What do you think?
Rate this book
What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds
What intelligence really is, and how AI’s emergence is a natural consequence of evolution.
It has come as a shock to some AI researchers that a large neural net that predicts next words seems to produce a system with general intelligence. Yet this is consistent with a long-held view among some neuroscientists that the brain evolved precisely to predict the future—the “predictive brain” hypothesis.
In What Is Intelligence?, Blaise Agüera y Arcas takes up this idea—that prediction is fundamental not only to intelligence and the brain, but to life itself—and explores the wide-ranging implications. These include radical new perspectives on the computational properties of living systems, the evolutionary and social origins of intelligence, the relationship between models and reality, entropy and the nature of time, the meaning of free will, the problem of consciousness, and the ethics of machine intelligence.
The book offers a unified picture of intelligence from molecules to organisms, societies, and AI, drawing from a wide array of literature in many fields, including computer science and machine learning, biology, physics, and neuroscience. It also adds recent and novel findings from the author, his research team, and colleagues. Combining technical rigor and deep up-to-the-minute knowledge about AI development, the natural sciences (especially neuroscience), and philosophical literacy, What Is Intelligence? argues—quite against the grain—that certain modern AI systems do indeed have a claim to intelligence, consciousness, and free will.
It has come as a shock to some AI researchers that a large neural net that predicts next words seems to produce a system with general intelligence. Yet this is consistent with a long-held view among some neuroscientists that the brain evolved precisely to predict the future—the “predictive brain” hypothesis.
In What Is Intelligence?, Blaise Agüera y Arcas takes up this idea—that prediction is fundamental not only to intelligence and the brain, but to life itself—and explores the wide-ranging implications. These include radical new perspectives on the computational properties of living systems, the evolutionary and social origins of intelligence, the relationship between models and reality, entropy and the nature of time, the meaning of free will, the problem of consciousness, and the ethics of machine intelligence.
The book offers a unified picture of intelligence from molecules to organisms, societies, and AI, drawing from a wide array of literature in many fields, including computer science and machine learning, biology, physics, and neuroscience. It also adds recent and novel findings from the author, his research team, and colleagues. Combining technical rigor and deep up-to-the-minute knowledge about AI development, the natural sciences (especially neuroscience), and philosophical literacy, What Is Intelligence? argues—quite against the grain—that certain modern AI systems do indeed have a claim to intelligence, consciousness, and free will.
600 pages, Paperback
Published September 23, 2025
51 people are currently reading
566 people want to read
Ratings & Reviews
Friends & Following
Create a free account to discover what your friends think of this book!
Community Reviews
Displaying 1 - 15 of 15 reviews
October 9, 2025
The author is a CTO at Google working in AI. The main idea of this book is that life that can reproduce itself has an internal model of itself. The same can be said of the self-- an entity with a self has an internal model of the world including a model of the entity itself. He writes:
"My contention is that theory of mind:
* Powers the “intelligence explosions” observed in our own lineage, the hominins, and in other brainy species;
* Gives us the ability to entertain counterfactual “what-ifs”;
* Motivates, and is enhanced by, the development of language;
* Allows us to make purposive decisions beyond “autopilot mode”;
* Underwrites free will;
* Operates both in social networks and within individual brains;
* Results automatically from symbioses among predictors; and
* Is the origin and mechanism of consciousness.
In a sense, theory of mind is mind."
Exploring information theory, cybernetics, fractals, thermodynamics, evolution, large language models, animals and brains, the author finds fascinating connections between these fields: so the evolution of brains, the thermodynamics of intelligence, the self-reference of LLMs, the fractal of minds (one of his core ideas is that minds are made of subminds which are themselves minds in a fractal way), and so forth.
For example-- why do moths fly the way they do, seemingly wasting energy? The author contends that their predators are so much smarter and faster (and inherently so-- there is no way a moth could evolve to be as smart as a bat) there is no way they can outwit or outrun them. So they use the only strategy open to them is to move chaotically and hope to evade by chance.
He says that LLMs are intelligent because they attempt to predict the future (the next token) and that is the nature of intelligence-- a kind of backwards causality that can locally overcome entropy by actively anticipating the future.
He has one of the most thoughtful treatments of split-brain patients I've seen.
Like everyone who thinks they have solved the hard problem of consciousness, he is missing the point. He has only solved the easy problem, of course.
The book is one of the first that I've read that really grapples with the question, since LLMs are so capable, what does that tell us about how the mind works?
You can read a copy for free here: https://whatisintelligence.antikyther... I would recommend that method because there are a lot of animated illustrations you would miss otherwise.
"My contention is that theory of mind:
* Powers the “intelligence explosions” observed in our own lineage, the hominins, and in other brainy species;
* Gives us the ability to entertain counterfactual “what-ifs”;
* Motivates, and is enhanced by, the development of language;
* Allows us to make purposive decisions beyond “autopilot mode”;
* Underwrites free will;
* Operates both in social networks and within individual brains;
* Results automatically from symbioses among predictors; and
* Is the origin and mechanism of consciousness.
In a sense, theory of mind is mind."
Exploring information theory, cybernetics, fractals, thermodynamics, evolution, large language models, animals and brains, the author finds fascinating connections between these fields: so the evolution of brains, the thermodynamics of intelligence, the self-reference of LLMs, the fractal of minds (one of his core ideas is that minds are made of subminds which are themselves minds in a fractal way), and so forth.
For example-- why do moths fly the way they do, seemingly wasting energy? The author contends that their predators are so much smarter and faster (and inherently so-- there is no way a moth could evolve to be as smart as a bat) there is no way they can outwit or outrun them. So they use the only strategy open to them is to move chaotically and hope to evade by chance.
He says that LLMs are intelligent because they attempt to predict the future (the next token) and that is the nature of intelligence-- a kind of backwards causality that can locally overcome entropy by actively anticipating the future.
He has one of the most thoughtful treatments of split-brain patients I've seen.
Like everyone who thinks they have solved the hard problem of consciousness, he is missing the point. He has only solved the easy problem, of course.
The book is one of the first that I've read that really grapples with the question, since LLMs are so capable, what does that tell us about how the mind works?
You can read a copy for free here: https://whatisintelligence.antikyther... I would recommend that method because there are a lot of animated illustrations you would miss otherwise.
Read
December 31, 2025** FAREWELL 2025 **
"What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds" by Blaise Aguera y Arcas offers a sweeping reinterpretation of intelligence that moves beyond human exceptionalism and biology-centered thinking. Rather than treating intelligence as something mysterious or uniquely human, the book argues that it is best understood as a computational process rooted in prediction. From the earliest chemical systems that gave rise to life to modern artificial intelligence models, the same underlying principle applies: systems that survive and succeed are those that can anticipate their environment and adjust accordingly. By reframing intelligence as prediction rather than consciousness or reasoning alone, the book invites readers to see AI not as a clever imitation of the mind, but as a genuine continuation of a process that has been unfolding for billions of years.
The story begins at the origin of life itself, when Earth was an unstable and hostile environment. Early life likely emerged not through sudden miracles, but through gradual chemical processes that allowed certain systems to sustain themselves. Deep-sea hydrothermal vents provided natural structures where chemical reactions could cycle and reinforce one another, creating primitive metabolic systems. These early life forms did not think or reason, but they did something crucial: they maintained themselves in changing conditions. From the very start, survival depended on responding to the environment in ways that favored stability. This basic requirement laid the groundwork for everything that would later be called intelligence.
As life evolved, major leaps occurred through cooperation and integration rather than isolated improvement. One of the most significant events was the merger of simple cells into more complex ones, resulting in mitochondria and eventually multicellular organisms. These evolutionary breakthroughs mirror developments in technology, where combining existing components in new ways leads to sudden increases in capability. The book draws a strong parallel between biology and computation, emphasizing that living cells process information, follow instructions, and replicate themselves much like machines executing code. From this perspective, life itself can be seen as a form of computation carried out by chemistry.
At its simplest level, intelligence emerges as the ability to regulate internal conditions while responding to external change. Even single-celled organisms demonstrate this. A bacterium navigating toward nutrients is not acting randomly; it is comparing present conditions to recent past experiences and adjusting its behavior accordingly. This comparison is a form of statistical inference, allowing the organism to predict whether it is moving toward or away from favorable conditions. Although primitive, this process captures the essence of intelligence: using information over time to guide action in the service of survival.
Inside every living system is an internal model of the world, however simple, that links perception, internal state, and action. Evolution gradually refines these models, favoring organisms that predict more accurately and respond more effectively. Over many generations, these predictive systems become more flexible and general, enabling increasingly complex behaviors. Once a system can model itself as part of the world it inhabits, it acquires a goal - continued existence. This goal-directed behavior is not imposed from outside; it arises naturally from the dynamics of prediction and feedback.
The book then turns to the history of computing to show how similar ideas emerged in machines. Early thinkers such as Leibniz, Lovelace, Turing, and von Neumann recognized that calculation could extend beyond arithmetic into logic, patterns, and even thought itself. However, much of early computing focused on rigid, rule-based systems designed for precision and efficiency. This approach shaped early artificial intelligence, which treated reasoning as symbolic manipulation and separated logic from emotion, learning, and uncertainty.
Over time, it became clear that this view did not match how real minds work. Biological neurons are not simple logic gates but complex, adaptive units embedded in feedback loops. An alternative tradition, known as cybernetics, emphasized learning, control, and prediction through continuous interaction with the environment. Although this approach was conceptually powerful, it lacked the computational resources needed to fully realize its vision. Still, its core insight - that intelligence arises from prediction and feedback - remained influential.
Modern AI began to take shape when computing power and data became sufficient to support learning-based systems. Artificial neural networks, inspired loosely by biology, allowed machines to adjust their internal parameters based on experience. Techniques such as nonlinear activation functions enabled systems to recognize patterns across varying conditions. As these systems trained on large amounts of data, they developed the ability to generalize, applying learned patterns to new situations. This capacity for transfer learning marked a major step toward more flexible intelligence.
A crucial breakthrough came with self-supervised learning, where models learn by predicting missing or future information rather than relying on explicit labels. By filling in gaps, systems are forced to build internal representations of the structure of the world. This mirrors how humans learn, constantly predicting what they will see next and correcting errors through feedback. Prediction, rather than explicit instruction, becomes the driver of understanding.
Language models exemplify this principle at scale. By training on the task of predicting the next word in a sequence, these systems implicitly learn grammar, facts, reasoning patterns, and aspects of human culture. Language functions as a compressed representation of thought, allowing minds to share internal models with one another. While language feels special to humans, it is simply one powerful medium through which prediction operates. Advances such as transformer architectures enabled models to handle long-range dependencies, allowing richer contextual understanding and more coherent outputs.
Despite their impressive capabilities, current AI systems still have limitations. They lack continuous learning and persistent memory across interactions, which can lead to inconsistencies in reasoning. However, prompting models to reason step by step often improves performance, highlighting the role of structured prediction in both human and machine thinking. These systems already demonstrate forms of intelligence comparable to that seen in many animals, suggesting that intelligence exists on a spectrum rather than as a single threshold.
Looking ahead, the book cautions against viewing AI as a sudden leap toward a distant superintelligence. Instead, it frames recent advances as part of a gradual transition toward more general-purpose systems. Just as earlier technological revolutions reshaped society by increasing interdependence, AI represents another layer of collective intelligence. This brings both expanded capabilities and new vulnerabilities, as humans become more reliant on complex systems they do not fully control.
In conclusion, "What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds" by Blaise Aguera y Arcas presents intelligence as a universal process grounded in prediction, feedback, and adaptation. From the chemistry of early life to the architectures of modern AI, the same principles recur across vastly different systems. Intelligence is not an exclusive human trait nor a sudden technological invention, but a natural outcome of systems that model their world in order to persist within it. By recognizing AI as part of this long evolutionary story, the book shifts the conversation away from fear and toward understanding, emphasizing that the future of intelligence will be increasingly shared, hybrid, and collective.
"What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds" by Blaise Aguera y Arcas offers a sweeping reinterpretation of intelligence that moves beyond human exceptionalism and biology-centered thinking. Rather than treating intelligence as something mysterious or uniquely human, the book argues that it is best understood as a computational process rooted in prediction. From the earliest chemical systems that gave rise to life to modern artificial intelligence models, the same underlying principle applies: systems that survive and succeed are those that can anticipate their environment and adjust accordingly. By reframing intelligence as prediction rather than consciousness or reasoning alone, the book invites readers to see AI not as a clever imitation of the mind, but as a genuine continuation of a process that has been unfolding for billions of years.
The story begins at the origin of life itself, when Earth was an unstable and hostile environment. Early life likely emerged not through sudden miracles, but through gradual chemical processes that allowed certain systems to sustain themselves. Deep-sea hydrothermal vents provided natural structures where chemical reactions could cycle and reinforce one another, creating primitive metabolic systems. These early life forms did not think or reason, but they did something crucial: they maintained themselves in changing conditions. From the very start, survival depended on responding to the environment in ways that favored stability. This basic requirement laid the groundwork for everything that would later be called intelligence.
As life evolved, major leaps occurred through cooperation and integration rather than isolated improvement. One of the most significant events was the merger of simple cells into more complex ones, resulting in mitochondria and eventually multicellular organisms. These evolutionary breakthroughs mirror developments in technology, where combining existing components in new ways leads to sudden increases in capability. The book draws a strong parallel between biology and computation, emphasizing that living cells process information, follow instructions, and replicate themselves much like machines executing code. From this perspective, life itself can be seen as a form of computation carried out by chemistry.
At its simplest level, intelligence emerges as the ability to regulate internal conditions while responding to external change. Even single-celled organisms demonstrate this. A bacterium navigating toward nutrients is not acting randomly; it is comparing present conditions to recent past experiences and adjusting its behavior accordingly. This comparison is a form of statistical inference, allowing the organism to predict whether it is moving toward or away from favorable conditions. Although primitive, this process captures the essence of intelligence: using information over time to guide action in the service of survival.
Inside every living system is an internal model of the world, however simple, that links perception, internal state, and action. Evolution gradually refines these models, favoring organisms that predict more accurately and respond more effectively. Over many generations, these predictive systems become more flexible and general, enabling increasingly complex behaviors. Once a system can model itself as part of the world it inhabits, it acquires a goal - continued existence. This goal-directed behavior is not imposed from outside; it arises naturally from the dynamics of prediction and feedback.
The book then turns to the history of computing to show how similar ideas emerged in machines. Early thinkers such as Leibniz, Lovelace, Turing, and von Neumann recognized that calculation could extend beyond arithmetic into logic, patterns, and even thought itself. However, much of early computing focused on rigid, rule-based systems designed for precision and efficiency. This approach shaped early artificial intelligence, which treated reasoning as symbolic manipulation and separated logic from emotion, learning, and uncertainty.
Over time, it became clear that this view did not match how real minds work. Biological neurons are not simple logic gates but complex, adaptive units embedded in feedback loops. An alternative tradition, known as cybernetics, emphasized learning, control, and prediction through continuous interaction with the environment. Although this approach was conceptually powerful, it lacked the computational resources needed to fully realize its vision. Still, its core insight - that intelligence arises from prediction and feedback - remained influential.
Modern AI began to take shape when computing power and data became sufficient to support learning-based systems. Artificial neural networks, inspired loosely by biology, allowed machines to adjust their internal parameters based on experience. Techniques such as nonlinear activation functions enabled systems to recognize patterns across varying conditions. As these systems trained on large amounts of data, they developed the ability to generalize, applying learned patterns to new situations. This capacity for transfer learning marked a major step toward more flexible intelligence.
A crucial breakthrough came with self-supervised learning, where models learn by predicting missing or future information rather than relying on explicit labels. By filling in gaps, systems are forced to build internal representations of the structure of the world. This mirrors how humans learn, constantly predicting what they will see next and correcting errors through feedback. Prediction, rather than explicit instruction, becomes the driver of understanding.
Language models exemplify this principle at scale. By training on the task of predicting the next word in a sequence, these systems implicitly learn grammar, facts, reasoning patterns, and aspects of human culture. Language functions as a compressed representation of thought, allowing minds to share internal models with one another. While language feels special to humans, it is simply one powerful medium through which prediction operates. Advances such as transformer architectures enabled models to handle long-range dependencies, allowing richer contextual understanding and more coherent outputs.
Despite their impressive capabilities, current AI systems still have limitations. They lack continuous learning and persistent memory across interactions, which can lead to inconsistencies in reasoning. However, prompting models to reason step by step often improves performance, highlighting the role of structured prediction in both human and machine thinking. These systems already demonstrate forms of intelligence comparable to that seen in many animals, suggesting that intelligence exists on a spectrum rather than as a single threshold.
Looking ahead, the book cautions against viewing AI as a sudden leap toward a distant superintelligence. Instead, it frames recent advances as part of a gradual transition toward more general-purpose systems. Just as earlier technological revolutions reshaped society by increasing interdependence, AI represents another layer of collective intelligence. This brings both expanded capabilities and new vulnerabilities, as humans become more reliant on complex systems they do not fully control.
In conclusion, "What Is Intelligence?: Lessons from AI About Evolution, Computing, and Minds" by Blaise Aguera y Arcas presents intelligence as a universal process grounded in prediction, feedback, and adaptation. From the chemistry of early life to the architectures of modern AI, the same principles recur across vastly different systems. Intelligence is not an exclusive human trait nor a sudden technological invention, but a natural outcome of systems that model their world in order to persist within it. By recognizing AI as part of this long evolutionary story, the book shifts the conversation away from fear and toward understanding, emphasizing that the future of intelligence will be increasingly shared, hybrid, and collective.
November 29, 2025
A brilliant book. It covers an immensely large view whilst maintaining an entertaining style. In each sitting I discovered a fascinating new connection. The main argument is provocative, arguing that computation is the central defining feature of living systems, and by extension that the advance of AI is part of a continuous process from the earliest forms of life on Earth.
A lot of contemporary writing on AI falls into one of two camps: breathless boosterism or paranoid critique. This work shows how both are misguided. Aguera y Arca comes across as a deeply considerate scholar, who is excited about the advances in this technology, but has a deep understanding of its proper place in human history and how to think about it holistically.
I would recommend it to anyone who wants to understand the world today, or its coming path.
A lot of contemporary writing on AI falls into one of two camps: breathless boosterism or paranoid critique. This work shows how both are misguided. Aguera y Arca comes across as a deeply considerate scholar, who is excited about the advances in this technology, but has a deep understanding of its proper place in human history and how to think about it holistically.
I would recommend it to anyone who wants to understand the world today, or its coming path.
November 10, 2025
the beginning of the book is great but quickly goes downhill afterwards
December 5, 2025
Predictie (voorspelling) staat centraal als het fundamentele mechanisme achter zowel biologische als kunstmatige intelligentie. Dit boek breekt met het idee dat intelligentie draait om symbolische logica of vooraf geprogrammeerde regels, en beargumenteert dat het vermogen om de toekomst te anticiperen de basisvoorwaarde voor intelligentie/cognitie is.
AyA laat zien dat voor biologische wezens de primaire functie van de hersenen niet is het ‘denken’ op zich, maar het bewegen en overleven in een complexe wereld. Een brein dat voortdurend reageert op nieuwe prikkels is namelijk traag en energieverslindend. Een brein dat voorspelt wat er gaat gebeuren en alleen reageert als de voorspelling niet klopt (een prediction error), is veel efficiënter. Wat wij ‘zien’ of ‘horen’ is daarom grotendeels een hallucinatie, gebaseerd op wat we verwachten, gecorrigeerd door zintuiglijke data. Dit sluit aan bij de theorie van predictive coding in de neurowetenschappen. AyA gebruikt dit biologische kader om de werking van moderne Large Language Models te duiden en te verdedigen, o.a. dat predictie het taalsysteem dringt om een intern model van de wereld te bouwen (predictie leidt tot representatie).
Sterk aan dit boek is dat AyA AI-modellen weghaalt uit de sfeer van magie of abstracte wiskunde en het plaatst in de context van de biologische evolutie. Minder sterk is dat de auteur soms last heeft van de Silicon Valley-bias en soms te grote filosofische sprongen maakt. De gelijkstelling van statistische predictie aan begrip, cognitie en bewustzijn is filosofisch zeer omstreden. Af en toe krijg je het gevoel dat complexe vraagstukken over wat het betekent om mens te zijn, gereduceerd worden tot optimalisatie-problemen.
AyA laat zien dat voor biologische wezens de primaire functie van de hersenen niet is het ‘denken’ op zich, maar het bewegen en overleven in een complexe wereld. Een brein dat voortdurend reageert op nieuwe prikkels is namelijk traag en energieverslindend. Een brein dat voorspelt wat er gaat gebeuren en alleen reageert als de voorspelling niet klopt (een prediction error), is veel efficiënter. Wat wij ‘zien’ of ‘horen’ is daarom grotendeels een hallucinatie, gebaseerd op wat we verwachten, gecorrigeerd door zintuiglijke data. Dit sluit aan bij de theorie van predictive coding in de neurowetenschappen. AyA gebruikt dit biologische kader om de werking van moderne Large Language Models te duiden en te verdedigen, o.a. dat predictie het taalsysteem dringt om een intern model van de wereld te bouwen (predictie leidt tot representatie).
Sterk aan dit boek is dat AyA AI-modellen weghaalt uit de sfeer van magie of abstracte wiskunde en het plaatst in de context van de biologische evolutie. Minder sterk is dat de auteur soms last heeft van de Silicon Valley-bias en soms te grote filosofische sprongen maakt. De gelijkstelling van statistische predictie aan begrip, cognitie en bewustzijn is filosofisch zeer omstreden. Af en toe krijg je het gevoel dat complexe vraagstukken over wat het betekent om mens te zijn, gereduceerd worden tot optimalisatie-problemen.
November 12, 2025
Blaise argues that "we only began to see general intelligence when we stopped training models to do specific tasks with supervised learning, and switched to the unsupervised regime—but this isn’t (yet) a widely held view.". I agree with 80% of what he's saying and the 20% are mainly about the future outcomes of his observations as a "Google insider". But then those (80 + 20) futures could be wildly different!
It really is all cybernetics (from the mid of the last centuary) or the cybernetics of cybernetics (2nd or 3rd order control of control - married to connectionism or neural networks). Operationalising a self or even a new form of consciousness or just linear memory (say) may be enough or the step too far.
Probably not a best intro to the discussion as it's more of an idiosyncratic view (so 3 to 4 points) but I've given it a 5 because of how much it made me consider and reconsider my own take on the most important development of our time (and at 60 I've lived and contributed to the precursor computing and communication revolutions and the positives and negatives there so I don't this lightly).
This review is for the online version which has some great graphics (though scrolling through them is a bit haphazard at times).
It really is all cybernetics (from the mid of the last centuary) or the cybernetics of cybernetics (2nd or 3rd order control of control - married to connectionism or neural networks). Operationalising a self or even a new form of consciousness or just linear memory (say) may be enough or the step too far.
Probably not a best intro to the discussion as it's more of an idiosyncratic view (so 3 to 4 points) but I've given it a 5 because of how much it made me consider and reconsider my own take on the most important development of our time (and at 60 I've lived and contributed to the precursor computing and communication revolutions and the positives and negatives there so I don't this lightly).
This review is for the online version which has some great graphics (though scrolling through them is a bit haphazard at times).
November 10, 2025
"Best Book Ever."
(Short 4 words version, including period)
Longer one, What is Intelligence? is just one great connection after the other mixing #AI #Biology #Neuroscience #Science #Technology #Computation #Evolution #Intelligence #Past #Present #Future and more in a unique and clever way, didactic but deep at the same time, updated and grounded, if you are alive, you should read it!
It is true that it helps if you are in one of the fields related, for example I already knew about 80% of what it covers, since I expend most of my time over the last decades researching and building AGIs and neurotechs, but even so, it is organized briliantly that engages you at every parragraph.
(Short 4 words version, including period)
Longer one, What is Intelligence? is just one great connection after the other mixing #AI #Biology #Neuroscience #Science #Technology #Computation #Evolution #Intelligence #Past #Present #Future and more in a unique and clever way, didactic but deep at the same time, updated and grounded, if you are alive, you should read it!
It is true that it helps if you are in one of the fields related, for example I already knew about 80% of what it covers, since I expend most of my time over the last decades researching and building AGIs and neurotechs, but even so, it is organized briliantly that engages you at every parragraph.
January 4, 2026
It is a challenging book, but also a very rewarding and thought-provoking one. I found the first part, on the biology of life and the idea of life as a computational problem, the most demanding, probably because I am not a biologist. The second part, which traces the evolution of AI, felt more familiar and easier to follow given my background in the field. I especially appreciated the discussion on the social evolution of humans, which I found both insightful and stimulating. I think the book could have been shorter. the last chapter is very philosophical and hard to follow.
December 31, 2025
What is Intelligence? (2025) repositions AI not as a looming alien mind, but as a natural continuation of life’s long, messy story of evolution, cooperation, and prediction. It weaves together bacteria, brains, cities, and neural networks to show how intelligence emerges wherever systems learn to model themselves and their world. It takes us through the past, present, and future of AI, while describing our place in it.
This entire review has been hidden because of spoilers.
December 6, 2025
This book eads like someone tried to rewrite the whole philosophy of mind syllabus after spending too much time inside Google Colab. It throws everything into one giant blender. The core idea that life and mind are basically predictive computation is cool, but it often feels buried under world-building and metaphors, so you keep waiting for the clean argument that never fully lands.
September 20, 2025
Full text available online at whatisintelligence DOT antikythera DOT org
November 14, 2025
I absolutely love this man's mind and how he writes. I can honestly say some of it goes above my head, but he does such a good job at making complexity accessible. Amazing mind.
December 30, 2025
Compelling look at abiogenesis, symbiogenesis, and emergent intelligence in evolution. Insights from AI research add complexity— and indeed attempt to explain complexity!
December 31, 2025
Intelligence is about predictions based on experiences.
January 6, 2026
Solid book. Explains the ways that the brain and compares this to how artificial intelligence works.
Listened via Blinkist at 2.4x speed. 35mins long.
Listened via Blinkist at 2.4x speed. 35mins long.
Displaying 1 - 15 of 15 reviews