I very much enjoyed this book, but, from the start, I was not sure whether it was a work of serious science or one of science fiction. By the end, I did not much care. If I was forced to classify it, I would say that it presents a set of plausible scenarios for the future--speculative science that could form the technical foundation for science fiction worlds.
This is Moravec’s second book, and therein lies his motivation. In his first book, Mind children , he predicted that by now we would have all sorts of “mobile utility robots to help us around the house.” That has not happened. So Moravec has written this book because he still believes that those robots are coming, and he wants his vision and his justification for his vision on the record to help create them. Even with this second chance, Moravec is doomed to continue to suffer the prophet’s fate: to overestimate changes in the near term, and to underestimate them in the long run.
The first three chapters are a prelude. Moravec gives a history of a few innovations, both technical and social, and some details of recent experiments with “perceiving” robots. He is not interested in the massive machines that do tough work in car factories or fine repetitive tasks under numerical control. Moravec is an intellectual descendant of Karel Capek, the Czech playwright who invented the term “robot” in his 1921 play Rossum’s Universal Robots (also R.U.R.); Moravec is writing about androids, in mind if not in body.
The concrete predictions start in chapter 4. By 2010, Moravec predicts that we will have robots with the intelligence of lizards, driven by computers capable of 3 kiloMIPS. These first-generation robots “will perform their job but also react to obstacles, missing supplies, little accidents. Within tightly circumscribed boundaries, they will seem to have genuine awareness [but] outside of that range, the impression will collapse, as their program repertoire runs out and they stop with an error code.”
The second generation will arrive about ten years later, with processing power of about 100 kiloMIPS and the intelligence of mice. “Second generation robots will learn to deal with circumstances not explicitly programmed. They can be trained or traumatized. They will have a behavioral character.” Moravec predicts that some of these robots will become pets. (Interestingly, on June 1, 1999, Sony introduced AIBO, a dog-like machine offered as the first pet robot.)
By 2030, Moravec predicts, we will have a third generation of robots driven by 3 megaMIPS of power, with the intellectual capabilities of monkeys. These robots will be able to simulate the real world at speeds slightly faster than human perception. This will give them the appearance of consciousness. They will have fully formed physical and psychological presence. They will be able to hold conversations but will be childlike, naive, and simple-minded.
The fourth-generation robots, arriving around the middle of the 21st century, will have human-scale intelligence driven by a 100-megaMIPS processor. These will be true reasoning machines that are fully capable of generalizing and distinguishing subtleties. These machines will have mental and physical abilities far beyond those of humans.
Moravec is a technologist, not an economist or sociologist, but that does not stop him from speculating on the societal impact of these robots. He predicts the end of scarcity, with all the changes that will create. People will not work anymore because all their needs will be met by robots. The service sector will decline. “Money will accumulate in industries, enriching just the few people still associated with them.” In the extreme, Moravec predicts an “evaporation of ownership [that] will end capitalism, but capital enterprises [operated by robots] will thrive as never before.”
By 2100, Moravec predicts a world that pushes out into space, inhabited by “Exes,” ex-companies and ex-humans that have moved beyond the physical, biological, and social boundaries that we now take for granted. “Exes will propagate less by reproduction than reconstruction, meeting the future with continuous self-improvements.” Moravec envisions exes creating new sentient organizational models, some like animated bushes, others like starfish, all with a fractal-like repetitiveness and scaling.
While all of this speculation is interesting and fun to think about, I felt unfulfilled in two areas. First, Moravec builds on our current models of electronic and mechanical concepts. That is reasonable because of his background, but much of what he forecasts might come first from the biological world, from genetic engineering and cloning. The robots of the future may be more like Mary Shelley’s Frankenstein’s monster than like Capek’s machines.
Second, I wish that Moravec had linked his thoughts to those of some outstanding science fiction writers who have given a lot of energy to considering the roles of robots and androids in the future. Moravec mentions Isaac Asimov’s famous Laws of Robotics, but just in passing, while Asimov himself devoted thoughtful volumes to how humans and robots might create a society together. More recently, the television series Star Trek: The Next Generation used the android character Data to explore with great sensitivity what it means to be human--and what it means to be artificial.
Moravec is doomed to frustration. Hopefully he will live long enough to see his specific predictions fail while, at the same time, falling far short of what really happens. Humans seem to need to create creatures in their own image, and we are quickly approaching the time when a variety of technologies will be available to help us do that. As with all innovation, there will be unimagined consequences, positive and negative. It sounds like great fun. I look forward to it eagerly.