★★★★★ · 2025 #AI#tech#philosophy

What Is Intelligence?

by Blaise Agüera y Arcas

The book in brief. Blaise Agüera y Arcas defines intelligence as fundamentally predictive: any system, biological or artificial, that models the world in order to anticipate it. His framework is functionalist, in the tradition of Turing and von Neumann: a heart, a cell or a mind is what it does, not what it is made of. Complexity grows through symbiogenesis, and human and machine cognition are now merging in the same way.

Key ideas

  • No natural boundary. Every line drawn around a cell, an organism or a planet is a choice of frame along two continua, scale and time.
  • What a life is. A life is a region of matter where computation is possible, free energy is available and dynamic stability can take hold.
  • Machine metabolism. Watt’s steam engine in 1776 is framed as a symbiogenetic event on a par with the engulfment that gave us mitochondria.
  • Perception as prediction. The account builds on Helmholtz’s unconscious inference: what reaches the brain is already a model, corrected when the world disagrees.

Why it matters to my work. In Reading What Is Intelligence? (2): Where Does A Life Begin? I use the functionalist frame to replace the argument over whether AI is “really” alive with three measurable questions: computation, free energy and robustness. The first reading note covers symbiogenesis, which is also my deeper frame in On Thinking with (AI) Machines. In Compressing a Causal Chain: What Has to Be True I carry the idea over to investing: a company, like the thesis written about it, must recompress when the regularities around it move.