PROF. ROBERTS ‐ N305: FINAL PROJECT
AI: A PARADIGM IN PARALYSIS? A DISAMBIGUATION OF ARTIFICIAL INTELLIGENCE J. Charles Schleutker 6/16/2008
Before getting into the harry details of Artificial Intelligence (AI), I would like to pose
simple question to the reader: What doe AI mean to you? Do you associate anything in particular with the concept? Before I started researching AI, I asked myself the same question, and could only associate a handful of fictitious instances from either a film or a book. I couldn’t put AI into a real‐world context or think of a practical application besides video‐games, which is hardly a bi‐product of the real‐world. This is part of the problem with AI research today; it is so ambiguous and ill‐defined that AI only presents more problems than concrete solutions. In fact, when evaluating AI’s performance in carrying day‐to‐day in comparison with humans AI is still considered sub‐human or performs worst than most humans. However, this doesn’t mean that AI research hasn’t provided any concrete solutions. In fact, medical diagnoses, stock trading, robot control, law, scientific discovery, and toys are the most recent benefactors of AI, while time sharing, GUIs, the computer mouse, and object‐oriented programming were all founded in AI laboratories. So why are researchers encountering more problems than solutions in AI? If AI “is both the intelligence of machines and the branch of computer science which aims to create it,” then why are researchers having so many issues creating intelligent machines? Being such a broad discipline, AI often defines their problems in the theoretical sense; here are just a few of AI’s short‐comings in comparison to man: deduction, problem‐solving, unconscious knowledge, the breadth of commonsense knowledge, planning, methods of machine learning, perception, motion, language processing, and creativity is only a handful of problems that burden computer scientists and our modern‐day machines. Some might criticize AI researchers as not clearly defining and understanding intelligence on a human‐scale, let alone bothering in an attempt to create it. Others may argue that AI is just presently under construction given the lack of
through representation of symbolic languages. This has catalyzed an interest in developing sub‐ symbolic or bottom‐up approaches to accurately imitate the intelligence of man in machines. ARTIFICIAL Made by human skill; produced by humans (opposed to natural); lacking in natural or spontaneous quality; *based on differential morphological characters not necessarily indicative of natural relationships. Artificial reinforces the notion of invention or ‘man‐made;’ in case of AI, artificial might accurately depict how man uses his tools, given his intelligence. Others might say artificial is merely proliferation of technology in accordance with AI.
INTELLIGENCE: soft‐core version; A very general mental capability that, among other things, involves the ability to reason, plan, solve problems, think abstractly, comprehend complex ideas, learn quickly and learn from experience. It is not merely book learning, a narrow academic skill, or test‐taking smarts. Rather, it reflects a broader and deeper capability for comprehending our surroundings—"catching on", "making sense" of things, or "figuring out" what to do. This is the Mainstream Science on Intelligence’s matter‐of‐fact definition of intelligence. It is, of course, defined via the context that man is intelligent.
INTELLIGENCE: hard‐core definition; “Individuals differ from one another in their ability to understand complex ideas, to adapt effectively to the environment, to learn from experience, to engage in various forms of reasoning, to overcome obstacles by taking thought. Although these individual differences can be substantial, they are never entirely consistent: a given person’s intellectual performance will vary on different occasions, in different domains, as judged by different
criteria. Concepts of "intelligence" are attempts to clarify and organize this complex set of phenomena. Although considerable clarity has been achieved in some areas, no such conceptualization has yet answered all the important questions. . .” This is the APA’s version of intelligence, while it is certainly most and probably more accurate then the previous contender; it merely reinforces the scientific community’s lack of evidence and tools for defining and measuring intelligence.
ARTIFICIAL INTELLIGENCE Major AI textbooks define artificial intelligence as "the study and design of intelligent agents,"[1] where an intelligent agent is a system that perceives its environment and takes actions which maximize its chances of success.[2] AI can be seen as a realization of an abstract intelligent agent (AIA) which exhibits the functional essence of intelligence.[3] John McCarthy, who coined the term in 1956,[4] defines it as "the science and engineering of making intelligent machines."[5]
At the present, the problem domain and serious application of AI have both taken the backseat to how scientists should approach AI altogether. Since AI research is presently generating more questions than answers some would characterize the field as a paradigm in paralysis, or more specifically a scientific discipline where the problems they are attempting to solve are beyond their current models of thinking. However, other AI advocates may argue that paradigm shift is about to begin, or given the several sub‐disciplines AI has spawned it is this research of which will catalyze the next paradigm shift in science; a paradigm shift is a dramatic event in modern day science where the a scientific discipline by means of deduction or induction that radically changes scientist’s world‐view of reality. As of just yet, computational intelligence (CI) is the self‐proclaimed successor of the broader more convoluted field known as AI while it comically refers to AI as GOFAI (Good Old Fashioned AI). CI is particularly different than AI in its adoption of entirely new algorithms and data‐sets of which might express something more in more complex terms. It is noted that CI relies on heuristic algorithms such as in fuzzy systems, neural networks and evolutionary computation. The other, less conventional, sub‐field of AI that has been generating attention is referred as
synthetic intelligence. (SI) It emphasizes the belief of many researchers that the intelligence of machines is not an imitation or in any way artificial; it is a genuine form of intelligence. According to this view, the term artificial intelligence is an oxymoron; implying that the intelligence is not intelligent. As absurd as SI really sounds it is in my view‐point that SI the closest present sub‐discipline in practice that reiterates Alan Turing’s, the founding father of AI, most original idea of intelligent machines.
The origin of AI was suggested in 1950 by an intelligent young man by the name of Alan
Turing; his claim, in a nutshell, was that since machines are capable of performing the same computational operations as a man then machines both, can and do think. It sounds a lot bolder in paraphrasing this claim, but if read it for yourself, you’ll see why: I propose to consider the question, 'Can machines think?'"[8] As Turing highlighted, the traditional approach to such a question is to start with definitions, defining both the terms machine and intelligence. Nevertheless, Turing chose not to do so. Instead he replaced the question with a new question, "which is closely related to it and is expressed in relatively unambiguous words".[8] In essence, Turing proposed to change the question from "Do machines think?" into "Can machines do what we (as thinking entities) can do?"[9] The advantage of the new question, Turing argued, was that it "drew a fairly sharp line between the physical and intellectual capacities of a man.[10]
The original claim is, ‘Can machines think?’ What is Turing suggesting by deliberately not defining the key terms necessary in answering such a question? Could it be that machines and intelligence are irrelevant to AI altogether? Next, consider his rephrasing of the claim from ‘Do machines think?’ into ‘Can machines do what we (as thinking entities) can do?’ At first it might not apparently be clear what Turing is trying to get it; his first proposal is whether or not a machine is capable of thought, if the machine is capable of thought, does it think? Lastly, his claim changes form one last time, while it seems to beg the question: Am I capable of thought?
Do I think? Can a machine do these two things I can do? Even today, nearly sixty years ago, this claim still holds a lot of weight given everything computers can’t do that humans still can. In fact, the concept Turing underlines in making such a claim will quite possibly be timeless until the advent of Strong‐AI, which is where AI is capable of “such traits as sentience, sapience, self‐ awareness and consciousness.” However, returning to Turing’s claim and presenting it in a more practical point‐of‐view given today’s technology, consider the following dialogue between me and my laptop: ME: What can I do? AI: Swim. ME: You can’t swim. What can I do with you? AI: Browse the Internet. ME: I can read. How/why did I learn to read? AI: You learned to read by reading. You read because, as a human, you use reading as a means of learning as a means of achieving a particular state in reality. Reading for you is the act of acquiring and storing the writer’s input as your means of producing a new output until your state is achieved. ME: Can you read? AI: Yes, but only to the extent of my given limitations of intelligence. ME: Can you read faster than me? AI: Yes. ME: Can you read what I read and help me faster achieve my desired state? AI: Yes, but once again, only in context of my intelligent‐set that you have provided for me.
The aim of this dialogue is supposed to help reinforce how Turing wants us to think about computers; while the plausibility of such a conversation may seem far‐fetched it is certainly not impossible. This conversational instance might at first not appear very practical given your
everyday person, but it helps illustrate how and why we would exploit a computer as a computational toolkit in answering questions like ’How can a computer do what I can do?’ Because the only way a computer can do what you can do is by showing the computer how to do it, which suggests a much more practical paradigm given the infinite state of intelligence and machines along with our present notion of technology. However, while the state of intelligence and machines is certainly infinite and ever‐increasing, my own intellectual capacity is not, since after all I am only human and will always have more problems than solutions.
Sources Cited "Artificial intelligence." Wikipedia, The Free Encyclopedia. 12 Jun 2008, 06:35 UTC. Wikimedia Foundation, Inc. 16 Jun 2008 . "Intelligence." Wikipedia, The Free Encyclopedia. 13 Jun 2008, 00:59 UTC. Wikimedia Foundation, Inc. 17 Jun 2008 . "Paradigm." Wikipedia, The Free Encyclopedia. 16 Jun 2008, 13:21 UTC. Wikimedia Foundation, Inc. 16 Jun 2008 . "Computational intelligence." Wikipedia, The Free Encyclopedia. 5 May 2008, 11:08 UTC. Wikimedia Foundation, Inc. 17 Jun 2008 . "Synthetic intelligence." Wikipedia, The Free Encyclopedia. 22 Apr 2008, 10:18 UTC. Wikimedia Foundation, Inc. 17 Jun 2008 . "Strong AI." Wikipedia, The Free Encyclopedia. 13 Jun 2008, 19:22 UTC. Wikimedia Foundation, Inc. 17 Jun 2008 .