Canadians embrace AI for banking questions, but human advice still wins on the decisions that matter

Artificial intelligence is becoming increasingly embedded in Canadian financial services, helping consumers understand their spending, navigate banking products and obtain answers without waiting for a customer service representative. Yet there is a boundary Canadians appear reluctant to cross.

Recent research suggests consumers are prepared to accept AI when the financial stakes are relatively low, but when decisions concern investments, retirement or other important questions about their money, confidence shifts decisively towards people. This creates a dilemma for Canadian banks. The technology promises faster service and greater personalisation, but deploying increasingly sophisticated AI agents does not necessarily mean customers will trust them with increasingly important decisions.

Comparable findings from the US underline the issue. According to The State of AI Adoption in America, research supplied by Channel V Media and based on a Prosper Insights & Analytics survey of 7,675 US adults, 82.6 per cent of Americans prefer dealing with a live person rather than an AI bot for banking.

The supplied research indicates that the preference persists among people already comfortable with AI. Some 76.8 per cent of executives surveyed still preferred a human for banking, while 74.3 per cent of Millennials did so. For Canada, independently reported evidence points towards a similar underlying phenomenon, although with a notable distinction: Canadians appear relatively enthusiastic about AI for simple banking interactions.

AI has a place in Canadian banking

Evidence from the 2026 TD AI Insights Report, conducted by Ipsos among more than 2,500 Canadian adults, indicates that 55 per cent of Canadians are comfortable with financial institutions using AI to track spending, while 53 per cent accept its use for calculating credit scores. Half are comfortable using AI-powered tools to assist with budgeting.

For quick enquiries, acceptance rises further. The research found 59 per cent preferred AI involvement when obtaining quick answers about fees or banking products, while 55 per cent favoured it for balance checks and routine transactions and 52 per cent for password resets and account-access problems.  This makes an important distinction.

For many consumers, the question is no longer whether they will interact with AI while banking. Instead, it is what they are willing to let AI do. Checking a balance is fundamentally different from determining whether somebody can afford a mortgage. Explaining what an investment product does is different from advising someone where to put their retirement savings. The Canadian research suggests consumers recognise this difference.

When the stakes rise, humans return

Despite acceptance of relatively straightforward automation, the TD research found that 71 per cent of Canadians have greater confidence in human intelligence than AI when dealing with financial matters.

Previous reporting on the study identified inaccurate information, cited by 61 per cent, privacy and security risks at 55 per cent and lack of accountability at 54 per cent among the principal concerns around AI.  This suggests the obstacle to greater adoption is not necessarily resistance to technology itself.

Instead, consumers appear to be asking a very traditional question: who is responsible if this goes wrong? An automated assistant which incorrectly explains a bank fee may cause irritation. An AI system giving erroneous information relating to retirement income, investment risk or borrowing potentially carries far greater consequences.

That distinction may become particularly important as generative AI develops from chatbots that respond to questions into AI agents capable of undertaking more complex tasks.

Canadian banks are themselves exploring this direction. Reporting from Digital Journal on TD’s approach to agentic AI, for example, found the bank emphasising governance and retaining human accountability as it develops the technology. TD executives described humans as remaining responsible for decisions and their consequences.

Privacy adds another dimension. The American research supplied by Channel V Media found 60.5 per cent of respondents were very or extremely concerned about AI and privacy. Canadian evidence similarly identifies privacy and security as significant obstacles to confidence in AI-powered financial services. In the TD research, privacy and security risks were among the leading concerns reported by consumers.

This matters because financial conversations can involve exceptionally sensitive personal information: income, debts, spending patterns, savings and future financial plans. The more personalised an AI banking service becomes, the greater the potential importance of explaining what information the system uses, what happens to those data and where responsibility rests for its outputs. AI systems do not need to be rejected for privacy to matter. Rather, customers need sufficient confidence to disclose the information required for the technology to become genuinely useful.

Trust may be the real banking innovation

The findings consequently suggest that the most successful application of AI in Canadian banking may not involve removing people from the process altogether. A more compelling model is likely to combine machine speed with meaningful human oversight. There is evidence for this. Some 59 per cent of Canadians said they were comfortable with banks using AI provided meaningful human oversight exists. Data protection, accountability for mistakes and human oversight were also identified as measures that could strengthen confidence.

For banks, this shifts the challenge from simply developing more capable AI towards demonstrating trustworthy AI. A sophisticated chatbot may answer questions rapidly, but consumers also need to understand when a person can intervene, how important decisions can be challenged and who ultimately takes responsibility.

These considerations are becoming more important as adoption spreads beyond banking. Canadian businesses more broadly are increasing their use of artificial intelligence, with recent reporting showing that 12.2 per cent used AI to produce goods or deliver services in 2025, compared with 6.1 per cent in 2024.