TD sees big role for AI-driven features
Behind-scenes advancements designed to help detect fraud, create better lending models
The biggest contributions artificial intelligence can make to financial institutions are ones customers may never see, said Toronto-Dominion Bank’s Tomi Poutanen.
While the bank released details of new AI-driven features in its mobile-banking application Thursday — such as digital nudges when a chequing-account balance is getting low, or a notice that a big bill is coming due — Poutanen, Toronto-Dominion’s chief AI officer, said his unit may be making its most important advancements behind the scenes.
One such use is combing through data to identify networks of bad actors who team up to cheat the bank’s insurance division, such as an auto-repair shop, towing company and scam artist working together on fictitious auto-insurance claims, Poutanen said. Another is using the granular data the bank has on its customers to create better models for deciding which customers are safe bets for loans, he said.
“If you get your credit decisions right and you manage fraud well, you can run a very efficient bank,” Poutanen, 48, said in an interview Wednesday. “That’s an area where AI is very applicable.”
For banks, smoothly functioning technology and the ability to prevent digital fraud have become more important during the coronavirus pandemic, which has forced much of financial institutions’ business online. Toronto-Dominion has seen more than one million customers enrol in digital banking since COVID-19 struck, and it now has more than 14 million active online and mobile customers across North America.
Poutanen co-founded Toronto-based AI firm Layer 6 Inc. and moved to TD when the bank bought the startup in January 2018. Layer 6 has kept its name within the bank and has more than tripled its staff since the takeover.
The unit is working on developing uses for its technology with every one of the bank’s business lines, including its capital-markets division, where AI can improve decision-making, he said.
“Where there is lots of customer data, and where the enterprise makes datadriven decisions, that’s where AI can help,” Poutanen said.