Just finished listening to an audio version of "Life 3.0: Being Human in the Age of Artificial Intelligence," a new book by Max Tegmark. His "My Mathematical Universe" is one of my favorites, so I was really looking forward to his new book. And he didn't disappoint. This is a gripping text for anyone interested in AI and the future of life on our planet and beyond. Without a doubt, this is the most important conversation of our times. If you fail to see why it is so important, consider this. Most AI experts and industry insiders predict that artificial general intelligence (human-level and above) will become a reality either in a couple of decades or within this century. If someone told you that aliens are heading toward our planet and will land in some 20 years, I guess a lot of people, I mean really a lot of people, would start to freak out. Politicians, the academe, etc. would start thinking hard what to do about it. We are in a similar situation with AI research. Considering the incredible progress made in the field, it is time to think how to make sure AGI, if/when created, is beneficial to humanity.
If you are wondering about the title of the book, Tegmark posits that Life 1.0 is life where both the hardware and software are evolved rather than designed (for example, bacteria). Humans, on the other hand, are examples of "Life 2.0": life whose hardware is evolved, but whose software is largely designed. By software Tegmark means all the algorithms and knowledge that we use to process the information from our senses and decide what to do-everything from the ability to recognize your friends when you see them to your ability to walk, read, write, calculate, sing and tell jokes.
Finally, Life 3.0 designs and upgrades both its software and hardware.
In summary he divides the development of life into three stages, distinguished by life's ability to design itself:
Life 1.0 (biological stage): evolves its hardware and software
Life 2.0 (cultural stage): evolves its hardware, designs much of its software
Life 3.0 (technological stage): designs its hardware and software.
In fact, considering the current progress with prosthetics, cochlear devices, etc., he says, we are more like Life 2.1 already. But he thinks it is more likely that AGI will be created faster than any cyborg-style mind uploading becomes a reality (for description of that vision, see an interesting book "The Age of Em" by by Robin Hanson).
Tegmark begins his book with a pretty realistic scenario of a superintelligence break-out. Reads like a sci-fi thriller but is in fact much better than what Hollywood has come up with so far on this subject.
He then explores the current state of research into machine learning and some breakthroughs in the field. Then he tries to imagine the near and more distant future. It takes a physicist to write a compelling vision how far life can progress if limited only by the laws of physics. This part of the book is truly mind-boggling even if most of it can hardly be achievable due to various limitations and possible cosmocalypses (also described by the author).
Then he explores the subject of consciousness. Many people view AGI as our descendants. Even if they choose to eliminate us, they will live on and continue the story of life in our part of the observable Universe. Well, what if what we create are zombies without any consciousness Tegmark discusses what consciousness could be, briefly considers Integrated information theory as a viable explanation of the phenomenon, but also enumerates the most common criticisms of the theory. "Consciousness is the way information feels when being processed in certain ways," he summarizes his own view of consciousness and speculates that it must be substrate-independent, similarly to remembering, computing and learning.
He finishes his book optimistically, describing the work he does at the Future of Life Institute he has founded. Just like his previous book, "Life 3.0" is a brilliant example of existential hope, something humanity really needs. If you want to read an enthrolling, captivating book on AI, choose this one. Not that it needs any promotion after it was praised by Elon Musk :).
Follow it up by Nick Bostrom's voluminous "Superintelligence: Paths, Dangers, Strategies," another great text on the subject.