Artificial Intelligence

Anthropomorphism- The ascribing of human motivation and characteristics to inanimate objets, animals, or natural phenomena.

For centuries, these two words have been a large part of scientists and artists lives. Scientists have been recently trying to create lifelike behavior by computer programs in a digital environment. Artists also try to create life, but the life they create is strictly on paper, canvas, or in sculpture. The artificial life is to be determined by the observer, the life is within themselves. By the 18th century, inventors and clock makers were creating mechanical creatures. For example, Jacques de Vaucanson invented a mechanical duck, that could flap its wings, eat, and excrete waste. This was a small step toward artificial life.

In the summer of 1956, researchers at Dartmouth College met to lay the ground work for artificial intelligence. In the last 40 years, it is believed that we have traveled only a short distance to that goal. But, people are looking at how far from the goal we are, rather than how far we have come. Simple tasks of reasoning for the human being need to be written in miles of code, each simple rule coded as hundreds of complex statements. Although, this roadblock has been scaled by many inventions.

The program titled Mycin, developed in 1975, passed the accuracy of the average physician in diagnosing meningitis in patients. Even though computers with artificial intelligence could take the rule of physicians in the future, despite accuracy and steady hands, they could never take the place of human ingenuity and improvisation. If a problem that has not been coded presents itself, will the artificial being be able to handle the situation by way of previous rules? Obviously, they would not be able to improvise something that is not in memory or quickly accessible.

Another question is raised, how do you teach a program common sense? How do you teach it to understand dialect? Would an artificial being ever be able to derive the subject of a conversation arriving half-way into it? These are minute things that humans do without thinking about it. You cannot teach a computer human emotion. It would be impossible for it to comprehend beliefs and faith. Most of the being's knowledge would contradict one another. It may know that Frankenstein was a monster, but also know that monsters are not real. These type of contradictions would bring many errors in problem solving.

A new type of intelligence that is beginning to be developed is materials that can anticipate failure, repair themselves and adapt to the environment. Many materials are used to animate inanimate objects, such as actuators and motors to simulate muscles, sensors that serve as nerves and memory, and communications and computational networks that represent the brain. The actuators are the materials that allow the object to adapt to the environment. They can change shape, stiffness, position, and other characteristics in response to temperature or electromagnetic fields. The use of these, and other devices, are to make intelligent objects, like a bridge, for example. The bridge would adjust itself when the weight is great or little or when the weather is bad. It can alert a watch guard when the structure may fail and can also repair itself.

There are four common actuators used today. They are shape-memory alloys, piezoelectric ceramics, magnetostrictive materials and electrorheological and magnetorheological fluids. Although these materials are being tested and developed for use in intelligent objects, they still have drawbacks. Shape-memory alloys adapt according to temperature change. They are slow due to the rate of change of temperature. Piezoelectric ceramics are quick acting by changing with the amount of voltage applied. Despite their high-speed reactions, nearly in the thousandths of a second, they do not exert the same amount of force that shape-memory alloys can. Even though, they are being used in optical tracing devices, magnetic heads, and adaptive optical systems for robots, inkjet printers, etc. An actuator that reacts to magnetic fields are magnetostrictive materials. This type of actuator is similar to piezoelectrics, it reacts quickly by rotating until it lines up with an external magnetic field. One common magnetostrictive substance is Terfenol-D. Terfenol-D is used in low-frequency, high power sonar. The final kind of actuator is made up of electrorheological and magnetorheological fluids. The micron-sized particles in this substance forms long chains when placed in an electric or magnetic field. All of these would provide an object with the ability to adapt to certain situations, considering it would be perfected and controlled by a computer brain.

Improved sensors can give an artificial being near the same sight and touch that a human has. Since the development of fiber optics, clearer visions can be seen. Giving an artificial being fiber optic vision, it would receive light nearly the same way that we do. Also, piezoelectirc polymers would give a being a unique sense of touch. The film that the polymers form and the sensitivity of them would allow the being to read Braille and distinguish grades of sandpaper. Polymers nearly 200 to 300 microns thick might replicate human skin.

Although all of these together may form a being, we must observe how all this data is manipulated. Neurons do not send every message they receive to the brain. Some messages that require quick reaction are sent to the spinal cord for a decision. Therefore, not all messages need to be sent to a central processor, some may be transported to some smaller decision maker that can handle simple problems that deal with reaction. Our modern day silicon chips are much faster than our neurons. But neurons can perform complex actions rapidly due to efficient networking. This is difficult to do with programming.

The artificial neural network is an attempt to solve this problem. This program mimics the functions of real neurons. This program can anticipate needs and correct mistakes. It would eliminate the need for the sensors to report directly to a central processing unit.

The largest step to date is a project headed by a professor, Rodney Brooks, that had violated the most important rule in the development of artificial intelligence. He built a robot that could maneuver through a cluttered room before it could reason. But, soon his idea caught on and other researchers followed his footsteps. Now he has taken a step beyond his robot that couldn't reason only move, he is developing Cog, a humanoid robot. After its hardware is complete, it will be turned on and it will be "born." Brooks provided the robot with pre-programmed desires, only, and hopes that it will learn its reasoning abilities as time moves on.

His main idea for Cog was to move away from mental maps and replace them with simple preprogrammed behaviors. He called a meeting with his 12 students\assistants to discuss consciousness. After hours of discussion, the group had decided to begin to build Cog and let the question answer itself. Brooks plans to have hormones built into its brain, he expects that some time after it is "born" it may recognize someone as its mother.

Although all of this is far down the road, we are much closer to artificial intelligence than most would like to admit. But the main roadblock is this: Will this created being ever have consciousness? There are many unknown and complex processes of the human brain that we may not be able to duplicate. All we can do is wait, see, and discover. Maybe someday Cog will pass the Turing test and prove to the world that artificial intelligence is possible, or maybe artificial intelligence is impossible and consciousness is only possible in biological creatures. But, of course, the question remains: what is consciousness?

Bibliography

"The Bad Boy of Robotics." Robert Langreth. Popular Science, June 1995.

"The Pursuit of the Living Machine." Simon Penny. Scientific American, September 1995.

"Intelligent Mateials." Craig A. Rogers. Scientific American, September 1995.

"Artificial Intelligence." Douglas B. Lenat. Scientific American, September 1995.

"Artificial Intelligence." Microsoft Encarta 96 Encyclopedia. (c) 1993-1995 Microsoft Corporation. Funk & Wagnalls corporation.

The Age of Intelligent Machines. Raymond Kurzweil. Massachusetts Institute of Technology, 1990.

Thinking Machines. Igor Aleksander and Piers Burnett. Alfred A. Knopf, INC. 1987. New York.