A habit already formed

Putting money questions to a generative AI is no longer a fringe behaviour. A survey relayed by Les Affaires puts numbers on the trend: 26% of Canadian pre-retirees and 11% of retirees already turn to artificial intelligence to support their financial planning. The catch shows up immediately after: among those who use it, nearly two thirds (65%) say they are only "somewhat confident" in the answers they get.

Translation: people use it, but they sense they should not swallow it whole. That unease is healthy — and it deserves unpacking. Where does AI genuinely help? Where should you be wary?

26%
of Canadian pre-retirees already use AI for their finances
11%
of retirees do the same
65%
are only "somewhat confident" in the answers

What AI genuinely does well

Generative AI is excellent at explaining, simplifying and structuring. It is a patient teacher, available at 11 p.m., who will never judge you for a "stupid" question. Concretely, it shines when you ask it to:

Used this way, AI is not a decision-maker: it is an amplifier of understanding. It turns an opaque subject into something you grasp well enough to take back control of the conversation.

Where it goes off the rails

The real danger is not that the AI gets something wrong now and then — it is the moment you ask it for a definitive answer to a question that requires some context about your situation. "Should I withdraw from my RRSP now?" has no right answer in the abstract. It depends on your income, your spouse, your other accounts, your horizon, your province.

Three blind spots come up constantly:

Blind spot What happens
Concepts mixed up It can confuse close rules (RRSP vs TFSA, limits, contributions) and state an error with total assurance.
Regional rules Quebec and Canadian tax rules (QPP, OAS, provincial credits) are not always current, nor clearly separated from U.S. rules.
Over-simplification To stay clear, it smooths away nuance — and the nuance is often where your decision actually turns.
The golden rule. Use AI to understand a question, never to settle a decision that commits your money. A number out of a chatbot is not a projection: it is a plausible hypothesis, to be verified.

"Confident" does not mean "correct"

An AI always answers in the same assured tone, whether it is right or wrong. That is disorienting: in a real conversation, someone's hesitation warns us to dig further. Here the assurance is constant. Hence the need for a simple reflex: every answer with a financial impact must be cross-checked — against an official source (Retraite Québec, the CRA), a professional, or your own numbers.

That is exactly why two thirds of users stay cautious. That caution is not a flaw in the tool: it is the correct way to use it.

The AI does not know your numbers — you do

Here is the limit people forget most often. A general-purpose AI knows nothing about your portfolio: how much you actually hold in a TFSA, what return your investments delivered, how fast you are saving, when you will hit your target. It reasons on averages; your retirement turns on precise numbers — yours.

The right method combines both:

Professional judgement is not replaceable. AI lowers the cost of understanding. It does not lower the cost of error. For anything hard to reverse — drawdown, withdrawal order, when to claim the QPP — a human eye that knows your file remains irreplaceable.

The right way to use it

Three principles to keep in mind the next time you open a chatbot to talk about retirement:

AI is a formidable starting point. It is never the finish line.

The survey statistics cited come from the article "L'IA peut-elle vraiment planifier votre retraite ?", Les Affaires, published on 2 July 2026. This text is an original analysis and reproduces no third-party passage.