Is AI really ‘intelligent’? This philosopher says yes
Why Read This
What Makes This Article Worth Your Time
Summary
What This Article Is About
Jane Goodall reviews What Is Intelligence? by Google researcher and polymath Blaise Agüera y Arcas (MIT Press), a sweeping philosophical and scientific argument that AI systems genuinely possess intelligence — not metaphorically, but because computation is the universal substrate of intelligence itself. Against the dominant view (captured in philosopher Daniel Dennett’s phrase “competence without comprehension”) that LLMs are merely sophisticated mimics, Agüera y Arcas argues that prediction is the foundational principle of intelligence across all life forms, from single-celled bacteria to the human brain to Large Language Models.
Drawing on foundational thinkers including Alan Turing, John von Neumann, and microbiologist Lynn Margulis, the book reframes intelligence as a property of systems rather than beings — defined by function, not consciousness. Goodall finds the ideas genuinely important and potentially paradigm-shifting, while noting the 600-page book can be unwieldy and occasionally loses itself in tangential excursions. She recommends it as a work to dip into rather than swallow whole.
Key Points
Main Takeaways
Computation Is the Substrate of All Intelligence
Agüera y Arcas argues that computation — not carbon-based biology — is the universal foundation of intelligence, making brains and AI fundamentally comparable.
Prediction Is the Core of Intelligence
From bacteria predicting survival sequences to neurons firing in patterns, Agüera y Arcas proposes that predictive pattern completion — not conscious thought — is what intelligence fundamentally is.
Intelligence Is a System Property, Not Personal
Rather than belonging to individual minds, intelligence is a property of systems — defined by whether they perform a function. A kidney has it; a rock does not.
Scale Unlocks General Intelligence
Agüera y Arcas’ turning point was accepting that in computation, “bigger really was better” — the key leap from narrow AI (chess-playing) to general AI (philosophical reasoning).
Margulis Challenges Dawkins on Evolution
The book sides with Lynn Margulis’ theory of symbiogenesis over Dawkins’ “selfish gene” — favouring biological complexity through cooperation over competitive natural selection.
Important Ideas, Uneven Execution
Goodall praises the book’s potentially paradigm-shifting thesis while noting it sprawls at 600+ pages, with tangential excursions that dilute its central argument’s force.
Master Reading Comprehension
Practice with 365 curated articles and 2,400+ questions across 9 RC types.
Article Analysis
Breaking Down the Elements
Main Idea
Intelligence Is Computation — In All Life Forms, Including AI
Agüera y Arcas makes the radical claim — reviewed here with cautious admiration — that AI intelligence is real because computation is the universal substrate of all intelligence, from bacterial prediction to human cognition, dissolving the line between biological and artificial minds.
Purpose
To Review and Contextualise a Paradigm-Shifting Book
Goodall writes to introduce a dense, important scientific argument to a general audience — explaining its core claims, mapping its intellectual lineage from Turing to Margulis, and offering a balanced verdict on both its ambitions and its structural weaknesses.
Structure
Contextual → Expository → Comparative → Evaluative
Opens with the AI intelligence debate, exposes the book’s central thesis (computation as intelligence substrate), traces its intellectual lineage through Turing and Margulis, presents the “Brainfuck” experiment, then closes with a balanced critical evaluation of the book’s merits and flaws.
Tone
Measured, Intellectually Engaged & Critically Fair
Goodall writes as a careful reviewer — genuinely excited by the ideas, willing to track their complexity across biology, computation, and philosophy, but unflinching in noting where the book overreaches or loses its focus for a general reader.
Key Terms
Vocabulary from the Article
Click each card to reveal the definition
Build your vocabulary systematically
Each article in our course includes 8-12 vocabulary words with contextual usage.
Tough Words
Challenging Vocabulary
Tap each card to flip and see the definition
A critical term for LLMs, coined by AI researchers, suggesting they generate plausible-sounding text by statistical pattern-matching without any genuine understanding of meaning.
“Scepticism turns to cynicism, often tinged with paranoia about how ‘stochastic parrots’ may start to control our lives.”
The scientific study of regulatory systems, communication, and control in animals and machines — a foundational field for understanding how intelligence and feedback loops work.
“He draws explanatory frameworks from microbiology, philosophy, linguistics, cybernetics, neuroscience and industrial history.”
Relating to a synapse — the junction between two nerve cells across which electrical or chemical signals are transmitted, enabling learning and memory in biological systems.
“The synaptic learning rules in single neurons give rise to local sequence prediction.”
A coined term used by Agüera y Arcas to describe the emergent state where formerly random code becomes a self-replicating, functional computational environment — analogous to life emerging from chemistry.
“The non-functional code or ‘Turing gas’ transforms itself into a ‘computorium’ of replicating code.”
The philosophical view that all events, including mental and biological ones, are fully determined by prior physical causes — leaving no room for emergent properties, purpose, or free will.
“Agüera y Arcas avoids both Sheldrake’s intuitive orientations, and the hard-headed constraints of mechanistic determinism.”
A background marked by both successes and serious failures or controversies — describing a mixed, often troubled past record rather than a consistently positive one.
“Research on intelligence has a chequered history, tainted by eugenics, statistical manipulation and a banal obsession with metrics.”
Reading Comprehension
Test Your Understanding
5 questions covering different RC question types
1According to the article, Agüera y Arcas argues that the brain is like a computer — a useful metaphor for understanding how biological intelligence works.
2What was the significance of the “Brainfuck” programming experiment described in the article?
3Which sentence best expresses Agüera y Arcas’ definition of intelligence as a property of systems rather than individuals?
4Evaluate the following statements about the article’s account of Agüera y Arcas’ intellectual influences and positions.
The article states that Agüera y Arcas engages in a direct and explicit polemical critique of Richard Dawkins’ theory of the selfish gene.
The Antikythera, after which the book’s publishing series is named, is described in the article as an ancient device discovered in a shipwreck.
Goodall’s review acknowledges structural weaknesses in the book alongside its important ideas, noting it is better suited to selective reading than cover-to-cover reading.
Select True or False for all three statements, then click “Check Answers”
5The article suggests that asking whether LLMs “have” intelligence rather than whether they “are” intelligent is a more productive framing. What can be inferred about why this distinction matters in Agüera y Arcas’ framework?
FAQ
Frequently Asked Questions
Philosopher Daniel Dennett coined this phrase to describe AI systems that perform tasks effectively — translating, summarising, reasoning — without actually understanding what they are doing. Agüera y Arcas rejects this framing because it assumes understanding must be distinctly biological or conscious. In his framework, if intelligence is defined by predictive function rather than subjective awareness, the distinction between “competence” and “comprehension” dissolves.
ANI refers to AI systems designed and trained for a specific, narrow task — like playing chess, recognising images, or translating text — with no ability to transfer that capability to other domains. AGI describes a hypothetical system with flexible, general-purpose intelligence capable of reasoning across diverse domains, much like human cognition. Agüera y Arcas argues that scaling up computation may be the key to bridging this gap.
Dawkins’ “selfish gene” theory frames evolution as driven by competitive advantage — organisms and genes that outcompete others survive. Margulis’ theory of symbiogenesis argues that major evolutionary leaps arise through cooperation and merger, not competition — as when mitochondria (once independent bacteria) became part of complex cells. Agüera y Arcas uses Margulis to support his view of intelligence as arising through aggregation and cooperation rather than competitive selection.
Readlite provides curated articles with comprehensive analysis including summaries, key points, vocabulary building, and practice questions across 9 different RC question types. Our Ultimate Reading Course offers 365 articles with 2,400+ questions to systematically improve your reading comprehension skills.
This article is rated Intermediate. While Goodall writes accessibly, the review introduces a dense conceptual framework spanning biology, computation, and philosophy of mind. Readers must track abstract distinctions — such as between metaphor and literal claim, or between “having” vs. “being” intelligent — and follow the intellectual lineage from Turing and von Neumann to Margulis. Familiarity with basic AI terminology (LLMs, AGI) will help, though is not strictly required.
Agüera y Arcas is a Google researcher and self-described polymath with a background in physics and computational neuroscience. His perspective carries weight because he writes from inside the AI industry — having worked at the frontier of large-scale machine learning — while simultaneously engaging seriously with evolutionary biology, philosophy of mind, and the history of computation. His affirmative position on AI intelligence thus cannot easily be dismissed as uninformed enthusiasm.
The Ultimate Reading Course covers 9 RC question types: Multiple Choice, True/False, Multi-Statement T/F, Text Highlight, Fill in the Blanks, Matching, Sequencing, Error Spotting, and Short Answer. This comprehensive coverage prepares you for any reading comprehension format you might encounter.
