Credit: Katerina Holmes from Pexels

Artificial intelligence is everywhere. In fact, each reader of this article could have multiple AI apps operating on the very device displaying this piece.

Despite this, many mechanisms governing AI behavior remain poorly understood, even to top AI experts. This leads to an AI race built upon costly scaling, both environmentally and financially, that is also dangerously unreliable.

Progress therefore depends not on escalating this race, but on understanding the principles underpinning AI. Mathematics lies at the heart of AI and investment in these mathematical foundations is the critical key to becoming a true global AI leader.

How AI shapes daily life

AI has rapidly become part of everyday life, not only in talking home devices and fun social media generation, but also in ways so seamless that many people don’t even notice its presence.

It provides the recommendations we see when browsing online and quietly optimizes everything from transit routes to home energy use.

Critical services rely on AI because it’s used in medical diagnosis, banking fraud detection, drug discovery, criminal sentencing, governmental services and health predictions, all areas where inaccurate outputs may have devastating consequences.

Problems, issues

Despite AI’s widespread use, serious and widely documented issues continue to showcase concerns around fairness, reliability and sustainability. Biases embedded in data and models can propagate discriminatory outcomes, from facial detection methods that perform well only on light skin tones to predictive tools that systematically disadvantage underrepresented groups.

These failures continue to be reported and range from racist outputs of ChatGPT and other chatbots to imaging tools that misidentify Barack Obama as white and biased criminal sentencing algorithms.

At the same time, the environmental and financial costs of deploying large-scale AI systems are growing at an extremely rapid pace.

If this trajectory continues, it will not only prove environmentally unsustainable, it will also concentrate access to these powerful AI tools to a few wealthy and influential entities with access to vast capital and massive infrastructure.

Why mathematics?

To address issues with a system, whether it’s fixing a car or ensuring reliability in an AI system, it’s crucial to understand how it works. A mechanic cannot fix or even diagnose why a car isn’t operating correctly without understanding how the engine works.

The "engine" for AI is mathematics. In the 1950s, scientists used ideas from logic and probability to teach computers how to make simple decisions. As technology advanced, so did the math, and tools from optimization, linear algebra, geometry, statistics and other mathematical disciplines became the backbone of what are now modern AI systems.

These methods are certainly modeled after aspects of the human brain, but despite the nomenclature of "neural networks" and "machine learning," these systems are essentially giant math engines that carry out vast amounts of mathematical operations with parameters that were optimized using massive amounts of data.

This means improving AI is not just about continuously building bigger computers and using more data; it’s about deepening our understanding of the complex math that governs these systems. By recognizing how fundamentally mathematical AI really is, we can improve its fairness, reliability and sustainable scalability as it becomes an even larger part of everyday life.

Canada’s path forward

So what should Canada do next? Invest in the parts of AI that turn power into dependability. That means funding the science that makes AI systems predictable, auditable and efficient, so hospitals, banks, utilities and public agencies can adopt AI with confidence.

This is not a call for bigger servers; it’s a call for better science, where mathematics is the core scientific engine.

Canada already has a national platform to advance this work: the mathematical sciences institutes the (Pacific Institute for the Mathematical Sciences, Fields Institute for Research in Mathematical Sciences, The Centre de recherches mathématiques, Atlantic Association for Research in the Mathematical Sciences, Banff International Research Station connect researchers across provinces and disciplines, convene collaborative programs and link academia with the public sector.

Together with Canada’s AI institutes (Mila, Vector, Amii) and CIFAR, this ecosystem strengthens both foundational and translational AI nationwide.

Canada’s standing in AI was built on decades of foundational research, work that preceded today’s large models and made them possible. Reinforcing that foundation would allow Canada to lead the next stage of AI development: models that are efficient rather than wasteful, transparent rather than opaque and trustworthy rather than brittle. Investing in mathematical research is not only scientifically essential, it is strategically wise and will strengthen national sovereignty.

The payoff is straightforward: AI that costs less to run, fails less often and earns more public trust. Canada can lead here, not by winning a computing power arms race, but by setting the scientific bar for how AI should work when lives, livelihoods and public resources are at stake.

This article is republished from The Conversation under a Creative Commons license. Read the original article.

Citation: How can Canada become a global AI powerhouse? By investing in mathematics (2025, December 25) retrieved 25 December 2025 from https://phys.org/news/2025-12-canada-global-ai-powerhouse-investing.html

This document is subject to copyright. Apart from any fair dealing for the purpose of private study or research, no part may be reproduced without the written permission. The content is provided for information purposes only.