What happened
- Nature published on November 9, 2026 a review of Big Math: The Hidden Codes That Run Our World, by mathematicians Steven Strogatz and Alex Townsend (Little, Brown, 2026). It is written by Anil Ananthaswamy, author of Why Machines Learn.
- The book argues that a little-seen mathematics, which they call “big math”, underpins technologies that change how people work, play, learn, shop, travel, communicate and heal.
- It opens with Regina Barzilay, a computer scientist who had breast cancer and, after finishing treatment in 2014, decided to build tools to spot early warning signs.
- With radiologists and other scientists, Barzilay developed a machine-learning model that predicts a person’s five-year breast cancer risk from a single mammogram.
Why it matters
The review arrives in a year when, according to the caption on the piece, AI tools have solved several mathematics and physics problems that had been open for years. Understanding what lies under those systems helps judge what they can and cannot do.
The authors warn they do not aim to make maths “cozy or accessible”, yet the book has few equations and is easier to read than other Strogatz titles. For people who work with AI models without a mathematical background, it offers a bridge.
The number
From hundreds of billions to trillions of numbers. That is the size of the matrices multiplied at the heart of generative AI, according to the review. The word “matrix” was coined in the nineteenth century by James Joseph Sylvester, who imagined it as a “mother”, an entity that gave birth to other mathematical objects.
Context
The book runs through historical examples. One is the Chinese mathematicians who solved accounting problems to estimate the edible rice from bundles still in the husk: a system of linear equations with matrices and vectors, the maths that underpins AI today, solved then by moving bamboo sticks.
Another is Leon Festinger’s 1949 idea that repeatedly multiplying a matrix by a vector can reveal hidden structures in social networks, a method later used to measure influence and detect clusters. There is also the board game Snakes and Ladders, which illustrates a Markov chain, a piece of Google’s early search algorithms: the next state depends only on the current one.
At Mamífero we have followed that crossover. We reported that Axiom verified the 7,654-page Stacks Project with agents, that an AI-assisted proof closed a 2008 problem on quantum advantage, that IBM and Lisbon turned zeolite synthesis recipes into steps with an LLM and that rewarding each search step makes agents more reliable.
What’s next
- The review announces no new result: it is a reading guide. What it does leave is a question about how much mathematical creativity machines can take on.
- Nature keeps a series on how AI is reshaping discovery in maths and physics, linked from the review itself.
- Readers who want to go further will find in the book the story of George Dantzig, a Berkeley student in 1939 who arrived late to a statistics class.
Bottom line
The mathematics that moves AI is, at its base, linear algebra with centuries of history. The book proposes reading current advances as one more stage of that line, not as a break.
Sources
- Nature (Anil Ananthaswamy), “AI has transformed maths — this is how we got here”, book review, November 9, 2026, DOI 10.1038/d41586-026-03627-6, link.
- Cover image: MathKnight ✡ (Talk) No machine-readable author provided. MathKnight as, CC BY 2.5, via Wikimedia Commons
Edited by Rodrigo Cornejo. How we select and verify: who writes these notes.
