Life and AI

For almost four billion years, life has been getting better at one thing: keeping what helps it renew itself. Genes kept those lessons first, then nervous systems, language, writing and science. Artificial intelligence is the newest store, trained on almost everything people have written down. This book asks what it has in common with the others, where it differs, and what the whole effort is for.

It starts inside a language model: how it holds knowledge and behaviour, why a small model that thinks longer can beat a larger one, and how context serves as memory. From there it widens to what understanding is, why every model leaves something out, causes versus correlations, the energy price of intelligence, error correction in DNA and in AI agents, whether AI can create knowledge, and what keeps a system open to change.

Life and AI are made of different stuff, yet they face many of the same problems: what to fix and what to keep flexible, what to compute in advance and what to work out on the spot, how to catch an error before it compounds. Setting their solutions side by side sharpens our understanding of both.

The book is written for readers who use AI and know its main ideas: students of AI and the sciences, programmers and curious professionals. It names real systems and recent studies, and keeps the basics in an appendix. It has 42 chapters in ten parts, 42 figures, a glossary, a reading list and an index.

Its answer to the question of purpose is a direction rather than a destination: more life, in more ways, for longer, understood ever better.