Standard Chartered Bank’s chief data officer is looking to apply artificial intelligence to grow new areas of business, not just to streamline operations.
“Artificial intelligence’s value in financial services is to expand the safe space in which we can operate,” said Shameek Kundu, Singapore-based CDO and chief information officer for enterprise functions.
“It leads to new customer segments or markets where traditional ways of assessing risk and managing aren’t well developed,” he added. “It’s about being able to do stuff you couldn’t do before.”
For example, adopting chatbots to cut back on the costs of call centers is fine, but that by itself may not compel the bank to adopt such technology. “But if it serves a new market segment, that’s more interesting,” Kundu said.
A.I. pilots
StanChart is working on a variety of pilots for A.I., including in credit models, fraud detection, preventing financial crime, and detecting cybersecurity threats.
Among those applications, most are to help the bank mitigate risks, but Kundu thinks the next step will be more proactive. “A lot of the work today is defensive in nature, but we can do more with clients.”
Artificial intelligence’s value is to expand the safe space in which we can operate
Shameek Kundu, Standard Chartered
However, the bank has been deliberately measured in adopting some of the edgier uses of A.I., such as incorporating alternative data sets to risk models. “We have to be more conservative than fintechs,” Kundu said. “We need to ask ourselves if there’s customer consent for us to use the data, and if the outcome is ethical.”
Although this slows A.I. adoption, Kundu argues it takes a big bank to properly scale the tech. “A.I. needs the right guardrails,” he said.
Data portability
The industry faces growing pressure for open banking, that is, sharing customer data with third parties such as fintechs via APIs, which is another driver of adopting models for machine learning, natural language processing, and other subsets of A.I.
But Kundu says the bank prefers to think in terms of data portability, rather than “open banking”. This is particularly relevant in Singapore, which last year introduced a regulatory framework for moving data among entities with customer consent. It creates opportunities for banks, not just for fintechs.
“It’s one thing for us to use data from our customer directly or even from another financial institutions,” Kundu said, “and something else to use a customer’s data from a telco, for instance – now that becomes interesting.”
Explainability of A.I. decisions is key
Shameek Kundu, Standard Chartered
Kundu says banks are well positioned to benefit from the sharing of data because of their governance and controls.
But actual open-banking services remain works in progress, in part because, first of all, such arrangements need a commercial driver. The bank has already announced it is working with Singapore fintech Silent Eight to use A.I. in compliance to fight crime.
More to come
Kundu says this is the type of partnership in which A.I. and machine learning makes a difference.
“Some ideas are conceptual and some are still being developed,” he said, although he says StanChart is close to deploying at least one credit-lending service with a third party tech company, which he declined to name.
Another project in the works, with a California tech firm, is to code for ethics in A.I., instead of rely on human judgment. “Explainability of A.I. decisions is key,” Kundu said. “Humans alone can’t govern A.I.”
A final experiment involves the supply chains of large corporate clients. StanChart is focusing on using payments data to extend credit limits or otherwise service suppliers or distributors. The bank doesn’t intend to automate loans but use data to expand its ability to provide financial services to those corporate ecosystems.
