Michelle Alexopoulos delivered a keynote address to the Ottawa Economics Association and the Canadian Association for Business Economics Spring Policy Conference. Speaking in Ottawa, Alexopoulos framed artificial intelligence not as a passing trend but as a potential general-purpose technology (GPT), a category that historically includes the steam engine, electricity, and the internet. The address, published by the Bank of Canada, is descriptive and contextual rather than a policy directive, but its institutional weight gives it significance well beyond an academic lecture.
Alexopoulos situated AI within a 75-year arc of development, noting that recent advances have made the technology far more powerful and accessible. Her remarks covered the potential for AI to lower business costs, improve efficiencies, support higher wages, reduce consumer prices, and spur new investment. She also acknowledged the harder questions: whether this wave of automation will differ from past transformative technologies in its effect on employment, and whether the rapid rise of AI-focused equities reflects genuine value or overinvestment. The Bank of Canada, she stated plainly, cares about AI because of its potential to significantly affect productivity, economic growth, employment, and inflation, and because it carries implications for financial system stability, including the risk that AI makes sophisticated cyberattacks easier to carry out.
Why AI's GPT Classification Matters for Financial Institutions
The GPT framing Alexopoulos used is analytically important and not merely rhetorical. General-purpose technologies share a specific profile: they begin with a narrow technological core, improve dramatically over time, spread across the whole economy, and generate spillover innovations that compound their original impact.No change strictly required for factual accuracy, but for completeness: Alexopoulos traced a lineage that also included the internal combustion engine as a GPT, noting that computers began as a spillover of electricity, which was itself a spillover of the steam engine, and that AI is now a spillover of computing. Each link in that chain reshaped labor markets, capital allocation, and institutional operations over multi-decade cycles.
For financial institutions, the GPT framing implies that AI adoption is not a single product decision but a structural transition with compounding effects. Institutions that treated prior GPT waves, particularly internet banking and mobile platforms, as optional upgrades often found themselves rebuilding their technology stacks under competitive pressure years later. Alexopoulos did not prescribe a response for financial institutions, and her remarks do not constitute regulatory guidance, but the analytical frame she offered is directly applicable to how boards and management teams at financial institutions should be thinking about technology investment timelines. The Bank of Canada's willingness to dedicate a deputy governor keynote to the topic signals that AI's macroeconomic footprint is now large enough to factor into monetary policy assessment.
What it means for credit unions and their technology investment calculus
What it means for credit unions operating in the $100 million to $10 billion asset range is that the Bank of Canada's analysis adds an institutional data point to a technology investment decision that many boards have been treating as premature. The cost-reduction and efficiency arguments Alexopoulos outlined, lower business costs and improved operational efficiency, map directly onto the operational pressures that mid-size credit unions face: thin margins, rising member service expectations, and competition from both large banks and fintech platforms. The National Credit Union Administration has not yet issued specific guidance on AI adoption frameworks for federally insured credit unions, which means institutions currently lack a regulatory backstop that would force the conversation. That absence can cut both ways. It creates flexibility for early movers and leaves laggards without a compliance deadline to anchor their planning.
The cyberattack risk Alexopoulos flagged is also directly relevant. Credit unions that are expanding digital service delivery, as many in the community-focused segment are doing, face a threat environment that the Bank of Canada now formally acknowledges is worsening as AI tools become more accessible to bad actors. Institutions considering AI-powered member-facing tools should be evaluating security architecture in parallel. Technology vendor partnerships are one avenue, as illustrated by recent digital banking platform renewals in the credit union sector, though vendor selection carries its own concentration risk. For institutions in underserved markets, the efficiency case for AI tools may be even more acute, given the documented difficulty of sustaining branch economics in low-density areas, a pressure point explored in our reporting on credit unions and banking desert competition.
What we're watching
- Bank of Canada Monetary Policy Report, July 2026: The next scheduled MPR will indicate whether AI productivity assumptions have been incorporated into the Bank's formal growth forecasts, a signal of how quickly the analysis is moving from speech to model.
- NCUA Board Meeting, Q3 2026: Any staff letter or supervisory guidance on AI governance for federally insured credit unions would materially change the compliance calculus for institutions currently in early-stage evaluation.
- Q2 2026 CUSO and fintech partnership disclosures: Credit union service organization filings that show AI-related capital commitments will serve as a leading indicator of how the $1 billion to $5 billion asset band is translating strategic intent into spending.
- Equity market correction threshold in AI-sector stocks: Alexopoulos flagged concerns about overvaluation in AI-focused equities.A significant correction in AI-focused equity markets would test whether credit union investment portfolios and correspondent banking relationships carry indirect exposure.