Why Machines Learn

Why Machines Learn: The Elegant Math Behind Modern AI does something most AI books avoid: it actually teaches the math. Anil Ananthaswamy walks readers through the algebra, calculus, and probability that make neural networks work. He does it with genuinely good storytelling rather than dry derivation.

Ananthaswamy holds a Master’s degree in Electrical Engineering from the University of Washington. He has built his career as a science writer, including a fellowship as an MIT Knight Science Journalism Fellow. That dual background, technical training paired with a writer’s instinct for narrative, is exactly what this material needs.

The chapters build steadily, starting with early neuron models and moving through vectors, matrix operations, and probability theory. Later chapters cover kernel methods, Hopfield networks, and the breakthroughs in image recognition that GPUs and large datasets made possible. Readers do not need to grasp every equation on the first pass. The book rewards a second and third read the same way any real math text does.

This will not be effortless reading for everyone, especially anyone rusty on linear algebra or statistics. It is worth the effort because it replaces vague AI metaphors with the actual mechanics underneath them.

Recommended for IT leaders and security practitioners who want to understand how AI actually computes. It goes well beyond how vendors describe it.

Originally published in ISSA Journal’s "The Cyber Library" column, May-June 2026 issue, by Jay Carson, Semper Sec’s Cyber Librarian. Read more at issa.org.