Ray Kurzweil is an acclaimed artificial intelligence (AI) innovator and entrepreneur known for developing technologies like optical character recognition, speech synthesis and recognition. He wrote this book as a sequel to The Age of Intelligent Machines (1990) which received great interest. Standing on the success of the first one, in this book Kurzweil sets out to extend his predictions into the next century, to claim that machine intelligence will surpass human intelligence and human species will merge with machine intelligence to evolve into post-biological "spiritual machines", freed from the limitations of its "biological hardware."
The book consists of three parts, made of twelve chapters. Each chapter is appended with casual conversations of the author with a fictitous reader called Molly. Part 1, titled "Probing the Past", opens with a chapter in which Kurzweil proposes The Law of Accelerating Returns. With this law, he generalizes the concept of Moore's law to claim that the exponential growth of computational power will continue beyond Moore's Law, through paradigm shifts in computation. In Chapter 2, he points out that if evolution, "a process which behaves only a quantum smarter than pure chance", is able to create human intelligence, then inevitably we will also create entities more intelligent than ourselves. The next chapter summarizes the views of different schools of thought on consciousness and implicitly hints that consciousness is independent of the computational substrate. In the last two chapters of this part, Kurzweil reviews the accomplishments of AI to conclude that although it had come a long way, it has not been and will never be perceived as successful since it is seen as "the study of how to make computers do things at which, at the moment, people are better." He suggests that intelligence can be created through the use of computational brute force and simple computational paradigms such as recursion, neural networks and evolutionary algorithms tied via knowledge and context.
Kurzweil starts Part 2, titled "Preparing the Present", by estimating the computational power needed to simulate a human brain which, he predicts, will be available for $1000 by 2025. Then he presents a compilation of technologies, such as optical, DNA and quantum computing, that may fuel the growth of computational power beyond Moore's Law. Armed with the computational power, Kurzweil moves on to suggest that reverse engineering of the brain will provide us the "software" for intelligence. He claims that the speed and resolution of brain scanning technologies, such as magnetic resonance imaging (MRI), will improve to make a full brain scan up to the synaptic level possible. The information can then be used to study computational properties of the brain or to simulate the brain on a different computational substrate. The next chapter is devoted to the technologies related to physical and virtual bodies for the twenty-first century. In his discussion, however, Kurzweil surprises his readers by skipping over the robotics technology, and emphasizing nanotechnology as the source of enhancements to our bodies and the world that we interact. Then, to free human intelligence from the physical limitations of the world, the author envisions different approaches of virtual reality ranging from suits creating a virtual tactile environment to spinal implants that tap directly into the nervous system for ultimate virtual reality experience. In the last chapter of this part, titled "1999", Kurzweil discusses the nature and neural correlates of spiritual experiences and how activities such as humour and art, at a certain extent, can be created by today's computers.
Kurzweil uses Part 3, titled "To Face The Future", to chronologically depict the advances in the technology and their effect on our lives in the next century. In this part, he relies more on his conversations with Molly who reports from the future. To summarize some of the predictions: By 2009, intelligent clothes linked through a wireless "body area network", monitor and assist people in their daily lives. By 2019, people begin to have relationships with virtual personalities, virtual artists emerge creating arts in collaboration with their human collegues. Virtual tactile simulations become possible allowing people to engage in activities like virtual sex. By 2029, machines surpass the computational capabilities of humans and start claiming to be conscious and have human-like spiritual experiences. Neural implants improving cognitive abilities of humans become common. Machines pass Turing test and start having long-term relationships with people. Spinal implants take over the virtual reality experiences making interactions with virtual personalities more natural. And by 2099, machines and humans merge into a single entity with almost all interactions being virtual.
The idea of human intelligence moving onto non-biological mediums and merging with machine intelligence, also proposed by Hans Moravec in Robot: Mere Machine to Transcendent Mind (1998), is exciting and the book succeeds in conveying it to its reader in a clear way. Through his conversations with Molly, Kurzweil also complements the technical aspects with discussions on the philosophical and social ramifications of the idea. However, the book is far from delivering a coherent and well-argued case for its predictions, making it irksome to read. Kurzweil seems to put together hypes surrounding the new technologies failing to melt them in a coherent vision. At the heart of Kurzweil's predictions lies the implicit assumption that different technologies, which are related to the future he envisions, will advance at the pace of computational power growth. I am not convinced by Kurzweil's arguments that this will be the case. For instance, it has been more than 50 years since McCulloch & Pitts, yet the computational view of brain that has emerged so far, is no different than the views of six blind men about the elephant in the well-known Indian tale. Neural interfaces, first hypothesized in the early 1960's, are still under experimentation and their use within the context of rehabilitation is still primitive. Grey Walter's Machina Speculatrix (1950's), mobile robots controlled by a two vacuum tube analog computer, were able to generate avoidance/approach, docking behaviors and learn; topics which are still considered active research topics in autonomous robotics research. In conclusion, the book seems to address non-specialist readers, and will not satisfy researchers who look for a fresh vision with well-founded arguments.