Mathematics did not arrive at artificial intelligence by accident. The tools that make machine learning possible, from matrix operations to probability chains, were developed over centuries by people solving entirely different problems. A new book by mathematicians Steven Strogatz and Alex Townsend, Big Math (Little, Brown and Company, 2026), traces that lineage with unusual clarity. The result is a portrait of how abstract mathematical thinking, often centuries old, quietly became the engine of the most consequential technology of the present era.
The Bamboo Sticks That Predicted Neural Networks
One of the most striking arguments in Big Math is that the mathematics central to AI is not new. Strogatz and Townsend describe how ancient Chinese mathematicians solved accounting problems, specifically estimating the edible rice yield from bundles of unprocessed rice, by manipulating bamboo sticks. The underlying structure of those calculations, expressed in modern terms, involves systems of linear equations, matrices, and vectors.
This is not a historical curiosity. Those same mathematical structures sit at the core of how neural networks operate today. The manipulation of massive matrices, containing hundreds of billions and even trillions of numbers, is described in the book as the heart of generative AI. The Chinese “reckoners” working with bamboo had no idea they were sketching the blueprint for a technology that would arrive roughly two millennia later.
The book also recovers a telling observation from nineteenth-century mathematician James Joseph Sylvester, who coined the word “matrix” and described it as a “mother, a generator, an entity that gave birth to other mathematical objects.” Strogatz and Townsend note that this framing feels eerily prescient given what matrices now produce.
From Social Networks to Search Engines: The Long Arc of Matrix Methods
The book does not stop at ancient history. It traces how matrix methods evolved through the twentieth century to become tools for understanding human relationships. Leon Festinger’s 1949 idea that repeated matrix-vector multiplications could uncover hidden structures in social networks eventually gave rise to methods for measuring influence, detecting clusters, and ranking individuals by their degree of connection.
That lineage leads directly to one of the most familiar applications in technology history. Strogatz and Townsend explain how the board game Snakes and Ladders can be used to understand a Markov chain, a mathematical formulation in which the probability of the next state depends only on the current state, not on any previous one. This structure was key to Google’s early search algorithms.
The book also recounts the story of George Dantzig, who as a graduate student at the University of California, Berkeley, in 1939 arrived late to a statistics class and copied two equations from the blackboard, assuming they were homework. He solved them days later and apologized for the delay. They were open, unsolved problems. The story later became the basis for a scene in the 1997 film Good Will Hunting.
These anecdotes are not decoration. They illustrate a pattern: mathematical tools developed in one context, often with no applied purpose in mind, repeatedly turn out to be exactly what a future technology requires.
Why the History of Math Is Also the History of AI
Here is what most coverage of AI misses. The public conversation about artificial intelligence tends to focus on recent breakthroughs, on what a model can do today versus what it could do last year. Big Math reframes the question. The capabilities that seem sudden are built on mathematical foundations that are anything but.
Regina Barzilay’s story, which opens the book, makes this concrete. A computer scientist who had breast cancer, Barzilay decided after completing treatment in 2014 to address the lack of tools for early detection. The technology she needed already existed in the form of machine learning, but it was being used primarily to recommend films and guide online purchasing behavior. By collaborating with radiologists and computer scientists, she developed a machine-learning model capable of predicting a person’s risk of developing breast cancer within the next five years from a single mammogram.
The mathematics behind that model was not invented for medicine. It was adapted from structures that had been accumulating for centuries. This is the deeper point Strogatz and Townsend are making: “big math,” as they call it, is the shared substrate beneath technologies that are reshaping how people work, learn, shop, travel, communicate, and heal.
Understanding that substrate does not require becoming a mathematician. But it does require recognizing that AI is not a rupture with the past. It is a culmination of one.
In Short
The mathematics powering AI, including matrices, vectors, linear equations, and Markov chains, was developed over centuries to solve problems that had nothing to do with computing. Big Math by Steven Strogatz and Alex Townsend traces that history from ancient Chinese accounting to Google’s search algorithms, showing that what looks like a technological revolution is also the latest chapter in a very long mathematical story.
Based on reporting from Nature.