​​Cautious Canadians are warming up to AI for everyday banking, survey reveals

Canadians are open to embracing artificial intelligence (AI) as a tool to enhance everyday banking, but they still want the human touch when it comes to making important decisions. That’s one of the findings of a recent TD survey tracking attitudes to AI in banking.  

The survey reveals that 59 per cent of Canadian respondents are comfortable with banks using AI, provided it’s guided by meaningful human oversight. 

“The survey findings line up almost exactly with how we think about AI at TD,” says Luke Gee, senior vice president, chief analytics and AI officer at TD. “Canadians want speed and simplicity for everyday tasks, like checking a balance, resetting a password and getting a quick answer about a product. But they also want a person involved when the conversation turns to big financial decisions.” 

Results indicate Canadians are comfortable using AI for routine financial tasks, such as tracking spending (55 per cent), reviewing credit scores (53 per cent) and budgeting (50 per cent). But they want to work with a human being on higher-stakes decisions, such as financial planning (55 per cent), product approvals (55 per cent) and retirement planning (53 per cent). 

“That’s a sensible distinction, and it speaks to trust,” says Gee. “AI can take care of the routine tasks quickly but people still want transparency, judgement and reassurance from another human when the stakes are higher. Our role is to deliver both to our clients.” 

TD has rolled out several AI solutions that enable faster and more effective services for clients, with humans still at the helm. Here are four everyday examples of what that looks like: 

For Canadians applying for a mortgage, the goal is a faster, smoother experience. TD’s first agentic AI model, launched this year, helps make that possible by handling pre-adjudication work on some mortgages and home equity lines of credit. It classifies documents, pulls out key information, calculates income, validates figures against policy, runs consent checks and flags inconsistencies before producing a summary memo for an underwriter to review. 

“That process used to take an average of 15 hours,” Gee says. “With the new model, it takes under three minutes, with more accurate results. The underwriter still makes the lending decision, and the client gets to ‘yes’ much faster.” 

When a client enters a branch with a question, front-line colleagues can use the TD Branch Virtual Assistant, a generative AI tool designed to help them source answers from policies, procedures and internal documents. 

“A question that could have taken several minutes to resolve can now take a matter of seconds,” Gee says. “We’re providing support faster than we ever have. Our colleagues are more confident — and I think that’s being felt by our clients too.” 

A new generative AI virtual assistant inside TD contact centres helps colleagues answer client questions by providing relevant guidance for them as the call is happening. Since it went live, call hand-offs have declined by 12 per cent.  

“That’s a meaningful difference for someone who just wants to sort out a charge on their statement or confirm a payment date,” Gee says. “The agent stays focused on the client as the information work happens in the background.” 

For someone with an insurance question, the TD Insurance (TDI) Virtual Assistant offers a great place to start. Users can access the TDI Virtual Assistant on the TD Insurance website and get answers to home, auto or small business insurance questions in seconds. They can do so without logging in, sharing personal information and without being a TD client.  

The TDI Virtual Assistant can only answer general insurance questions. When users have questions about their accounts, quotes or claims, the virtual assistant directs them to contact a TDI licensed professional for more information. 

TDI is the first home and auto insurer — and first among Canada’s major banks — to introduce a client-facing generative AI chatbot such as this. 

TD’s Trustworthy AI team evaluates each AI model deployed at the bank on five checkpoints before they go live: privacy, security, fairness, accountability and explainability. The team keeps monitoring each model after it’s deployed. Transparency with clients sits inside that framework. 

“Where we have client-facing virtual assistants in market, we design them so that the option to speak with a person sits alongside the digital experience, not behind it,” Gee says. “If a client wants to speak to a person, they can speak with their advisor, contact our client service team or escalate through our normal channels.” 

There is no scenario where a client is told that a decision is final because a system said so. “A person is accountable, and a person can explain.” 

TD continues to work toward a hybrid future where colleagues and AI work together, particularly on the moments that matter most to clients. That means more colleagues are equipped with better tools, more clients are getting faster answers and more time is spent on the conversations that need a human. 

“Our job is to use AI to help take the friction out of banking, give our colleagues more capacity to focus on clients and never lose sight of the fact that this is about helping real people make decisions that matter to them,” Gee says. “That’s what we mean when we say the future of AI in banking is more human.” 

For more information on how TD is using AI to make banking better for its clients, please click here. 

This story was created by Content Works, Postmedia’s commercial content division, on behalf of TD.