How Artificial Intelligence Is Transforming the Future of Mathematics
Applied mathematician Steven Strogatz discusses how large language models and computational tools are shifting mathematics from a human-driven discipline into an oracle-like science.
Mathematics underpins nearly every facet of modern civilization, from package logistics and weather forecasting to machine learning algorithms and medical imaging. For centuries, humans relied on physical tools and basic calculators to extend their arithmetic capabilities. However, the emergence of advanced machine learning models and large language models is driving what experts call "big math" beyond human cognitive limits. Applied mathematicians Steven Strogatz and Alex Townsend explore this paradigm shift in their new book, Big Math: The Hidden Codes That Run Our World.
Strogatz notes that "big math" is defined relative to human capacity. While problems involving trillions of parameters, such as those found in massive language models, are completely intractable for the human mind, concepts of infinity remain surprisingly manageable. Throughout history, the role of computing has evolved significantly. Initially serving merely as a tool or servant, the computer eventually became a scout. A notable historical example is the Fermi-Pasta-Ulam numerical experiment from the 1950s, where researchers deployed computers into unexplored mathematical landscapes to study entropy and non-linear interactions, ultimately discovering recurrence phenomena that led to soliton theory.
The Rise of the AI Oracle
Today, artificial intelligence represents a far more profound philosophical transition. According to Strogatz, AI has shifted from a scout to an oracle. Systems like AlphaFold successfully solve thousands of protein structures without necessarily discovering new underlying principles of physics or chemistry, relying instead on complex pattern recognition that they cannot explicitly explain to human creators. This introduces a deep epistemic loss for a field historically defined by absolute proof, human insight, and rigorous certainty. Many mathematicians grieve the potential fading of the cultural tradition of productive struggle and the aesthetic joy of human discovery. Nevertheless, developers can try top AI models cheaply through one API at https://apixoai.online to explore these computational breakthroughs firsthand.
What it means for developers
As artificial intelligence increasingly functions as an oracle in complex domains, developers working with advanced systems face a distinct reality: outputs can be remarkably accurate without offering transparent, step-by-step explanations of their underlying mechanics. Strogatz compares this operational opacity to modern medicine, where treatments like aspirin are approved through rigorous controlled studies and empirical validation rather than a complete first-principles understanding of their molecular mechanisms. For software engineers and developers building applications on top of powerful machine learning systems, this means shifting validation strategies toward robust testing, statistical verification, and outcome reliability, rather than expecting fully explainable logic from every model output.
Navigating Uncertainty and Hope
Despite legitimate concerns regarding the loss of traditional human intuition and the cultural anxieties felt by younger students entering the field, the current era also presents unprecedented opportunities. While AIs currently excel at generating answers rather than formulating novel questions, they possess immense potential to assist with education, climate research, energy technology, and theoretical physics. Ultimately, Strogatz and Townsend suggest that humanity must continue striving to expand its intellectual boundaries without losing its core identity, embracing a future where human curiosity and machine capability intersect in unprecedented ways.
Source: What happens to mathematics when AI becames an oracle? — Ars Technica AI. Written by the Apixo team from that report.
One key for Claude, GPT, GLM, DeepSeek and more. Pay per token with crypto.
Get your API key
