The Commonwealth Bank has embedded data science people in almost every part of the group’s operations, hoping their domain knowledge will lead to more AI-powered optimisations.
Chief data scientist Dan Jermyn told the AI Australia by Eliiza podcast that the bank had split its AI and data science efforts - which come under a broader “decision science” effort internally - between embedded capabilities and a more centralised AI Labs team.
Eliiza is a Melbourne-based data science consultancy.
“We think about our AI and data science capability in two ways,” Jermyn told the podcast in a wide-ranging interview.
“We have a large number of people who are embedded within the business, working every day allocated to our product, channel or customer teams, who’ll really understand the nuts and bolts of those areas, and understand what our customers need from a specific type of product or channel.
“That domain knowledge allows them to be much more useful and helpful in coming up with solutions to make things better.
“Then we also have what we call our AI Labs team, and they’re really focused on research and development, making sure that as a bank we’re at the cutting edge of what’s happening in AI, with an eye to making sure that we are developing capabilities then that can actually be used within the bank and across the group.”
Jermyn said that collectively, the two paths had so far exceeded expectations with the number of use cases produced.
“Certainly we have found any number of applications of data science - so many that we’d never even thought of at the outset,” he said.
“In more or less every aspect of the bank right now, we have somebody from a data science team in there thinking about simplification, automation, and ways to do things differently and better for our customers.”
Some decision science projects at CBA are better known than others, such as its Pega-powered customer engagement engine (CEE) used to prompt staff about the most effective thing to talk to customers about next; and its benefits finder, which automatically assesses a customer’s eligibility for more than 230 benefits and rebates and then guides them through the application process.
Jermyn added that as much of CBA’s AI work, in particular, was on the “cutting edge”, the bank had fewer places externally to turn to validate some of its ideas and directions.
Hence, he said, CBA worked closely with academia, including the 3A Institute at ANU and a long-running collaboration with UTS.
“In some ways we’re probably leading Australia in a lot of our application of AI,” Jermyn said.
“That’s excellent but it also means that we’re limited in who we can look to in terms of how we should be doing things.”