How to Create a Mind by Ray Kurzweil - A review

Ray Kurzweil has written a masterpiece combining hard cold facts about biological structure of brain and its functioning with his own theories for everything that is still to be discovered. He makes a convincing case for advent of human like intelligent machine by 2030. It is natural to not want to believe all his predictions because that would mean accepting defeat from our own creation.

Ray explains that brain is made of 300 million identical processing units, each with several inputs and one output. The processing unit, when triggered evaluates a combined score of all its inputs and if the combined score is higher than the threshold for output, the output is triggered. Each input has a certain associated weight. The strength of each input combined with its weight is used for calculating the combined score. The output of most of the 300 million processing units become inputs for one or more of other processing units. Ray calls these processing units as pattern recognizers with the input output connections making a hierarchical structure. Pattern recognition is a general word for any task that the brain needs to do: reading, conversation seeing or whatever else. The book postulates that everything we know and do is stored as patterns in these processing units, organized in a hierarchical manner. When we hear something, the chain of patterns starting from the smallest component of that sound are recognized by the processing units, feeding into the higher level combinations of sounds( words), which in turn feed into still higher level pattern(sentence) and so on to meaning of what was said. All these patterns are learned by the brain over time. In other words learning is the process of programming the 300 million processing units and building the hierarchical relationships. 

Ray claims that with IBM Watson and Apples Siri, we are starting to build computers that operate in a similar manner. These are generic information processors that learn and than use their knowledge to answer questions. They are not programmed with all of their knowledge but learn a large amount of their knowledge just like humans do. With these fundamentals in place, Ray addresses a host of questions about consciousness, creativity, feelings, innovation and ultimately the singularity of man and machine.

It was certainly a very informative experience for me to listen to this book. My brain certainly stored a lot more patterns in the process. I cannot say that I buy all the theories that are presented in the book, but they sound plausible and force you to think about ways to contradict or support them. I would recommend this to everyone. Anyone who has experience in the field of computing and is moderately curious about the brain will enjoy the book for sure.