorporated generative AI into a business intelligence tool used by frontline bankers to stay abreast of market developments.
It is also expanding usage of GitHub Copilot to help engineers develop code, and enabling experimentation with a private ChatGPT-like interface it calls Z-GPT.
Mahoney was buoyant about generative AI’s potential, particularly compared to previous emerging technologies that have tried to make their mark in the finance sector.
“I don’t think there’s a more exciting time in our industry than at the moment than the opportunities that GenAI and its predecessor machine learning offer us,” he said.
“The advantages across our businesses I think are astounding.
“I used to say a few years ago that when blockchain and DLT [distributed ledger technology] came along that it was a solution waiting for a problem, but I think GenAI is a solution already solving problems. That’s my attitude to this.”