AFTF 2023: AI, the right tool for the right job

The rise of generative artificial intelligence, or AI, is a transformative opportunity for the financial services sector, according to Gerard Florian, Group Executive, Technology at ANZ. Like any tool, it has its uses – if used in the right way.

Speaking at an ANZ internal event, Florian said how the technology was applied would be critical in determining its success at an organisation like ANZ.

“Generative AI is a tool that helps us to very quicky interpret large sets of data to give us a steer,” he said. “It’s more creative and useful for answering questions like, ‘does this contract match these policies?’ Or responding to prompts like, ‘summarise changes in this regulation’.

“I would not use AI to write a regulatory compliance report, where there is no room for error. But I would use it for a trend analysis to help shape strategy, for example. Context matters.”

A recent report from Microsoft and the Tech Council of Australia suggested generative AI technology could help add between $A5 billion and $A13 billion annually to the professional and financial services sectors in Australia by 2030.

The broader economic opportunity could be between $A45 billion and $A115 billion a year in that time, the report found.

Powerful

Lisa Vasic, ANZ Managing Director, Transaction Banking, told the event programs like ChatGPT put “a powerful AI tool into everyone’s hands”, giving users value even if they knew little about it.

“The scale and pace of progress is unlike anything we’ve experienced before,” she said. “Generative artificial intelligence is the most exciting tech leap forward since the internet.”

But what makes it different from existing technology - and so valuable? According to Florian, it’s the way the machine ‘thinks’.

“Traditional computing is deterministic,” he said.  “It uses a ‘If this, then that’ approach, which works for rules-based scenarios. We don’t want things to be open to interpretation in that area.

“AI computing is probabilistic – a ‘based on this data, it is probable that’ approach. The computer can now review the data and predict an outcome. And depending on how much data you train it on, you can get a different answer.”

For banks like ANZ, this can help answer complex questions, Florian said - like if transactions are fraudulent, or where sentiment in a market is currently heading.

Risk

Speaking at the same event, ANZ Banking Services Lead Nigel Dobson said as a risk-averse organisation, the bank has been careful about how it approached generative AI - although it has allowed the popular application to be used by staff (with strict controls).

“We didn’t block access for staff because we wanted everyone to learn how it works,” he said.

Dobson said the bank had worked hard to build our awareness of what AI tools do when given access to information, ensuring nothing sensitive is used.

“Our top priority is to protect ANZ data, so all our normal data protection rules apply,” he said, noting the ANZ was working on governance, risk and regulatory responses to AI as well.

Florian said ANZ was moving carefully with generative AI and would continue to do so until the bank had more confidence in the strengths and weaknesses of the technology.

“There will need to be checks and balances for some time yet,” he said. “We have to get the answers right and manage any ambiguity with confidence – particularly when it comes to customer safety.”

Beyond the risk, the focus of AI is what it can bring to ANZ’s customers, Dobson said.

“And because we have such rich data sets, our customers expect us to be at the forefront of this, and help them make better, more informed decisions,” he said.

Responsible

Florian said ANZ was taking a responsible and open approach to generative AI – and in fact had been doing so for some time.

“We’ve been using machine learning and AI at ANZ for some time already – for risk management, in user authentication, in cyber defence, in reading market signals,” he said.

Florian said the bank had established an Artificial Intelligence Centre of Experimentation in Bengaluru, India, where it had “more than two dozen official experiments in progress”.

At ANZ, the focus on AI is improving its quality of work through three areas, Florian said. The first was augmentation, where AI can be used to assist with a task, which can be helpful for engineers or financial coaches. The second is automation, where AI is embedded in tools to perform tasks.

The final application is orchestration, which helps the bank do things differently, or organise things in a unique way.

“At ANZ, AI is making knowledge management more effective, making software engineering more efficient, building better credit models, and revolutionising customer service, eventually,” Florian said.

What happens next is anyone’s guess. New types of AI are evolving quickly, according to Florian, and new applications will continue to emerge.

“The scale and pace of development is dizzying,” he said. “Generative AI will transform virtually all industries, and the possibilities are immense.”