Summary of Ray Kurzweil’s: The Age of Spiritual Machines: When Computers Exceed Human Intelligence. I read this book in 1999 and it changed my thinking about the future.
The Inexorable Emergence
Sometime in the next 100 years machines will surpass human intelligence. Computers today already surpass humans in activities like playing chess, diagnosing some medical conditions, buying and selling stocks, and guiding missiles. Still human intelligence is far more flexible than machine intelligence. Computers can’t describe objects on a table, write a term paper, tie shoes, distinguish a dog and a cat or appreciate humor. One reason for this is that computers are simpler than the human brain, about a million times simpler. But this difference will go away. Computers are doubling in speed every twelve months, and they will achieve the memory capacity and computing speed of the human brain around 2020.
Still this won’t allow computers to match the flexibility of human intelligence because the software of intelligence is as important as the hardware. One way to mirror the brain’s software is by reverse engineering—scanning a human brain (which will be achievable in this century) and copying its neural circuitry into a neural computer of sufficient capacity. We might be able to train a system of parallel neural nets to understand language, model knowledge, and read and understand written materials. Although today’s computers cannot do this their abilities are improving. Computers will be able to understand what they read in this century. Computers can then read all the world’s literature and gather knowledge on their own.
After computers reach a human level of intelligence, they will go beyond it. They already remember and process information better than we do. Computers process trillions of facts and search huge databases almost instantaneously, while we have a tough time with a few phone numbers. The combination of human-level intelligence and the speed, accuracy, and memory capabilities of computers will move computers beyond human intelligence. While our neurons are marvelous, they are hardly optimal. Most of their complexity supports of life processes, not computation and information analysis. Neurons are slow compared to electronic circuits which are millions of times faster. Once computers reach a human level of ability in understanding abstract concepts, recognizing patterns, and more, they will apply their abilities “to a knowledge base of all human-acquired knowledge.”
Most of us think of evolution as a billion-year drama that led to human intelligence. But the creation of greater than human intelligence will dispel the notion that we are the most intelligent creation. This realization will have profound effects on our thinking, lives, and selves. A variety of philosophical questions will be asked. Are computers thinking, or just calculating? Are humans thinking or just calculating? Are there any important differences between human thinking and machine thinking? Will we ever consider our machines conscious? If someone scans their brain through a “noninvasive scanning technology … and downloads their brain to their personal computer,” are the people who come out of the machine the same ones that entered?
The Law of Time and Chaos
In the first few moments of the universe important events and paradigm shifts happened very fast. As the universe aged, eons of time separated cosmologically significant events. So time, thought of as the interval between significant events, has been moving very slowly for most of cosmic evolution. But with cultural evolution the intervals between significant events has shortened; time is now moving exponentially faster. (What Kurzweil calls “the knee of the curve,” the spot at which exponential growth is about to take off, is basically the same phenomena that futurists call “the singularity.”) We can see this in the pace of cosmic evolution: almost 10 billions years until the earth’s formation; a few billion more for life to evolve, hundreds of millions of years till the emergence of primates, millions of years till the emergence of humanoids, the emergence of homo sapiens a mere ½ million years ago, modern humans just a few thousand years, agriculture just 10 thousand years ago, and the scientific revolution just a few hundred years ago. In short, time is speeding up by which Kurzweil means that the intervals between salient events are shrinking, i.e., there is an exponential curve which plots the nature of change.
Technology—fashioning and using tools ever more sophisticated tools—is “evolution by other means.” Just as DNA records the information of biology, technology records its own records in books, libraries, and computer databases.
(If you are interested in this idea, explore the concept of memes. Memes, like genes, preserve and transfer information only much faster because, while genes go from one body to another, memes spread from one brain to another. And you can spread a meme—an idea or belief for example—very quickly. I just spread the “meme” meme to all the people who read this. And the meme could be a meme for any idea: immortality, evolution, a god, patriotism, etc. Many contemporary thinkers believe that memes infect brains which then spread them; some think of people as memebots. For more see the work of Richard Dawkins, who introduced the term in his book, The Selfish Gene. In addition, there is an entire literature on memes.)
Technology expedites the process of evolution. Consider that homo sapiens sapiens appeared only 90,000 years ago, a mere moment in cosmic time, becoming the lone hominoids a mere 40,000 years ago. Still it took tens of thousands of years to figure out how to sharpen both ends of stones to make them effective! And the pace of technological change has accelerated remarkable since then. The 19th century saw technology increase dramatically fast compared to the 18th century and increased unbelievably fast compared to the 12th century! In the 20th century major technological shifts happened in decades or in some cases in a few years. A hundred years ago there was no heavier than air flight or radio. When I was born there were no wireless phones and not many color TVs. When many of my readers were born there were no cell phones or internet.
Technology has enabled our species to dominate the earth, exercise control over our environment and survive. A most noteworthy technological innovation has been computation “the ability to remember and solve problems…” The machines created by our technology do more than just increase our strength, they solve a vast array of problems. Moreover increases in computation is governed by Moore’s law: every two years or so the surface area of a transistor is reduced by 50%, thus you can put twice as many transistors on an integrated circuit. Thus every two years you get twice the computer power for the same cost. This trend should continue for another fifteen years or so after which it will break down when transistor insulators will be but a few atoms wide. (Then we will need quantum computing.) Kurzweil makes the point—and backs it up with statistics, pictures, and graphs—that this trend of doubling of computing power goes back to the beginning of the 20th century. To really understand what will happen in the 21st century and beyond we need to look closely at the exponential growth of technology that will bring about vast changes in the near future.
Crucial to Kurzweil’s argument is what he calls “the law of time and chaos.” Consider that some processes start out fast and then slow down—salient events in cosmic evolution or in the biological development of an organism—while others start slow and then speed up—the evolution of life forms or technology. Why is this? The answer is that the law of time and chaos shows a relationship between the speed of time and chaos. If there is a lot of chaos or disorder in a system, the time between salient events is great and thus time can be said to slow down. As the chaos decreases and order increases, the time between salient events gets smaller and time speeds up. The “law of increasing chaos” denotes the former while the “law of accelerating returns” denotes the latter. The law of accelerating returns is what most interests Kurzweil. While the universe increases in disorder or entropy, evolution leads to increasing order (information for the purpose of survival) and complexity. Technology is evolution by means other than biology. (It is part of cultural evolution which, unlike biological evolution, is now moving extraordinarily fast.) Technological evolution speeds up because “it builds on its own increasing order.” In other words, technological evolution, like other forms of evolution, “builds on itself.” What this all leads to is the following:
a) evolution builds on itself, thus
b) in an evolutionary process order increases exponentially, thus
c) time speeds up exponentially, thus
d) the returns accelerate.
This law of accelerating returns is what drives cultural and technological evolution forward. And the rate of these returns, which then build on themselves to create higher returns, is growing exponentially. Thus technology is most significant as it provided the means to store the information from previous brains. What this all means is that change is increasing exponentially—the near future will be radically different than the present.
Building New Brains
The Hardware of Intelligence – To build intelligent machines we need: 1) formulas (recursive search, self-organizing networks, and evolutionary improvement); 2) knowledge (for a process to achieve results); and 3) computation. The human brain is strong because of its parallel processing but weak in terms of its slow computational speed.
For this reason, DNA-based evolution will eventually have to be abandoned. DNA-based evolution is good at tinkering with and extending its designs, but it is unable to scrap an entire design and start over. Organisms created through DNA-based evolution are stuck with an extremely plodding type of circuitry …
But the Law of Accelerating Returns tells us that evolution will not remain stuck at a dead end for very long. And indeed, evolution has found a way around the computational limitations of neural circuitry. Cleverly, it has created organisms that in turn invented a computational technology a million times faster than carbon-based neurons (which are continuing to get faster). Ultimately, the computing conducted on extremely slow mammalian neural circuits will be ported to a far more versatile and speedier electronic (and photonic) equivalent.” (101-02)
Achieving the Hardware Capacity of the Human Brain – Kurzweil argues that computers will equal the computing speed of a human brain by about 2020. The memory capacity of the human brain should also be equaled at about that time. “Taking all of this into consideration, it is reasonable to estimate that a $1,000 personal computer will match the computing speed and capacity of the human brain by around the year 2020…” (105) A single personal computer will have the computing speed and memory capacity of a small village by 2030, the population of the US by 2048, and a trillion human brains by 2060. “If we estimate the human Earth population at 10 billion persons, one penny’s worth of computing circa 2099 will have a billion times greater computing capacity than all humans on Earth.” (105)
Computing Substrates of the 21st Century – Kurzweil knows that Moore’s law cannot hold forever: “For the immediate future, Moore’s Law will continue with ever smaller component geometries packing greater numbers of yet faster transistors on each chip. But as circuit dimensions reach near atomic sizes, undesirable quantum effects such as unwanted electron tunneling will produce unreliable results.” (106) He offers a number of scenarios to deal with this issue:
1) 3 dimensional computing – “Already, venture-backed companies … are competing to build chips with dozens and ultimately thousands of layers of circuitry.”106
2) computing with light – “Optical computing uses streams of photons (particles of light) rather than electrons. A laser can produce billions of coherent streams of photons, with each stream performing its own independent series of calculations.” (106)
3) computing with the machinery of life –
A new field called molecular computing has sprung up to harness the DNA molecule itself as a practical computing device. DNA is nature’s own nanoengineered computer and it is well suited for solving combinatorial problems. … Applying actual DNA to practical computing applications got its start when Leonard Adleman, a University of Southern California mathematician, coaxed a test tube full of DNA molecules …to solve the well-known “traveling salesperson” problem. …. It is an ideal problem for a recursive algorithm, although if the number of cities is too large, even a very fast recursive search will take far too long. Professor Adleman and other scientists in the molecular-computing field have identified a set of enzyme reactions that corresponds to the logical and arithmetic operations needed to solve a variety of computing problems. Although DNA molecular operations produce occasional errors, the number of DNA strands being used is so large that any molecular errors become statistically insignificant. Thus, despite the inherent error rate in DNA’s computing and copying processes, a DNA computer can be highly reliable if properly designed. (107)
There are other possibilities including a) the brain in the crystal, b) the nanotube, and c) quantum computing. Of all the technological possibilities to replace digital computing this is the one with the most promise and the one K discusses most. Unlike digital computing which relies on bits of info which are either on or off:
Quantum computing, … is based on qu-bits … which essentially are zero and one at the same time. The qu-bit is based on the fundamental ambiguity inherent in quantum mechanics. The position, momentum, or other state of a fundamental particle remains “ambiguous” until a process of disambiguation causes that particle to “decide” where it is, where it has been, and what properties it has …In a quantum computer, the qu-bits would be represented by a property–nuclear spin is a popular choice–of individual electrons. If set up in the proper way, the electrons will not have decided the direction of their nuclear spin (up or down) and thus will be in both states at the same time. The process of conscious observation of the electrons’ spin states–or any subsequent phenomena dependent on a determination of these states–causes the ambiguity to be resolved. …The key to the quantum computer is that we would present it with a problem, along with a way to test the answer. We would set up the quantum decoherence of the qu-bits in such a way that only an answer that passes the test survives the decoherence. The failing answers essentially cancel each other out. (110-11)
The next few pages argue that quantum computing is feasible, it difficulties can be overcome, and it will provide encryption that cannot be broken. Most importantly, he gainsays the arguments of those who say conscious machines are impossible. This all brings us to the main point of the chapter: how we can and will reverse engineer the human brain. Kurzweil suggests that we might begin the process of understanding how to reverse engineer a human brain by “freezing a recently deceased one.” (121) We could examine one layer at a time to scan every neural connection. We would proceed a layer at a time until we have a three dimensional model of the brain. Even better we could examine living brains of persons who are about to die and consent to the procedure. And with better noninvasive scanning technologies we will eventually be able to scan living brains completely, the way we scan our bodies with MRIs.
Ultimately, we would want to map the entire brain, including its synapses, in fact mapping it synapse by synapse, a capability that will result from the continual increase in scanning ability. After scanning the brain we should be in a position to understand it, at least its overall pattern, thus slowing unpeeling the onion. And the next step is downloading our minds into computers. After understanding the important algorithms of each brain region we should in theory be able to accomplish this. We don’t need complete understanding but we do need the ability to copy a brain’s pattern. The technology will be bad at first but should improve gradually.
And what will we find when we do this? To objective observers the new “person” will appear to be identical with the person originally scanned. From the subjective point of view this issue is more difficult and arguments can be made on both sides. If we are basically the pattern of our consciousness, then it seems the new person is the same as the original. If our essence is our body, then the person doesn’t survive. But for the copied and transferred person across the divide, she will think she is the original person and won’t believe she committed suicide by surviving in this way. She might wonder if she is the same person after the upload, but she’ll still be glad she still exists. Kurzweil believes that in the second half of the 21st century the process of taking the leap will begin.
“Initially there will be partial porting—replacing aging memory circuits, extending pattern-recognition and reasoning circuits through neural implants. Ultimately, and well before the 21st century is complete, people will port their entire mind files to the new thinking technology.” (126) While nostalgic for our carbon-based bodies, we will quickly get over it as we find that porting ourselves allows us to extend our minds. “
Remember that $1000 of computing in 2060 will have the computational capacity of a trillion human brains. So we might as well multiply memory a trillion fold, greatly extend recognition and reasoning abilities, and plug ourselves into the pervasive wireless-communications network. And while we are at it, we can add all human knowledge—as a readily accessible internal database as well as already processed and learned knowledge using the human type of distributed understanding. (126-28)
And this means that if you use this technology your software is no longer dependent on your hardware. And your evolving mind file won’t be stuck with the circuitry of the brain but can evolve well beyond that. Your mind file can be transferred from one medium to another just as files are transferred from one computer to another. And “our immortality will be a matter of being sufficiently careful to make frequent backups. If we’re careless about this, we’ll have to load an old backup copy and be doomed to repeat our recent past.” 129
Building New Bodies
What about building new bodies? Will we want to download our personal evolving mind files into our original bodies, upgraded bodies, nanoengineered bodies, or virtual bodies? Kurzweil claims that body and brain will likely evolve and be enhanced together—gradually. We are already further along with body transformation than with brain transformation with titanium devices, artificial skin and heart values, pacemakers, etc. So we might want to re-build our bodies completely. Of course most of us are attached to the warmth and softness of our bodies, so we might start by enhancing our bodies cell by cell using genetic therapies of some kind. But we can only go so far with this, and that’s because of the limitations of DNA-based cells that depend on protein synthesis. (They are stable under narrow temperature and pressure ranges, sensitive to radiation, etc.) Kurzweil agrees with Moravec that such bodies—however well they were enhanced—would just be second-rate robots.
Instead lets use nanotechnology to rebuild the world, atom by atom. While our machines of today are marvelous and precise compared to stone blades, they are primitive when viewed from the atomic level; our mechanisms:
moved atoms in great thundering statistical herds. … Nanotechnology is technology is technology built at the atomic level: building machines one atom at a time. … We have … proof of the feasibility of nanotechnology: life on Earth. Little machines in our cells called ribosomes build organisms such as humans one molecule at a time … Life on Earth has mastered the ultimate goal of nanotechnology, which is self-replication. (137)
The molecular building blocks of life are of limited use. And just as human-created computation will exceed natural computation, 21st century physical technology will exceed the capabilities of amino acid based nanotechnology of nature.
The holy grail of nanotechnology is that nanomachines are intelligent, replicative, and able to manipulate things at the nano level. Important theorist like Eric Drexler and Ralph Merkle have shown the feasibility of self-replicating nanobots. The possibilities for nanotechnology to transform the world are endless: they could build inexpensive solar cells to replace fossil fuels; be launched in our bloodstream to improve the immune system, destroy pathogens, cancer cells, rebuild diseased organs (the inspiration for the cryonics movement. (Ralph Merkle, formerly of XeroxPark and inventor of encryption technology, told me in personal correspondence that if nanotechnology works, cryonics almost certainly will work.) More possibilities include: reconstruction of bodily organs and systems, reverse engineering of human neurons and any cell in the human body. Moreover
Food, clothing, diamond rings, buildings could all assemble themselves molecule by molecule. Any sort of product could be instantly created when and where we need it. Indeed the world could continually reassemble itself to meet our changing needs, desires, and fantasies. … NT will permit objects such as furniture, buildings, clothing, even people to change their appearance and other characteristics—essentially to change into something else—in a split second. (140)
Will we want to do this?
There is a clear incentive to go down this path. Given a choice, people will prefer to keep their bones from crumbling, their skin supple, their life systems strong and vital. Improving our lives through neural implants on the mental level, and NT enhance bodies on the physical level , will be popular and compelling. It is another one of those slipper slopes—there is no obvious place to stop this progression until the human race has largely replaced the brains and bodies that evolution first provided. (141)
Kurzweil admits there are dangers from nanotechnology, self-replication run amok and the intentional hostile use of the technology. But the rewards may be worth the risks.
Of course we don’t need real bodies at all. In a virtual reality a virtual body would do fine. Kurzweil traces the development of virtual reality from the primitive computer games of the 1960s to the state of the art virtual today (1999). But in the next decade faster computers will continue to make virtual reality more realistic and inviting. Later in the 21st century we won’t need to enter a booth to experience virtual reality. “Your neural implants will provide the simulated sensory inputs of the virtual environment—and your virtual body—directly in your brain.” (144)
And that’s not all. By the late 21st century “the real world will take on many of the characteristics of the VW through the means of NT “swarms.” (145) Intelligent nanobots can merge and create something called Utility Fog. Spaces filled with utility fog go unnoticed but are able to “simulate any environment by creating all sorts of structures.” A Fog environment can be anything you want. Utility fog creates a virtual reality in the real physical world. Other minds can be simulated in the utility fog as “Fog people.” Furthermore, “there are a variety of proposals for nanotech swarms, in which the real environment is constructed form the interacting multitudes of nanomachines.” (145) What this all leads to is a situation in the last 21st century when “we will have to select our body, our personality, [and] our environment …” (146)
The next section deals with virtual sex, sexbots (sexual robots), neural implant sex, and the great sex inside utility fog. Kurzweil then delves into spiritual experiences which we already know have a physiological basis. When we completely understand their neurological correlates we can have them at will, if that is what we want. (Instead of sex I suppose.) The well-known work of neuroscientists from UCSD who have found “what they call the God module, a tiny locus of nerve cells in the frontal lobes that appears to be activated during religious experiences.” (152) This “neurological basis for spiritual experience has long been postulated by evolutionary biologists because of the social utility of religious belief.” (153) At any rate, a full understanding of this research would allow us to augment it. “Twenty-first century machines … will do as their human progenitors have done …[and] connect with their spiritual dimension.” 153
The Rest of the Universe Revisted
Kurzweil has argued that “the emergence of machine intelligence that exceeds human intelligence … is inevitable.” (253) This results from the Law of Accelerating Returns which also applies to the rest of the universe. But in what way?
To begin to answer this question Kurzweil speculates that life is both “rare and plentiful” in the universe. It is very rare compared to the immensity of the universe but very plentiful in an absolute sense since the universe is immense. Since intelligent life probably exists on other planets, Kurzweil gives us a sense of the kinds of thresholds they would go thru. He thinks the evolution of life forms could be thought of as one threshold, the evolution of intelligence as the next, then the evolution of technology, followed by the evolution of computation, followed by the merger of a species with the technology it has created. “At this stage … the computers are themselves based at least in part on the designs of the brains … of the species that originally created them and in turn the computers become embedded in and integrated into the species’ bodies and brains.” (255-56)
But there are many possible ways that this prediction might fail: nuclear war, self-replicating nanobots, and software viruses. Kurzweil guesses that we have “a better than even chance of making it through.” But he adds: “I have always been accused of being an optimist. (I don’t know how to figure the odds of us making it thru at all, much less thru to the age of spiritual machines. But I’m less optimistic than Kurzweil.)
Kurzweil concludes his book with a stirring and dramatic section concerning the relevance of intelligence for the universe. While the common view is that intelligence is impotent when compared to the mighty forces of the universe, Kurzweil disagrees. Intelligence does affect physical forces; our intelligence thwarts gravity and manipulates physical forces in other ways. Yet the density of intelligence is very small. If you imagine that intelligence will increase exponentially with time, then it will eventually become a worthy competitor for the big universal forces.
Thus he concludes: “The laws of physics are not repealed by intelligence, but they effectively evaporate in its presence.” What then of the fate of the universe? Kurzweil concludes: “the fate of the Universe is a decision yet to be made, one which we will intelligently consider when the time is right.”