“A 600 pages long book, a VP/Fellow at Google who has been instrumental in the development of autopredicting machines since the 1990s, might be a definitive take on AI, or at least where we’re at with it today. In short, it takes up the idea that prediction (which is AI’s superpower) is not only not a limiting factor to its attributed intelligence, but could be the core of intelligence in general, ours and the one we share with other living beings, rooted in the ubiquitous fact that life itself is an autopredicting machine.

From there on, 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, ripping apart in the making decades of false assumptions made by philosophers on the nature of intelligence and the alleged “hard problem of consciousness”.

I chose What is Intelligence? because, here again, albeit in a very different way than 2000 Meters to Andriivka, it is a reminder of how the partition between life and technology is an artificial construct, but in a good way that is. Moreover, it tends to let us believe that the faster we get conscious of it, the closer we might actually come to a breakaway from the growing darkness of our civilisation.” Mark Alizart

« Blaise Agüera y Arcas, computational scientist and head of Technology & Society at Google, in What Is Intelligence? Lessons from AI About Evolution, Computing, and Minds, proposes a unified vision of intelligence as a predictive faculty shared by all forms of life—from the human brain to artificial neural networks. Starting from the hypothesis of the “predictive brain,” the author argues that thinking means above all anticipating the future, and that the ability of machines to build predictive models reveals a profound continuity between biological processes and AI systems. Drawing on evolutionary biology, physics, neuroscience, and machine learning, Agüera y Arcas describes living beings as a “phase of self-modifying computational matter,” in which evolution itself appears as an algorithmic exploration of possible forms of life and mind. The book intertwines scientific and philosophical questions—from the computational nature of living systems to the social origin of intelligence, from the relationship between models and reality to the themes of entropy, time, and consciousness. Contrary to much of the literature on AI, the author goes so far as to attribute to some contemporary artificial systems an authentic form of intelligence, consciousness, and perhaps even will, with radical ethical and political consequences. Due to its theoretical breadth and clarity of exposition, What Is Intelligence? emerges as a point of reference in the debate on what it means to “think” today—and how the artificial intelligence revolution is forcing us to redefine the boundaries between human, living, and machine. » Emanuele Coccia

‘WHAT IS INTELLIGENCE?’

Selected both by Mark Alizart and Emanuele Coccia

Type: Book

Publisher: MIT Press