The Complexity of the Process of Thinking - Essay Example

Can machines think? Section Number of Can machines think? The Oxford English Dictionary defines the word ‘think’ in its verb form as (1) be of the opinion, (2) judge or consider, (3) exercise the mind positively with one’s ideas, (4) consider; be more mentally aware of, (5) form or entertain the idea of; imagine to oneself, (6) choose mentally; hit upon, (7) have a half-formed intention, (8) form a conception, (9) reduce to a specified condition by thinking, (10) recognize the presence or existence of, (11) intend or expect and (12) remember. The process of ‘thinking’ is defined as ‘using thought or rational judgment.’ The complexity of the process of thinking is evident in the very fact that the meaning of the word ‘to think’ can be interpreted in so many different ways. The definition of the process of thinking is also left open-ended enough to accommodate as many if not infinitely more choices and applicability. However, scientists who are involved in what can be termed as thinking capability or ‘Artificial Intelligence’ in machines work on “the premise that all cognitive activity can be explained in terms of computation. This premise has a long and illustrious tradition in Western philosophy, starting with Aristotle and Plato, who believed that thought, like any other physical phenomenon, can be unraveled using scientific observation and logical inference. Gottfried Leibniz, who equated thought with calculation, set the stage for George Boole’s treatise on propositional logic boldly titled “The Laws of Thought.” The advent of computers and the progress made in symbolic computation led to a new branch of computer science envisioned in Alan Turing’s “Computing Machinery and Intelligence.” (Reddy, 1996, p. 86) The answer to the question ‘Can machines think?’ will therefore depend on two fundamental conceptions. First, the definition of the term ‘thinking’ or ‘to think’ will have to be fixed in the context of the man-machine equation, and second and more important, it will have to be determined whether machines are capable of thinking in the context of this definition, and if they are, then to what extent they are capable of doing so in comparison to human beings. Since the faculty of thinking is directly related to intelligence, the capability of thinking, either in man or machine, will also be a function of intelligence. Human intelligence is thus translated to Artificial Intelligence (IA) in machines. The world at large is divided into two distinct camps which hold radically opposing views on the capability of machines to think. The intensity of conviction in both the camps and the basic reasons behind such beliefs is well expressed in Wilkes (1953, pp. 1230) and holds good even today: “Two contrary attitudes are common. In the first place there is a widespread, although mostly unconscious, desire to believe that a machine can be something more than a machine, and it is to this unconscious urge that the newspaper articles and headlines about mechanical brains appeal. On the other hand, many people passionately deny that machines can ever think. They often hold this view so strongly that they are led to attack designers of highspeed automatic computing machines, quite unjustly, for making claims, which they do not in fact make, that their machines have human attributes. Such people are often misled by the use of technical terms based on physiological analogies; a good example is the use of the word memory," for the part of the machine in which numbers are stored.” The objective of this paper is to attempt to reach a conclusive answer to the question ‘Can machines think?’ by fixing a definition for the process of thinking in human beings and then applying it to the domain of machines. However, since almost all machines in the world today, including robots and robotics utilize the basic principles of computers and computer science, the study will try to achieve greater objectivity by focusing on computer systems as a generalized form of machines. The question in such a case could be reframed as ‘Can computers think?’ There could be many possible and plausible answers to this question: i. No a computer cannot think (like a human). ii. Yes, a computer can think (like a human). iii. Yes, a computer can think, but not to the extent a human can (but may do so in the future). iv. Yes, a computer can think, and it can think better than a human being. Significance of the Study The rapid development of Information and Communication Technology (ICT) has brought in unprecedented changes in the way of living of people. Computer systems, the Internet and the Web have become a way of life itself. Computer technology has developed by leaps and bounds. Systems are growing more and more intelligent in the sense that they can perform more and more complicated and complex tasks. Not only so, computing technology has also entered realms which were hitherto thought to be the exclusive domain of the human mind. On the other hand, the Digital Divide is also being bridged at a very rapid rate. ICT is permeating into almost every nook and corner of the world. In such a scenario it is very essential that people understand, comprehend and appreciate the exact character, potentials and limitations of the very technology that is affecting their lives. Knowing the answer to this question, which has perplexed generations, could place every individual in a better position to adapt to and utilize the rapidly proliferating technologies. Something the individual has to, whether he is acquainted with the answer or not. In the history of machines, especially with the computer, there have been instances where things have been blown out of proportions leading to awe and then to disillusionment. This happened with the technological inventions of the period 1880 – 1930, as also with the advent of the computer in the early period of around 1940 to 1960. People were led to believe in the myth that computers were ‘awesome thinking machines’. This mythical representation of the early computer could be traced back to sensational and fictional reporting of a large section of the media who were not familiar with the technology and reported only with the intention of captivating the readers (Martin, 1993, pp. 120) Such a perception of the computer could have led to awe and a certain element of hesitancy on the part of the individual to embrace or utilize the new technology – an attitude that could adversely affect the popularity of any machine or technology. There could be other consequences. Hendler (2006) cites in his editorial an example of how, on the eve of the Wright Brother’s precedent-setting flight at Kitty Hawk, the US army had been in fact been cutting funds to flight research because of the compelling arguments that human flight was an impossibility. When Simon Newcomb, one of the most famous scientists of his day remarked, “Flight by machines heavier than air is unpractical and insignificant, if not utterly impossible,” people were more willing to believe him than the two bicycle repairmen from Dayton, Ohio. False beliefs and conceptions therefore put a restraining bridle on scientific and technological development. It works in the other way too. The case of Artificial Intelligence, which is very pertinent to the paper under consideration, is a prime example. John McCarthy introduced the conception of Artificial Intelligence in 1956. But AI was soon over hyped and it resulted in disappointment and a downslide in IA research funding. Moravec (1998) wrote: “In the 1950s, the pioneers of AI viewed computers as locomotives of thought, which might outperform humans in higher mental work as prodigiously as they outperformed them in arithmetic, if they were harnessed to the right programs... By 1960 the unspectacular performance of the first reasoning and translation programs had taken the bloom off the rose.” There was however a resurgence in the interest in AI in 1981. Though terms like ‘intelligent knowledge-based systems’, ‘expert systems’, ‘knowledge engineering, and ‘multi-agent systems’, were used more in place of AI, it was nevertheless what was termed as the ‘fifth generation’ AI in the context of the Japanese endeavor. But yet again, the potential of AI was overestimated. AI promises on speech recognition and natural language understanding, machines rivaling human intelligence and educating themselves, all went undelivered. By 1990, AI itself was again relegated to the back burner. However, with the advent of the Internet, interest in AI has been revived again albeit in combination with the Human-Computer Interaction (HCI) approach. The basic question in this roller coaster history of Artificial Intelligence has always been the one under consideration of this paper: Can machines think? It applies not only to Artificial Intelligence but the whole gamut of technological development as the ultimate objective of all technological development is to achieve human-level intelligence in machines and execute tasks and functions far beyond human endeavor. Having the right answer to this crucial question could therefore provide the proper perspective to all technology developments and provide the long-term orientation required in all such initiatives. Literature Review What is Intelligence? For the pioneers of Artificial Intelligence, intelligence was the accurate application of well-defined rules to a set of symbolic objects (Grudin, 2006). If intelligence is strait-jacketed so, then computer systems that perform complicated calculations, or the Deep Blue machine that outplays grandmasters in chess can very well be termed intelligent machines – machines that can think. Few would however agree to such a definition of intelligence. It does not take into account the uncertainties of life. That is perhaps the reason why probability theory has been added to ‘New AI’. But will the addition of probability theory suffice to make AI all encompassing? According to another school of thought, perception plays an important role in human intelligence. Much of human reasoning and decision making is said to be based on perception rather than measurements. “Perceptions are intrinsically imprecise, reflecting the bounded ability of human sensory organs and ultimately the brain, to resolve detail and store information. It is this imprecision that places computation and reasoning with perceptions beyond the reach of symbolic logic and probability theory.” (Zadeh, 2007) Zadeh goes on to postulate what he calls the Computational Theory of Perceptions in which the key idea is dealing with perceptions through their description in a natural language. Perception is therefore equated to its description, and computation with perception is actually computation with information that is described in natural language. Fuzzy Logic is applied for this purpose. Zadeh defines Fuzzy Logic as a logic which mirrors the ability of human mind to reason with information which is imprecise, uncertain and partially true. Intelligence assessment and definition thus take a very different course. In Robotics too, things have not remained the same. The ‘sense-act-think’ paradigm was taken to be the operational definition of a robot and as a broad roadmap for robotics research (Siegel, 2003). Mobile robots doing real work in a real world has changed that paradigm. It is essential that robots are able to communicate. Unprecedented development in image capture (sensing) and reproduction (display) technologies in the 20th Century has made it possible to add ‘communication’ to the list of functional requirements in robots. Teleoperation has accorded robots almost unrestrained mobility whereas telepresence has given them the sensing or perception ability. There are more innovations on the technology horizon. Web Intelligence (WI) is a new perspective from the view point of Brain Informatics (BI). It is a new interdisciplinary field that studies the mechanisms of human information processing from both the macro and micro viewpoints by combining experimental cognitive neuroscience with advanced information technology (Zhong, 2005). Bi-directional benefits are envisaged from WI and BI. New understanding new understanding of human intelligence through brain science will yield a new generation of Web intelligence research and development (i.e. BI for WI), and Web intelligence portal techniques will provide a powerful new platform for brain science (i.e. WI for BI) (Zhong, 2006). We finally come to Intelligence Science which is defined as a cross discipline that is dedicated to joint research on the basic theory and technology of intelligence. It employs Brain Science, Cognitive Science, Artificial Intelligence and others to unravel the mystery of human intelligence. Brain Science is concerned with the study of the human brain to deduce intelligence patterns at the molecular, cellular and behavioural level. Cognitive Science studies human mental activity such as perception, learning, memory, thinking, consciousness, etc. Artificial Intelligence, as we know, seeks to replicate and simulate human intelligence in machines. In Intelligence Science these three disciplines work together to explore new concept, new theory, new methodology (Shi, 2006). It is therefore evident that all development in technology is proceeding towards the ultimate goal of replicating human intelligence in machines. The outcome of this study will determine to what extent success has been achieved in this fascinating endeavour. Methodology This paper will adopt a three-pronged approach to arrive at the answer to the question it poses: i. The Backdrop: The methodology adopted will be based on what is known as the Turing Test. The Turing Test was originally conceived to address the question of a thinking machine. An answer derived at by the Turing Test would perhaps be more acceptable in its finality. ii. General Impression and the Media: To complement the findings arrived at on the basis of the Turning Test, this paper will undertake a thorough examination of the popular concept of the computer as a thinking machine as portrayed in the media and as imbibed by a larger section of the population. It will also attempt an analysis of the role and influence of the media in promoting machines in general and computers and robots in particular, as possessing intelligence equivalent to that of human beings. iii. Scientific Perspectives: The paper will also endeavor to view the question of machine intelligence vis-à-vis human intelligence from the perspective of various allied but distinct scientific approaches such as Artificial Intelligence and Robotics that use newer concepts of Fuzzy Logic, Web Intelligence, Brain Informatics, Intelligent Agents, etc. The inferences drawn from these three approaches will be combined together to culminate in an ultimate answer. While the methodology section of this paper incorporates a detailed description of the Turing Test and how the paper will attempt to make the Turing Test the backdrop of its methodology, the Literature Review section of this proposal dwells on the body of information available for the general impression created by the media and the scientific perspectives. British Mathematician Alan M Turing was the one who first figured out how to build a programmable computing device – the universal Turing Device. All programmable computers today in essence use the Turin principle. It was also Turing who first raised the question, ‘Can machines think?’ in 1950 (Turing, 1950). However, later Turin went on to opine that the question he had raised was a bad one – a question that led only to meaningless debates and controversy over definitions. He termed it, “too meaningless to deserve discussion” (Turing, 1950). In its place, Turin proposed the Imitation Game or the Turin Test (TT) to find out whether a machine or a computer is actually capable of thinking. The Imitation Game or Turing Test provides a method to test whether machine is capable of thinking or not. The Imitation Game is played with a Man, a Woman and an Interrogator or judge who can be of either gender. The man and the woman are hidden from the interrogator behind a screen, but both communicate with the judge by teletype. The aim of the interrogator is to find out though questions out through the teletype which is the woman and which is the man, whereas the aim of both the man and the woman is to convince the interrogator by their answers through the teletype that he or she is the woman and the other is not. The interrogation asks questions in written natural language and receives the answers in written natural language. Questions can be on any subject imaginable. The man therefore has to deceive the interrogator and the woman has to aid him in this deception. Turing then proposes that a computer replace the man, and both the woman and the computer try to convince the interrogator that both are a woman. “According to Turing, the new agenda to be discussed, instead of the equivocal ‘Can machines think?’ can be ‘What will happen when a machine takes the part of the man in this game? Will the interrogator decide wrongly as often when the game is played like this as he does when the game is played between a man and a woman?’”(Saygin, et.al., 2000, pp. 465). The computer, at the time when Turing had conceived the Imitation Game, had no chance at all. But Turing had made some bold prophesies. He had said that by 2000, computer technology would advance to a stage in which the Interrogator would not have more than 70 percent chance of making the right identification after five minutes of questioning. The Turing’s Test still remains a challenge, and an acceptable standard by which the intelligence of any machine can be assessed. This paper however does not propose conducting the Turing Test on any machine. Rather, literature of the findings of Turing Test conducted in all these years will be analyzed to evaluate the actual consolidated outcome. The Turing Test has never been carried out exactly the same way as Turing originally described it, but different versions have been tried out. Since 1991 Hugh Loebner has been organizing a so-called Annual Loebner Contest (Loebner, 1994). It has been the most well known forum for participating in the Turing Test. Though no program has been able to pass the unrestricted Turin Test, the quality of the participating programs has increased from year to year. This paper will examine in detail the performance of well known programs such as ELIZA the natural language system of Joseph Wizenbaum, PARRY of Kenneth Colby and SHRDLU of Terry Winogard (Maybury, 1990) in a bid to find out to what extent the advancement of technology has enable machines to think by the standards of Turing. References 1. Reddy, R., 1996, The Challenge of Artificial Intelligence, Computer, October 1996, 0018-9162196185 00 @ 1996 IEEE 2. Wilkes, M.V., 1953, Can Machines Think? Reprinted from Discovery (England) vol. 14, p. 151; May, 1953, in Proceedings of the I.R.E., October 2006 3. Martin, C.D., 1993, The Myth of the Awesome Thinking Machine, Communications of the ACM, April 1993/Vol. 36, No. 4 4. Hendler, J., 2006, Fly, but not like an Eagle, IEEE Intelligent Systems, IEEE Computer Systems, 1541-1672/06/$20.00 © 2006 IEEE 5. Moravec, H., 1998, When will computer hardware match the human brain? Journal of evolution and technology, 1, 1. 6. Turing, A. (1950), Computing Machinery and Intelligence, Mind 59(236), pp. 433–460. 7. Syagin, A.P., Cicekli, I. & Akman, V., 2000, Turing Test: 50 Years Later, Minds and Machines 10: 463–518, 2000, © 2001 Kluwer Academic Publishers 8. Loebner, H.G (1994), In Response, Communications of the Association for Computing Machinery 37, pp. 79–82 9. Maybury, M.T., 1990, The Mind Matters: Artificial Intelligence and its Societal Implications, IEEE Technology and Society Magazine, June/July, 1990. pp. 8 -15. 10. Grudin, J., 2006, Turing Maturing: The Separation of Artificial Intelligence and Human-Computer Interaction, Interactions, pp. 54 – 57. 11. Zadeh, L.A., 2007, Towards Human-level Machine Intelligence, Department of EECS, University of California, Berkeley, pp. 9 – 10 12. Siegel, M., 2003, The Sense-Think-Act Paradigm Revisited, Intenational Workshop on Robotic Sensing, Orebro University, Orebro, SWEDEN, 5-6 June 2003 13. Zhong, N., 2005, Web Intelligence Meets Brain Informatics: An Impending Revolution in WI and Brain Sciences, Advances in Web Intelligence, Springer LNAI 3528. pp. 23-25. 14. Zhong, N., 2006, How to Make “Web Intelligence (WI) meets Brain Informatics (BI)” Successfully?, Proceedings of the 30th Annual International Computer Software and Applications Conference (COMPSAC06) 15. Shi, Z., 2006, On Intelligence Science and Recent Progress, Conference on Cognitive Informatics, ICCI’06. Read More