AI through the years has progressed very far, but not far enough to where we have robots running around the house like C-3P0 and R2-D2. We have only minimal intelligence inside robots, and only enough to do basic functions they are programmed to do. Even though problems exist with Artificial Intelligence, the history, progress and discoveries, and even future developments prove there have been positive steps in the science of Artificial Intelligence.
In 1920, a Czech playwright, Karel Capek, wrote R.U.R., Rossum’s Universal Robots. The popularity of the play was so huge, the word Robot replaced the old term, automaton, out of use in almost every language used on Earth. (Asimov 863) With the advances of robotic science and artificial intelligence, the two terms have almost become synonymous. In the 1950’s, the word Artificial Intelligence was coined, but even before the word existed, AI researchers had been searching for ways to make machines think and act like humans (Masci 1).
The Robotic Laws, published by Issac Asimov in 1942 in a short story called I, Robot, is a set of four laws that are to be set in a robot’s mind if it is ever gains free-willed consciousness. The “zeroth” law states, “A robot may not injure humanity, or through inaction, allow humanityto come to harm.” This “zeroth” is the highest order law and cannot be overridden by any other law (Maulis 2). The first law states, ”Robots may not injure a human being, or, though inaction, allow a human being to come to harm.” The second law says, ”A robot must obey orders given by humans, except where orders would conflict with a higher order law.” The third law says, “A robot must protect its own existence as long as such protection does not conflict with higher order laws” (Clarke 1).
A fully conscious mind is one of the most complex problems in today’s sciences. There is nothing we know more about, but at the same time, there is nothing harder to explain and understand. Even with the major discoveries and equipment scientists have, we still have just a rudimentary knowledge of the basic workings of the brain, but we have no idea of the complex internal workings needed for true artificial intelligence (Chalmers 1).
Pessimists say one problem in creating AI is having it have essential human attributes like common sense and the ability to appreciate and understand the subtleties of meaning in language (Masci 2). Another problem for AI’s is Robotic Law deadlock. Deadlock is when a robot cannot decide between choices because doing any choices will violate a law. A deadlock problem is the key feature in several Asimov stories about robots. He constructed the type of standoff referred to as the “Burridan’s Ass” problem. An example is “two humans threatened with equal danger and the robot is unable to make a strategy to protect one without sacrificing the other.”
Another example AI scientists use is the “Hay fever and Jockey” problem:
If someone sees a jockey with hay fever and knows there is a lot of goldenrod on the track, he will not bet on that horse. While making that decision, he sees many other things but knows how to block them out, but scientists doubt that a computer would be able to come to the same conclusion because of the inability to block out the irrelevant information (Masci 2).
These problems show there is a need for extra programming that helps a robot solve deadlock. There have been several theories developed by scientists to help solve the deadlock problem. One way is to change a variable randomly and then re-evaluate the situation. If no way is found, then the robot randomly chooses a way and follows it without it questioning itself (Clarke 5).
Another problem is we are thinking that we have to build the AI like a human, but we need to be thinking in the sense of making AI like a computer. The human brain is made of DNA and proteins with a watery background created by three and a half billion years of evolution and natural selection. The computer is made of transistors and microprocessors with a semiconductor background created by scientists working in labs for fifty or sixty years, and with the two intelligence’s backgrounds being so different, just going with the current strategy of attacking the problem may not be the best way of solving it (Travers 866).
“The problem of verifying whether an intelligent being is self-conscious is even more complex” (Buttazzo 3). One way to explain this better is to say that if intelligence is the expression of externalbehavior that can be measured by tests, self-consciousness is an internal state that cannot be measured by any means. In addition, coming from a philosophical view, it is not possible to prove consciousness in another intelligence because this is only known by its possessor (Buttazzo 3).
One test that almost solves that problem is called the Turing Test. Created by Allen Turing in a 1950 article in the Mind Journal, this test tries to determine whether a computer is semi-intelligent. During the test, two people sit in separate rooms with computers hooked to the challenged computer with a chat program running. The two people converse with the AI and after a time limit expires, the people try to guess who the computer is. If the computer successfully fools all the test participants, it is deemed intelligent. Up to this day, no computer has fully fooled any human. So, until the day, the computer that is rated the best gets the prize money (Masci 6).
Another dilemma facing AI researchers is how to teach the robot. Several ways have been developed, but the one that seems more promising is called Top-down learning. Top-down learning was developed for Douglas Lenat’s robot Cyc. This is done by feeding the computer little bits of information that would seem silly to a normal person. For example, if you told Cyc that “Babies cannot drive”, “Babies go to daycare”, and “Speeding gets you a ticket”, it would figure out that “Babies cannot speed to daycare so Babies cannot get tickets (Masci 4).
One shining light in the world of AI is the company known as IBM. In 1997, IBM built Deep Blue, a chess playing machine that went against the world chess champion, Garry Kasparov. The six game match ended up with Deep Blue winning 3 to 2 against Kasparov. Even though Deep Blue may have won, it does not begin the robot’s domination of mankind. In fact, it is far from it, while Deep Blue calculated over 2 billion moves a second and seemed intelligent enough to play chess, it required humans to move and inputmoves into Deep Blue. One major part of consciousness Deep Blue did not have was self-awareness. When Martin Gardiner observed Deep Blue he said, “It knew no more it was playing chess than a vacuum cleaner knows it is cleaning a rug” (Masci 1).
Another thing Deep Blue did not have was the storage necessary to house an AI “brain”. A study done several years ago showed that “the human brain has 10^12 neurons and each neuron makes 10^3 connections for a total of 10^15 synapses. In artificial neural networks, a synapse can be represented with a floating point number, which needs 4 bytes of memory” (Buttazzo 4). By doing the math, the total comes up to 4.096 * 10^15 bytes needed or 4,096 petabytes. To put this in Layman’s terms, say that every computer has a 20-gigabyte hard drive in it. To make this amount of storage possible, you would have to link up about 2 million computers. A popular formula named Moore’s Law can tell us how much storage we will have in a specific year. Putting the above information into the formula, it comes out to the year 2029. An even scarier proposition is that the year 2029 is when the movie Terminator was set in (Buttazzo 4).
If an AI breakthrough is achieved, the world will certainly be a different place. According to Moravec, AI’s will be so much smarter and more capable than their human forebearers that they will quickly replace people in every endeavor from writing to computer science. However, many call Moravec’s ideas preposterous. Even scientists who envision the coming of AI say that humans will always have the ultimate control over their creations. Others, like Lenat, say that humans will use AI to enhance their own capabilities and thus will be able to stay one step ahead of the machines (Masci 2).
When looked at from this angle, the AI will not replace us, but rather help us as our friend and ally in the future.
Asimov, Issac. Asimov’s New Guide to Science. New York: Basic Books, Inc. 1984.
Bedell, Doug. “Rise of the Robots.” Dallas Morning News 18 May 1999:1F
Buttazzo, Giorgio. “Can a Machine Ever Become Self-aware.” Online. Internet. 28 Mar 2001
Chalmers, David J. “Facing up to problems of Consciousness.” Journal of Consciousness Studies. Online. Internet. 4 Apr 2001
Clarke, Roger. “Asimov’s Laws of Robotics.” Xamax Consultancy. Online. Internet. 27 Mar 2001 http://www.anu.edu.au/people/Roger.Clarke/SOS/Asimov.html
Kitano, Hiroaki Ph.D. “Direction of the relation between robots and humans.” Online. Internet. 28 Mar 2001 http://www.symbio.jst.go.jp/~tmatsui/pinodesign.htm
Masci, David. “Robots”. Congressional Quarterly 14 Nov 1997
Maulis, R. J. “Robotics” Online. Internet. 30 Mar 2001 http://student.fortlewis.edu/rjmaulis/Final.htm
Travers, Bridget, ed. Gale Encyclopedia of Science – Volume 5. Detroit: Gale Research, 1996.
Wilson, Jim. “The machines of Star Wars Episode 1.” Popular Mechanics June 1999: 76
This is my high school senior paper, so please ask if you are going to use anything out of it, I will usually be glad to let you use it.