Ray Kurzweil:
"How to Create a Mind" (2012)

(Copyright © 2012 Piero Scaruffi | Terms of use )
A better title would have been "How to Embarrass Yourself" but Kurzweil knows that he can rely on a devoted Silicon Valley audience (more like a church) that will take seriously anything written in this book.

He was also not lucky that he published this book in 2012. Hinton is not mentioned at all, despite the fact that he was already one of the scientists revolutionizing A.I., and precisely in 2012 he and his students created the famous AlexNet that popularized "deep learning". Kurzweil never mentions Bengio, and never mentions LeCun, despite the fact that they had already laid the foundations for the new A.I. Kurzweil, still attached to the Watsons and Deep Blues of IBM, was hopelessly left behind by the very field that he wanted to champion. He spends several pages hailing the successes of IBM's Watson, which was so discredited later in the decade to be retired in 2020. He repeatedly hails Google's self-driving cars as if they were about to be commercially available within a few years. I checked again in 2022 and not a single self-driving car is available commercially ten years after Kurzweil was mesmerized by them.

Anyway, Kurzweil argues that there exists a general law, the "Law of Accelerating Returns", that applies to both biological and technological evolution. I wonder how many biologists and how many historians would agree. Historians would point out that we haven't returned to the Moon and that the only supersonic passenger plane (the Concorde) has been retired. But of course it all depends on what you use as your metric. I personally think that there was incredible progress between 1880 and 1920, when almost all of today's appliances were invented, plus car, airplane, radio... plus Quantum Theory and Relativity. But i am sure he studied history carefully and decided that progress is faster in our age (this is a sarcastic statement). He is also confident that "we are rapidly reverse-engineering the information processes that underlie biology, including that of our brains". That's not what i hear from neuroscientists, who think that we are far from understanding how the brain works. The progress has mostly been in tools that enable us to see what happens inside the brain, not in understanding exactly what that does. In fact, opinions change yearly on what happens inside the various parts of the brain (Kurzweil greatly simplifies the structure of the brain in this book, probably using a Wikipedia page).

As usual in all of his books, he congratulates himself for having made so many correct predictions (they were mostly obvious and shared by every intelligent beings and he failed in all the important ones) and makes two more that are notable: 1. Other intelligent species don't exist; and 2. The Singularity will happen in 2029. Time will tell.

When he is not congratulating himself, he takes a lot of detours, talking about Einstein and summarizing old theories of the brain and of computer science. He states with absolute certainty what the thalamus, the hippocampus, the cerebellum and so on do and then, after another digression into creativity and love, and another digression into experiments of brain simulation, and another digression into early electronic computers, and after another digression into the books on consciousness that were popular when he wrote the book, he concludes that it all boils down to a simple process of hierarchical pattern recognition.

This theory of how the brain works is a variation on the theory by Jeff Hawkins and Dileep George, published in several papers and then summarized in the book "On Intelligence" (2004), an idea that in 2005 gave rise to their startup Numenta. It is really difficult to say anything about a book that talks about brain and mind and reveals from the onset only a superficial knowledge of decades of neuroscience and philosophy of mind.

Again, most of the book is half a self-congratulatory essay on his many (self-selected) accomplishments and half a superficial and amateurish survey of popular books on philosophy of mind and popular books on neuroscience. There are only a few pages about his own theory, which he repeatedly claims to have "proven" but the "proof" is simply quoting those popular books which, back then, had advanced similar ideas.

The chapter "The Pattern Recognition Theory of Mind" is so amateurish that i could not finish it. It is based on hearsay (not on actually neuroscience) about the functioning of the brain and hardly mentions any of the discoveries of the last 20 years. There is no science at all: just a sort of granma's fairy tale about how neurons work. There is no discussion of what has to happen at the physical level for his pattern recognizer to actually work. What he is describing is simply an electronic machine that he has in mind for carrying out some pattern recognition tasks. That machine might well be useful for some tasks but it is pretty much like describing a wheelbarrow and claiming that it can be used to dig a tunnel under the Everest. He seems to think that all neurons are identical, there is only one type of neurotransmitter, that the nervous system ends at some specific point (the neck?), and that the endocrine system is just decoration.

After some vague statements about consciousness (if i understand correctly, Kurzweil simply thinks that a machine that does what i do is conscious and there's no need to prove it), he retells stories told before about how the future of technology is about to change the very nature of our lives. Yawn.

The book ends on another comic note when Kurzweil hails the "discovery" of particles that travel faster than light, a "discovery" (announced in 2011) that was later found to be caused by an improperly attached fiber-optic cable and an improperly calibrated clock oscillator.

Maybe it's his publisher who has to be derided. Maybe Kurzweil had no intention of using the title "How to Create a Mind" and the even more ridiculous subtitle "The Secret of Human Thought Revealed". This stuff belongs to the science of marketing toothpastes and canned soups.

See also: Kurzweil, Ray: "The Singularity is Near" (2005)
Kurzweil, Ray: "The Age of Intelligent Machines" (1990)
Kurzweil, Ray: "The Age of Spiritual Machines" (1999)