The Singularity is Near for Artificial Intelligence and Man

By

Joe Chiarenzelli

Ray Kurzweil has been at the cutting edge of technological and scientific exploration since his graduation from MIT (Kushner 2009, 56-61). In a time of global strife it is necessary to look to our leaders in fields of high practicality to help us envision and plan for the future. Ray Kurzweil’s model of information acceleration followed by a singularity is very accurate and can be compared to historical and current data to prove that, if no other factors come into play, the current path of technology, and artificial intelligence specifically, will continue in correspondence with Kurzweil’s model. This process can be mapped by discussing the theory behind the acceleration of information, the data that currently points to a trend very similar to an exponential acceleration of information, and what constitutes a valid prediction about the future of technology and science.

 

 Ray Kurzweil, in his books, outlines the basic theory of information acceleration by compiling the assumed paradigm shifts in human history. In all Ray uses 15 sources to compile his list of paradigm shifts, these sources range from the Encyclopedia Britannica to the prominent scientist Carl Sagan (Kurzweil 18)(Kurzweil 2008, 10-10). A paradigm shift, in this situation, can be defined as the scientific and cultural framework on which human life operates. When graphed on a logarithmic scale versus time, these points form a straight line (Kurzweil 17). This means that when plotted on a normal scale there is exponential growth (Kurzweil 18). This growth is representative of the exponential increase in paradigm shifts as time progresses. However, on a small level this curve is not smooth. Each technology takes the shape of an S, a technology exponentially gains acceptance and after that it reaches an upper limit of permeance. But, Kurzweil illustrates that exponential growth gets around this by showing that when a technology reaches its limit it is replaced by a new technology that continues the S curve. So, when looked at from a macro level the curve looks smooth, but it is in fact a series of S curves. Due to the fact that paradigm shifts result from a shift in an obsolete to a more efficient and advanced model, they are inherently representative of technology and information itself. Thus, in illustrating the exponential nature of paradigm shifts through human chronology, Kurzweil provides a solid base for his discussion of an apparent exponential nature to this increase.

Furthermore, Ray provides more evidence behind this trend in information and technology. Gordon E. Moore, in 1965, described the nature of a certain type of technological acceleration, namely transistors that could be placed on a chip inexpensively. This relationship over time also illustrated an exponential increase, having a doubling time of two years (Lundstrom 210-211). Moore was immortalized when the term Moore’s law was coined to describe this exponential increase in transistor count. However, transistor counts are not the only area of technology to increase exponentially. Over time the cost per transistor has also decreased exponentially, which, when combined with Moore’s law allowed a massive increase in widely available computer power that resulted in the generation of other exponential trends (Kurzweil 62). Computer performance per unit cost, in direct correspondence to Moore’s law, doubles every two years. Virtually all areas of information technology exhibit this exponential trend, from price per bit transferred to the pixel per cost value of digital cameras. However, this trend does not only address subjects within the field of information technology.

Kurzweil, in a speech sponsored by the technology, entertainment, and design program (TED), outlined a similar process in mapping the human genome, a biological proposition. The project was originally slated to last fifteen years, beginning in 1990 (Kurzweil 2008). In the first five years the most advanced technology of the time had only mapped one ten-thousandth of the human genome (Kurzweil 2008). In the next five years there was still no significant progress, still having sequenced only a fraction of the entire genome (Kurzweil 2008). If this was a linear progression, obviously the project never could have been concluded within the original time-line, however, its progress took the shape of an exponential function. This increase in the rate at which decoding the DNA was performed resulted in the human genome being sequenced on time, with most of the work done in the last five years (Kurzweil 2008). Kurzweil also gives the example of the sequencing of SARS and HIV, “HIV was sequenced in fifteen years, and we sequenced SARS in thirty-one days.” (Kurzweil 2008). Kurzweil also cites the importance of biological systems as a model for exponential growth by explaining that the evolution of DNA required billions of years, while the Cambrian explosion, a period when great biologically diverse growth occurred, took only 10 million years after that. In Kurzweil’s theory even biological systems demonstrate this powerful growth.

This exponential process also dictates how we interact socially. Kurzweil states that it took fifty years to adopt the telephone, while cell-phones were adopted in eight years (Kurzweil 2008). This shows an increase in the consumer’s rate of adoption of technology. Kurzweil makes the point that, if one is to analyze the adoption rate of different information technologies the most recent are adopted in a much shorter time period, less than a decade, as compared to earlier ones such as, television or radio (Kurzweil 2008). The more connected the world becomes the more it is possible to exchange ideas and data at increasing speeds, causing a further natural acceleration of information.

An important point in Kurzweil’s theory is that even though technologies have limits, they facilitate the growth of technologies which can come to take their place (Kurzweil 43-44). This conception is highly visible in the world today. Recently the DVD format has been being replaced by both Blue-ray and HDVD, both of these formats are greatly improved entertainment formats that satisfy the demand for greater resolution in both audio and video (Bachman 1). Obviously then, although Moore’s law does have projected limits based on the current paradigm of flat chips with smaller and smaller transistor sizes, developments in chip design such as layering chips or developments on the basic substrate of computing such as moving computing to a quantum format provide possible replacements for these technologies (Preskill 469-486) (Clark et al. 1451-1471) (Kurzweil 111).

Artificial intelligence within this system could be seen to follow a very similar exponential path. Although having been originally conceived in promethean myths exhibiting man’s creation from clay, an unthinking material, the concept never received any legitimate scientific inquiry until our current century (Buchanan 53-60). Contemporary artificial intelligence research is divided into two sub-groups, strong AI and weak AI. Weak artificial intelligence is based on creating an intelligent system for addressing a certain problem and can already be performed in some fields, (Cheng et al. 237-247) (Bassam et al. 773-780) (Alvarez-Estevez 7778-7785). However, strong artificial intelligence represents a more difficult goal. A strong artificial intelligence would be able to transcend human abilities and apply its intelligence to any field of its choosing (Anderson and Copeland 371-378). While limited success has been achieved in the field of weak artificial intelligence using different methods, as can be seen in weak AI systems tailored to a specific task (Cheng et al. 237-247) (Bassam et al. 773-780) (Alvarez-Estevez 7778-7785), there is still a long way to go if we are going to construct a machine that can surpass our own intelligence. However, research in this field is developing at an accelerating rate, causing a massive growth in the field in the latter part of the 20^th century and the beginning of the 21^st. The major force behind the creation of smarter, faster, and more self-reliant computers is the exponential growth of the field, this growth builds on its self synergistically. For example, as information technology develops exponentially we are better able to evaluate and apply information coming from intelligent machines, this, in turn, allows us to create intelligent machines which can handle more of the data analysis load. It is this progression, of artificial intelligence creating artificial intelligence that will magnify and amplify the current research into a whole capable of replicating and even exceeding human intelligence.

Ray Kurzweil does find a conclusion to this acceleration of information however, he calls it the singularity. Once this self-perpetuation graduates to a post-human level of intelligence the artificial intelligence’s can take over their own growth and development, further increasing the rate at which their intelligence is developed (Kurzweil 8-9). The singularity will result in a sustained unfathomable increase in technology and intelligence which, for Kurzweil, will create a utopian society in which all our needs are met and exceeded (Kurzweil 8-9, 220-440). Kurzweil believes that advances in nano-technology will allow our bodies to be augmented and maintained by nano-particles creating a post-human meld of man and machine (Kurzweil 226-258). He further believes that there will be an increasing inter-face between the human mind and networks like the internet, this will allow creation of virtual realities for human enjoyment along with scientific exploration (Kurzweil 226-258). After this point humans will expand into the universe and process most of the inanimate matter, to harness its inherent computing power and, by doing so, create an “awakened” universe. This viewpoint is typical of utopian futurists who believe that man’s progress will develop and serve the common good; this ideal is shared by prominent thinkers like Buckminster Fuller. While this represents a best case scenario for human-kind’s technological development, there are many thinkers who disagree.

In the 2007 book “What is Your Dangerous Idea: Today’s Leading Thinkers on the Unthinkable”, a host of scientists discuss the possible pitfalls in the acceleration of technology and information. The psychological and artificial intelligence fields are on a collision course for melding into one science of intelligent systems, for this to happen however the existence of free will must be disproved. Disproving the concept of free will is central to artificial intelligence, because it means that the complete human experience can be modeled by algorithmic calculations. This concept presents some very tricky situations for future societies. Paul Bloom says in “What is Your Dangerous Idea” that “the widespread rejection of the soul would have profound moral and legal consequences. It would also require people to rethink what happens when they die, and give up the idea (held by some 90 percent of Americans) that their souls will survive the death of their bodies and ascend to heaven. It is hard to get more dangerous than that.” (Brockman 4-6). If it was true that no one had a soul (free will), but rather was only a set of algorithms, then there would be no choice but to rethink the entire concept of moral action and free will. If the human is only a product of mathematical algorithms, then is it ok to punish a human being for doing what they are required by the laws of mathematics to do? The under-lying problem here is an admission of no free-will meaning that every action must result from algorithmic processes and, if I have no free-wil,l than how can I be held accountable for any crime? Another idea discussed in this book is the anthropic principle this says, essentially, that the universe acts and behaves the way it is because it is perfectly attuned to life and, if this was not the case there would be no opportunity to come to conclusions on how the universe works since there would be no sentient entities to examine it.

From this principle Robert Shapiro claims that, “The origin of life would be a natural (and perhaps frequent) result of the physical laws that govern the universe. This latter thought falls directly in line with the idea of cosmic evolution, which asserts that events since the Big Bang have moved almost inevitably in the direction of life. No miracle or immense stroke of luck was needed to get it started. If this turns out to be the case, then we should expect to be successful when we search for life beyond this planet. We are not the only life that inhabits this universe.” (Brockman 65-68) This also represents an obstacle in our under-standing of the universe, because it will result in a major rethinking of our position within the universe and whether or not there is any purpose to our carrying on in any function because, inevitably, there will be others to take our place. However, these possibilities represent the potential of science to send us into an existential crisis, a host of other thinkers think that technology will directly destroy the world or create a dystopian future.

Under the shadow of a nuclear holocaust, most humans believe that if a nuclear war can be avoided we are free and clear to develop a world that will benefit all. However, with the current economic crisis and rising global tension, many parties believe there are many more possibilities for other technological crises. Virgin media has compiled a list of ten technologies that could destroy the world. With crises with computer viruses it is of the utmost importance to understand and guard against viruses. If a worm was to infect all the computers connected to the internet and turn them off, we could see massive mortality. Train schedules, airports, and bus systems all rely on computers to regulate arrival and departure times; this reliance is only being compounded over time. If there is a virus which can turn all of these computers off, then obviously there will be catastrophic consequences (Virgin Media 2009). On an even more macabre note, what if a worm, after the advent of post-human artificial intelligence, could infect these systems and cause them to turn on their makers? Of course even if we assume there is an anthropogenic intent behind a dystopic world event such as, nuclear war, other sorts of advanced weaponry, complete invasion of privacy, or even mind control, that does not rule out the possibility of some sort of human technology having unforeseen consequences.

Recently there has been quite an uproar about the activation of the large hadron collider in Switzerland. Many groups are clamoring to stop the LHC from reactivating, because on initial activation it encountered technical problems, due to the nature of particle accelerator’s unforeseen consequences (Overbye 2009). Particle accelerators work by smashing together subatomic particles to examine their constituent parts. When these collisions occur many subatomic particles are formed, like higgs bosons and quarks, but there is also a possibility of creating a very small black-hole (Overbye 2009).. Physicists explain away the risk by citing Hawking’s radiation, a process by which small black holes evaporate through gamma ray emission, but given the previously untested nature of these experiments nothing is a foregone conclusion. While it is doubtful the large hadron collider will create any significant threat to the earth’s safety, as technology accelerates we will further be able to manipulate our world in untested ways. One of the key ingredients in nanotechnological development is the creation of self-replicating nano-particles. While it is presumed these particles will have, encoded in them, a stop code to prevent them from creating too many copies it is always possible, en route to this technology, that there could be a release of an unstoppably self-replicating mechanistic system that will voraciously devour the planet at a faster and faster rate. While these developments are dangerous they are a necessary step on the technological curve that Kurzweil proposes, these risks make many conclude that we may want to stop the explosiveness of our technological growth rate to consider the implications and inherent risks these technologies present.

In my opinion Kurzweil’s arguments have several flaws. Firstly, the fact that the overall exponential curve is composed of smaller S curves creates a point of criticism. There is no absolute guarantee that whenever a limit is reached a new technology can be generated to replace it. To be conservative, all technology must stay within the bounds of physics. For instance, there is no possible way, known to us, to transcend the physical laws of the universe. Because of this there can be no possible transcendence of the speed of light or the other cosmological constants that bound the known universe. Also, the ability for humans to control the new observational knowledge of the universe may also be limited. Kurzweil outlines major exponential curves in computing technology, but this does not mean that these can be transmitted into actual practical usages. For example, even if we can somehow get to the point where we can simulate a human intelligence using great processing power, it would require an unbelievable ability to simultaneously analyze and determine the position of all neurons, which number in the trillions. At a small level like this, there is a certain indeterminacy that is necessitated by Heisenberg’s uncertainty principle which says that it is impossible to know the position and velocity of a particle simultaneously. However, Kurzweil’s writing provides un-provable counter arguments to this. He claims that we have no idea what the technologies will be that replace the current but that they will emerge, something that is impossible to prove or disprove without some sort of precognition. Also, he believes that it is possible to transcend the limits of physical constants, but again he cannot say where exactly they will come from. While it is impossible to prove or disprove these statements, they are differences of opinion that can make an individual doubt or embrace his vision.

Ray Kurzweil’s theories certainly have a strong theoretical background and an almost irrepressible conclusion. What we can learn from his general argument is that human progression, while chaotic and unstable in the short term, follows a predictable curve. Due to the predictability of this curve we can then discuss what future developments will occur within our society and how these developments will change our current culture paradigm and what it means to be human. While the nature of these curves are irrefutable, the end result, or singularity, is of an opinionated nature and has no historical precedent to be compared against. Because of this there is much conjecture on the nature of a singularity and what it will entail. Some believe in a purely utopian future where every whim and wanting can be satisfied through the use of technology. Others believe that this technological revolution will carry with it existential and social consequences that will throw us into a cycle of self questioning. Dystopians believe that the coming technological revolution will create far too many possibilities of the earth’s elimination for one not to occur. In essence there is almost no question, if one accepts Kurzweil’s trends, for their not to be a singularity, but, at this point, the nature of a singularity is open to conjecture.

References

Álvarez-Estévez, Diego, and Vicente Moret-Bonillo. 2009. Fuzzy reasoning used to detect apneic events in the sleep apnea-hypopnea syndrome. Expert Systems with Applications 36, (4) (05): 7778-85.

Anderson, David, and B. J. Copeland. 2002. Artificial life and the chinese room argument. Artificial Life 8, (4): 371-8.

Bachman, Justin. 2008. Samsung's BD-P1500: Blu-ray, priced right. Business Week Online (11/12): 1-.

Bassam, A., D. Ortega-Toledo, J. Hernandez, J. Gonzalez-Rodriguez, and J. Uruchurtu. 2009. Artificial neural network for the evaluation of CO₂ corrosion in a pipeline steel. Journal of Solid State Electrochemistry 13, (5) (05): 773-80.

Brockman, John. 2007. What is your dangerous idea? :Today's leading thinkers on the unthinkable. 1st ed. New York: Harper Perennial.

Buchanan, Bruce G. 2005. A (very) brief history of artificial intelligence. AI Magazine 26, (4): 53-60.

Cheng, H. D., Xiaopeng Cai, and Rui Min. 2009. A novel approach to color normalization using neural network. Neural Computing & Applications 18, (3) (05): 237-47.

Kurzweil, Ray. “The Acceleration of Information.” Lecture, TED lecture series, TED, 6/24/2008.

Kurzweil, Ray. 2008. The singularity: The last word. IEEE Spectrum 45, (10) (10): 10-.

Lundstrom, Mark. 2003. Moore's law forever? Science 299, (5604) (Jan. 10): 210-1.

Preskill, John. 1998. Quantum computing: Pro and con. Proceedings: Mathematical, Physical and Engineering Sciences 454, (1969, Quantum Coherence and Decoherence) (Jan. 8): 469-86.

Progress in silicon-based quantum computing. 2003. Philosophical Transactions: Mathematical, Physical and Engineering Sciences 361, (1808, Practical Realizations of Quantum Information Processing) (Jul. 15): 1451-71.

Wireless energy transfer - 10 ways technology could destroy the world - pictures - digital - virgin media. [cited 4/3/2009 2009]. Available from http://www.virginmedia.com/digital/galleries/how-technology-will-destroy-the-world.php?ssid=10 (accessed 4/3/2009).