From DNA to Computer Programming: The Essential Condition for the Existence of Artificial Intelligence. by Dr. Von Kopf (1998)
Marvin Minsky’s definition of artificial intelligence is “the science of making machines do things that would require intelligence if done by man” (The Analytical Engine 299). Having this definition in mind, do you believe a calculator can think? It can perform almost instantly the processes that would take a human many times more minutes to perform. If so, can it be considered intelligent? One cannot simply rely on intuition in answering these questions because the answer would inevitably involve several notions that are not easy to define: “intelligent”, “intelligence”, and “artificial intelligence”. There is still no agreement on what artificial intelligence is, and whether it can exist, and if it can, then, what the main condition for its existence is. Because the ability to learn determines, at most, the presence of intelligence, and since a specially designed program can make a computer to learn on its own, then, the main source of creation of artificial intelligence is not the size or power of computer, but a precise program. It would be most appropriate to begin with defining human intelligence. As it appears, intelligence is not a luxury that every living thing possesses, but rather a higher quality of life capable of “learning”. Paul M. Churchland says: “A system has intelligence just in case it exploits the information it already contains, and the energy flux through it … in such a way as to increase the information it contains, … and that seems to be the central element of intelligence” (173). The ability to learn is what differentiates intelligent organisms from non-intelligent ones; and the proficiency in the ability to learn determines the level of intelligence. Artificial intelligence, then, must be a system built by man and capable of learning. Going back to the calculator, one can easily state now that it hardly possesses any intelligence because although it performs quite complicated tasks, it cannot learn on its own. The given definition of artificial intelligence seems to be very comprehensive and straightforward; however, there are different opinions on whether artificial intelligence exists and whether it can exist at all. For some authors, the existence of Artificial Intelligence is obvious and even natural. Robert Sokolowsky in his article Natural and Artificial Intelligence draws a bridge between these two kinds of intelligence. “When thinking is embodied in the written word, there is something artificial about it”, says he (48). He elaborates on the fact that humans use words to encode information during thinking process. As a consequence, according to Sokolowsky, whenever we use written words to encode information, the same process that usually takes place in our minds during thinking is performed outside of our brains. Based on this comparison, all written information can be considered artificial intelligence. This approach not only reduces both the natural and artificial intelligence to the use of words, but it is in disagreement with our definition of intelligence. “Thinking…essentially is rational manipulation of mental symbols” (Haugeland 70), which are rarely words, but rather visual, acoustic, olfactory (sense of smell), and who knows what else. Similarly, computers mostly use different kinds of semantic symbols rather than whole words in their operation. But most important is that Sokolowsky’s definition of artificial intelligence – a man-made system that operates using written word —does not include the learning aspect in itself. According to Sokolowsky, a book is an example of artificial intelligence, even though it does not possess an ability to multiply the information it contains on its own. Therefore, this definition should be considered simplistic and too broad. In contrast to Sokolowsky, for whom the existence of artificial intelligence is obvious, other scientists think that existing computers are far from being truly intelligent yet. They perceive a computer as a machine that does not possess its own intelligence, but rather demonstrates the intelligence of a human who built it. Really, most of contemporary computers are only able to obtain information by means of a human feeding information to it. In this sense, the opinion that artificial intelligence does not yet exist is justifiable. At the same time, most of the scientists consider the existence of artificial intelligence possible in future, given that necessary conditions are achieved. However, these scientists see the origin of the problem in the size and power of the existing computers. They believe that in order to build a computer capable of establishing as many bonds within itself as a human brain does, “a great deal more processing power” is needed (Waltz 206). The cost and the size of such a computer become the most important issue. To decide if this is really a source of the problem, it would be helpful to look again at a natural brain. At one time we used to compare the human brain with that of other animals in terms of having more or less convolutions (ridges) of gray matter. It appeared, at that time that human brain had the biggest number of convolutions. However, it is known now that “dolphins, and particularly the sperm whales, have brains with more convolutions than the brain of a man” (Cohen 60). There is still no scientific proof that these animals posses the higher intelligence than humans. With consideration to the ratio between the total body weight and the weight of the brain, it is scientifically proven that a bigger brain does not constitute to a higher intelligence level. For example, moles and shrews “have more favorable body – brain ration than a man does” (Cohen 58) without exhibiting much intelligence. Therefore, the size of the brain is not as important. It does not mean, however, that the problem of size and power of computers is to be overlooked. Since brain has millions of processors, each connected to thousands of others, while even most advanced parallel computers have a few thousands processors, each connected to perhaps a hundred others, computers obviously fall behind. This means that in order for computer to become “capable of massively parallel computations” (Decker and Hirshfield 304), it has to be bigger and more powerful. The point not to be missed here is that there is yet another condition to be satisfied before artificial intelligence can exist, and it is a condition without which even an oversized and extremely powerful computer will not function. This condition is the presence of an appropriate program. If not the size of the brain and not the quantity of gray matter is most important to a level of intelligence, then what? It must be the same thing that makes a human brain works faster and be able to solve more complicated tasks than a sperm whale. Some other properties of our brain must be different -- brain chemistry or the way it works, for example. In other words, human brains must be programmed differently. This leads us to the main source of both natural and artificial intelligence – the appropriate programming. Our brain works only because it is programmed to do so. If all the nerve cells were connected properly, but still did not “know” how to use these connections, we would not be what we are. Everyone who knows biology can not deny the fact that we are programmed -- because what else is DNA if not a coded program for our functioning. When it comes to computers, there is the very same requirement for successful functioning as with humans — appropriate programming. Even the most expensive, most sophistically built machine will not work without a proper program. This fact explains why the size and power are, although essential, but not the primary conditions for the existence of artificial intelligence. It is not until we program our computers to learn on their own that they would become intelligent. As it is seen now, both natural and artificial intelligence have in common the principle of ability to learn. "We do not wish to claim there that intelligence only a property of organisms” (Yazdani 270). Artificial intelligence can exist given the necessary conditions are satisfied. The size and power of existing computers has to be increased in order for them to be prepared to perform “intelligent” tasks. However, they would not be able to perform these tasks until the appropriate programming enables them to do so. Taking in consideration the rate at which computer science is developing these days, we may be at the threshold of a new generation of artificial intelligence. Literature Bibliography Works Cited
|
|
|