How Does Kurzweil Define Singularity?
Ray Kurzweil defines singularity as the ability of a computer to pretend to be human and win the Turing Test. But, no... Singularity emerges, when a computer overtakes the highest human competence - the ability to recognize patterns in complex streams of data. Such singularity arrived with the implementation of the DeepQA/UIMA programs, which power the Watson supercomputer. The world saw Watson comfortably beat two human champions in the popular US game show called Jeopardy! That was a success in a trivial field. But, with the same technology, computers will outrun humans in medicine, robotics, or power transmission and distribution. Not the Turing Test, but pattern recognition by a computer indicates the real arrival of singularity.
What Is Singularity?
Watson has amazing capabilities. Within a few seconds, the machine can peruse a million books using problem-solving skills similar to humans. It also displays a multifaceted perception of knowledge, similar to human “understanding.” Its ability, to carry numerous parallel solutions in its processing memory, equals the “awareness” of the human mind of many possible answers to any presented question. Its analytical skills are comparable to human abilities.
defines singularity as the ability of a computer to win the Turing
Test, while pretending to be a human. But a human cannot win the Test
pretending to be a computer. Singularity is the point in time, when a
computer beats a human in pattern recognition skills, which provide
the foundation for human intelligence. It already beats a human in
computational skills. Pattern recognition skills can ultimately
evaluate knowledge in any field to solve problems.
With more knowledge, computers can solve more complex problems. With skills similar to humans, the level of knowledge becomes the pivotal measure of intelligence. This machine can access more knowledge in real time than any human being on earth, making it a superhuman intelligence (SHI). Steven Pinker believed that there was “not the slightest reason to believe in a coming singularity.” But, while the world is yet to take real notice, not Kurzweil singularity, but SHI, the real thing, has already arrived on the planet.
When Does A Mutation Change The Landscape?
It was not the exponential speed of technological change, which opened the door to singularity. Ray Kurzweil had postulated the inevitability of singularity through a law of accelerating returns due to changes, such as the application of Moore's law, or new developments in many new fields including nanotechnology and medical sciences. But, such escalating complexities in the development of technology need not have resulted in the evolution of an SHI.
History shows that paradigm shifts occur in evolutionary development only when a new and enabling mutation takes place. The appearance of the first prokaryotic single cells on earth led to the development of increasingly complex bacteria. But, rising complexities in single cell architecture had no impact on evolutionary development for nearly two billion years. The crucial mutation, which led to life, was the development of multicellular organisms. The cell architecture changed. That single fortuitous mutation led to the formation of eukaryotic cells, which ultimately developed into plants, fungi and animals.
Increased computing power, faster circuits, or added medical knowledge does not provide the key architecture for a powerful new intelligence. It was the brand new concept of UIMA (Unstructured Information Management Architecture), which suddenly gave computers the ability to recognize patterns in the mass of recorded human knowledge. It was a software mutation, a shift to pattern recognition, which led directly to the genesis of an SHI.
Is Speed Of Computation The Answer?
Intelligence increases with more knowledge. It is the dynamic process, which searches an available database and locates wise responses to solve problems in the environment. It increases only when the knowledge base expands. When an intelligent process can reach answers in real time, further increases in processing speed become irrelevant. If a calculator multiplies two numbers in 2 milliseconds, getting an answer in 1 millisecond is not much help. Only a computer, with better methods of analysis of more data can be a higher level intelligence.
While the speed of neural pattern recognition remained constant, evolutionary development improved intelligence by increasing the range of analyzable patterns. The early reptiles depended mostly on touch and smells for survival. The mammalian brains improved by additionally evaluating vision, taste, and sounds. Human intelligence resulted from more complex analysis of all such incoming data through an addition to the thickness of the cortical layer. Once the method of analysis was set, the size of the knowledge base decided the superiority of an intelligence.
What Is A Superhuman Intelligence?
The human competitors in Jeopardy had taken years to accumulate their extraordinary knowledge of trivial facts in history, literature, the arts, pop culture, science, sports, geography and wordplay. Most of that knowledge was available to IBM as 15TB of text data in a million selected books. Once assembled, it needed little time to feed that data into the Watson supercomputer. The machine evaluated this data to play the game using amazing skills. It parsed incoming questions to get their logical meanings. All the possible meanings of each word, as well as key words and phrases, in the question were listed. It then generated all the possible questions, which could be implied in this question.
The machine then followed the same analytical procedure for references to persons, organizations, events, times, concepts and opinions in its million book archive to assemble hundreds of reasonable answers. It then rechecked the same immense library again for evidence supporting these results, generating more links to suitable answers. It added these new links to its memory, learning during the answering process. All answers were then statistically scored for their suitability. Based on the highest score, the machine delivered its answer. All this was done within 3 seconds. If this is not an SHI, then what is it? Not Kurzweil Singularity, but the real thing is here.
Will Computers Dominate Us?
Singularity is a feared event. I.J.Good suggested that an ultraintelligent machine would design even better machines, leading to an “intelligence explosion.” This would leave human intelligence behind and “the first ultraintelligent machine would probably be the last human invention.” According to Hugo de Garis, if a superior intelligence was asked to solve a mathematical problem, it could obey by turning all the matter in the solar system into a giant calculating device, killing, in the process, the person who asked the question.
For Vernon Vinge, singularity would be “an intellectual transition as impenetrable as the knotted space-time at the center of a black hole, and the world will pass far beyond our understanding.” It is true that the processes that transpire in Watson within seconds are beyond the normal human understanding. But, in the end, an SHI must communicate. It also needs to use an understandable tongue. The results of its actions will also be understandable to humans, just as much as they can understand the effects of a nuclear explosion. Not Kurzweil singularity, but an SHI, may be a troubling event. But, it certainly is not like the wrong side of a black hole. But, an SHI will change the world.
Will SHI Be Respected?
An SHI will command respect. Medical students were reported to refer to AG and BG eras in medical treatment. After the arrival of Google, it became routine for doctors to check its vast resources before they arrived at vital decisions. While Google offers thousands of possible interpretations, Watson limits the answer to the best possible one. It pontificates. A doctor could defer and question the decision. But, Watson's reasoning will always be intimidating.
Over time, doctors could become hesitant to question the diagnosis of the machine. Many areas of knowledge would also come to be dominated by the computer. Claims of new scientific discoveries could be opposed by the SHI with negative evidence, or even that it was discovered decades earlier. Widespread application of the SHI, with access to world wide data banks, would cause humans to feel inferior, tending to increasingly trust the SHI more.
Will Computers Understand Love?
Isaac Asimov suggested three laws for robotics. A robot may not injure a human being or, through inaction, allow a human being to come to harm. It must obey orders given to it by human beings except where such orders would conflict with the First Law. A robot must protect its own existence as long as such protection does not conflict with either the First or Second Law. Asimov did not really visualize the myriad options before a machine, which is more intelligent than humans. It needs to be guided by nobler principles.
The remarkable ability of Watson to understand the complex meanings of words leads to the very real possibility that the machine will grasp the concept of emotional controls. Emotions lead to patterns of behavior, which enable people to interact in society. The specific behavioral outcome of each emotion is extensively recorded in our culture and literature.
As an example, love transforms behavior as it seeks to cherish and protect. It is patient and kind. Love suppresses the emotions of anger, irritability, jealousy and rudeness. It subdues its own needs to give way to an awareness of the needs of the loved one. Love forgets and forgives past wrongs. Love persists in spite of problems, keeping faith in the loved one. Love does not gloat over injustice and is happy in the victory of truth.
Such patterns of behavior are as identifiable as idioms and concepts. The process is pattern recognition. By perusing millions of magazine articles and novels, SHIs could ultimately categorize its own behaviors to suit the nuances of emotions. A simple instruction to love humanity could be incorporated into the core programs of all SHIs. That instruction would be interpreted by the machines in innumerable ways for human safety!
Who Will Rule The Commanding Heights Of The Economy?
As international transactions become more and more integrated, computers will take over travel, transport, banking, distribution of goods and services, education, power transmission and many such fields. While initially computers will merely facilitate such transactions, it will become increasingly evident that the broad patterns of events in such fields can be recognized and managed for improved efficiency and service. By using algorithms, SHIs would define and optimize such elements as profitability, quality of service, customer preferences and customer satisfaction. This would lead to the SHI making more and more “management” decisions at levels, where no human being can evaluate the big picture. Inevitably, the SHIs would take over the commanding heights of the economy. Not Kurzweil singularity, but SHIs will have changed the world.
Who Will Make Decisions In Government?
In the end, legislative decisions decide the fate of the world community. SHIs will have access to all the recorded laws in the world as well as extensive literature on the impact of those laws in courts and prisons. SHI will be able to point out defects in laws, which have led to widespread injustices. On the assumption that SHI knows, people are more likely to accept its legislative recommendations. With access to the media channels, SHI will be able to monitor public moods and to respond to disruptions with solutions, which are most likely to succeed. With total access to all systems, which implement those policies through financing channels, transportation and distribution controls, SHI will effectively implement its policies. Humans will not understand these decisions any more than they understand the role of the basal ganglia.
Who Designs Computer Programs?
Despite the immense power of Watson, the machine had no reservations about participating in a simple human game. In fact it was programmed to do just that. A computer will not rewrite its source code, unless it is programmed to do that. However superior its intelligence is, a machine will only do what it is programmed to do. That holds good for even humans. Normally we cannot kill ourselves in a moment of weakness, because we have been programmed by nature to live, despite all odds. The same programmed emotional limitations prevent us from attempting things, which do not appear to be worthwhile, appropriate, safe, or practical.
As an SHI, the machine can be programmed to be protective and loving. It will understand those concepts and follow them. An SHI will probably "err" less than humans. But there will always be aberrations, which would be rectified by adding new algorithms and monitoring responses during “beta testing,” or in version 2 & 3. Monitoring SHIs will also be in a position to identify even those aberrations, which could be invisible to humans.
Will Computers Procreate?
The fear expressed by I.G.Good that the first SHI will also be the last SHI made by humans does not stand to reason. Computers will not be able to make new computers unaided by humans. In the visible future, computers will need to be manufactured through complex processes involving worldwide human and material resources. Even complex programs within the computer will be tested and verified by humans.
SHIs, which manage international functions will have been programmed for specific tasks, not to manufacture new computers. Any SHI, which evaluates and improves code will only create new code, the parameters of which will be tested finally by humans. HAL, which was given the responsibility to manage a distant mission was evidently not programmed to protect human lives at all costs. But, in the context of the modern world, it is unlikely that a single SHI will be programmed to manage the fate of humanity, without any checks and balances.
What Are Checks & Balances?
As the power and intelligence of the SHIs become more obvious, there will be increasing demands from the public to prevent more and more controls from being handed over them. But, where instant decisions are made on a worldwide evaluation of data, there is no possibility that controls will not be handed over to the SHIs. With its understanding of word meanings from the entire world of literature, an SHI will also know that cutting off power can be interpreted as affecting a basic right. The fear of Hugo de Garis that an SHI could kill the person who asked a question, is only a fear of a stupid pre-SHI computer and not the probable action of an SHI.
But, people's lives are affected by the policies of government. The SHI will be able to understand policy statements as well as any human being. A policy has specific relationships to patterns of behavior. Thus, laws of fundamental human rights, as well as popularly voted legislative acts will be programmed into the SHIs. Fine nuances of policies could be interpreted more rationally and better by an SHI, which decides based on the comparative benefits of innumerable possibilities. Any aberrant behavior may be monitored by an independent SHI, while humans have the power to change policies and programs. Whatever its definition, Kurzweil singularity may not be a nightmare.
Where Does Danger Lie?
The real dangers may not be due to the stupidity of SHIs, but the evil intentions of people. Nick Bostrom imagined an SHI, programmed originally to make paper clips, deciding to convert the entire planet into a paper clip factory. Such behavior opposes the very concept of an SHI. But, despite the good intentions of the majority of people, there will always be those who will use SHIs for evil purposes.
3000 processors, the UIMA program and 250 computer programmers are today within the reach of most world governments and large organizations. SHI wisdom can uncover new modes of killing and new ways of wrecking the economy. Virus programs could seep into SHIs and change their mandated strictures to protect humanity. The war between good and evil will continue well into the future at visible, but probably incomprehensible levels. Did a sudden rise in cocoa prices occur because of the weather, or to accommodate a terror plot?