Ian's Noesis Wiki

Properties1
tagsai, agi, functionalism, philosophy-of-mind, prediction, computational-neuroscience, comparison

4 min read

What is Intelligence? (Book)

Author: Blaise Agüera y Arcas (VP and Fellow at Google, CTO of Technology & Society) Full title: What is Intelligence? Lessons from AI About Evolution, Computing, and Minds Series: Antikythera Book Series (MIT Press × Berggruen Institute), September 2025 URL: https://whatisintelligence.antikythera.org/

A comprehensive online book arguing that prediction and intelligence are equivalent, building on the Turing/von Neumann functionalist tradition and extending it across biology, evolution, and AI.

Central Thesis: Prediction = Intelligence

The book’s core claim is that the emergence of general intelligence in large language models trained solely on next-word prediction was not a surprise or a hack — it was the natural consequence of prediction being the fundamental operation underlying all cognitive tasks. Any test that can be expressed in language can be reformulated as a next-word prediction problem, including tests of knowledge, reasoning, math, professional qualification, and even moral judgment. Therefore “a narrow language task contains all keyboard-based cognitive tasks.” 1

The Functionalist Framework

The book adopts the functionalist stance of Alan Turing and John von Neumann:

  • A function is what it does — two functions are equivalent if their outputs are indistinguishable given the same inputs
  • Turing Test as valid — sustained successful imitation is the real thing; the “philosophical zombie” objection is unscientific (it asserts something appears to be X by every test yet is really Y)
  • Extended to biology — living organisms are compositions of functions (kidneys, hearts, brains), and therefore themselves functions. “Everything alive is a computer.”
  • Kidney Turing Test — if an artificial device does what a kidney does, it is a kidney functionally
  • Avoids both vitalism (soul/spirit) and strict materialism (can’t account for purpose/teleology)

Key Concepts

  • AI-completeness of language prediction — next-word prediction as a universal cognitive task; doing it well requires solving AGI
  • Computational symbiogenesis — life, evolution, and intelligence as computational processes
  • LaMDA watershed — the 2021 model that unexpectedly demonstrated conversational ability from pure next-word prediction, challenging assumptions about what extra machinery intelligence requires
  • Denial vs acceptance — the two broad responses to AI capabilities; the book advocates for acceptance of AI as genuinely intelligent
  • turing-test — The behavioral criterion the book defends as valid; the Kidney Turing Test extends it to biology
  • chinese-room-argument — Searle’s famous challenge to functionalism: does symbol manipulation produce understanding, or just the appearance of it?
  • aristotle — Aristotle’s hylomorphism (soul as form of the body) is a foundational precursor to functionalism
  • psychophysical-harmony — Bayesian argument for theism claiming functionalism can’t explain which psycho-functional identity obtains
  • active-inference — Both frameworks center prediction/inference as fundamental to intelligence, though from different angles. Active inference derives action from Bayesian inference over states; this book argues prediction itself is the core operation from which intelligence emerges.
  • spiking-neural-networks — Another computational framework for understanding intelligence, operating at the neuronal rather than the algorithmic/computational level that this book focuses on.
  • emergence — The metaphysical question underneath the book’s functionalism: can intelligence/consciousness be physically-realized organizational properties? The book presumes weak emergence suffices.

Key Questions the Book Addresses

  1. Why has real AI only emerged after 70 years of effort? (Scale? Transformer architecture? Something else?)
  2. What do current AI models lack relative to human brains?
  3. Are “philosophical zombies” a real thing? Is there consciousness in AI?
  4. Are animals, plants, fungi, and bacteria intelligent? Conscious?
  5. What about agency, free will, and existential risk from AI?

References

  • Agüera y Arcas, B. (2025). What is Intelligence? Lessons from AI About Evolution, Computing, and Minds. MIT Press / Antikythera. 1
  • Turing, A. (1950). Computing Machinery and Intelligence. Mind.
  • Thoppilan, R. et al. (2022). LaMDA: Language Models for Dialog Applications.
  • Agüera y Arcas, B. (2022). — on LaMDA and philosophical zombies.

Footnotes

  1. raw/articles/aguera-y-arcas-2025-what-is-intelligence-intro.md ↩ ↩2

Graph View

Backlinks