2008-04-17T00:32:00.003-04:00

Turing’s Concepts Breach the Wall of Artificial Intelligence

Ever since man invented the computer, scientists have hoped to someday make an “intelligent agent.” An Intelligent Agent is an automated device that acts with its environment, at which it learns from the inputs, that enters its database. Alan M Turing came close to finding a possible way to define Artificial Intelligence, but his theories may have impeded our progress toward realizing Artificial Intelligence’s full potential. This paper shows the different theories that have been introduced in Artificial Intelligence. It also discusses issues around the scientific community, becoming aware of the great, truly powerful ideas behind these theories, and scientist need to better understand the human mind to fully create Artificial Intelligence.

Researchers know little about the human mind, so to gain the knowledge to make a machine truly “intelligent” researchers need to more fully study and understand the conscious processes involved in the brain’s functions. Since the beginning of mankind, the human race has made every effort to learn from the sciences of today’s world. As man has evolved into the super race of today, the race has been interested and wants to know everything possible. The thirst for knowledge and understanding compels man to “strive” for some kind of perfection. Humankind wants to understand the reasons why what were once concepts have been proved accurate. Although, the one science that is the hardest to realize has become Artificial Intelligence (AI).

Alan M. Turing, a mathematician from London, made great strides in cracking spy messages for the military, and gained great knowledge about the potential that the computer could gain with further advancement. The first line that Alan M. Turing wrote in his article, Mind, “I propose to consider the question, ‘Can machines think?’ (Turing 433-435). Turing’s concepts have altered the world of computer sciences since October, 1950. The Turing Test, (originally “The Imitation Game”) was an idea that a machine could convince an interrogator that it was human. If the machine could in fact, prove that it was human, and then logically this would, in theory, make it intelligent. Then to pose this question, Is this intelligence, or just intricate clockwork? Robert M. French (French 2-3) has found that The Turing Test is by far the most popular among all theories, and has advanced the most discussion within the realm of Artificial Intelligence. The reason why Turing’s concepts are the most discussed is because his ideas are possible. Programming a machine thousands of responses is one thing, but then in reality being able to use these responses intellectually is another. The latter activity actually makes a machine intelligent.

Turing’s bizarre ideas is what made people start to believe that Artificial Intelligence is possible. Although, the struggle came up, how will this intelligent agent become modeled in such a way that it is plausible? That is the reason for his ideas to being so bizarre. It is impossible that a programmer could give a machine enough responses for it to take on any endless conversation. For over 50 years, researchers have tried to progress in the science of Artificial Intelligence but have always found themselves back tracking because of Turing's theories on Artificial Intelligence.

Now the question is, “What were Turing’s concepts on building an Intelligent Agent?” In more depth, The Turing Test is based on the observations of three people, a man, woman, and lastly an interrogator. The object of the test was for the man to convince the interrogator that he is a woman, while the woman will try and convince the interrogator that she is the real women by revealing true answers. What does this have to do with Artificial Intelligence? Turing’s plans were to replace the man with a machine (Intelligent Agent) and this is where the question, “Can machines think?” was formulated. When you take away a human and add a machine, this is where the interrogator makes the assumption. The interrogator’s job is to then ask both the human, and the machine questions. Then if the machine can “fool” the interrogator; that is, make him believe “it” is a human this, would make it intelligent.

A vital question needs to be explored while researching Artificial Intelligence: “How do we make a machine think as we do?” Marvin Minsky a professor of Massachusetts Institute of Technology, states in order to build a successful “humanlike” robot we would need to, “enable them to think in ways like ours.” (Minsky 1-3) He makes another point that, the troubles are to overcome this situation because even with all the knowledge man utilizes scientist still contains so little about how our minds work. Minsky has come up with many different models of the brain including; one, two, and three part models of the brain. Yet, the one other method that Marvin Minsky brought up is that the thought process of the brain can be looked at as a thought cloud. An example that Minsky formulates to look at this idea of a thought cloud was a “human community,” and all of the people have different occupations, and multiple intelligences. All of these humans have their own piece to put in the puzzle in order to make it complete, and Minsky says that, “…it may be better to think of a large network of smaller machines.” The one problem with looking at it in this representation is that it reveals how unclear the model of a thought cloud essentially is because it shows the large view but does not show the individual pieces of the “network” which are the “smaller machines.” Marvin Minsky’s main point is, “Our own human brains are too complex for us to envision all at once.” Therefore in order to find the underlying process of the brain, different models of the brain will need to be examined in order to piece the puzzle together and come out with the finished conscious process of the human brain.

So now ask the question again, ‘Can machines think?’ According to Hayes and Ford (Hayes and Ford 972-977) we should “abandon” the concept of The Turing Test and spend more research time on what they call “A General Science of Cognition.” This would be the focal point of capabilities such as, analogy making, vision, generalization, and categorization. Hayes and Ford do not follow the conceptual ideas of The Turing Test, because as they state that, “What extra functionality would such a machine provide?” There view is that of Marvin Minsky’s, that breaking apart into different models is significant to find our way to the bigger picture. Another fixation that Hayes and Ford bring up is they state that, “To what extent do machines have to act like humans before it becomes immoral to damage or destroy.” By this picture, they are assuming that if we reach a time in science where a machine can reproduce intelligence, will it be immoral to destroy it or “kill it.” This brings up Jamais Cascio thoughts on, should Artificial Intelligence or a “thinking machine” be able to obtain human rights. Cascio (Cascio 1) believes that researchers have not made any recent “breakthroughs” but “steady advances” have brought us closer to self aware, intelligent machines. So if we come to a point in time where machines become just aware as we are, how could it be moral for us to destroy them? When machine makes it to this point in time, this is where the movies Irobot, the Terminator and other futuristic Sci-fi movies come into play. In such movies the machine community becomes so intelligent that they find out they can rebel, and exist without the human races help, so along with their machinery we need to apply rules to them.

So, will researchers ever find a way to intelligent machines? According to Marvin Minsky (Minsky 2) he stated that, “Within a generation the problem of creating ‘artificial intelligence’ will be substantially solved.” Although later Minsky’s prediction quickly changed when artificial intelligence had, ‘undergone a radical shift’ when he stated that artificial intelligence is by far the most difficult ever ‘undertaken’ by science. Then the discussion on The Turing Test went from if a machine managed to pass the test would it be considered intelligent or would it even be possible for any machine to pass a test like The Turing Test. (Kolata 1237)

Since the human race, “strives for perfection” researchers are always trying to find the “right” way to do something. Marvin Minsky said in a speech that researchers have made large strides but stopped making progress since 1980. This is because researchers are trying to perfect Artificial Intelligence. Minsky said, “Say you have six pretty good ways of doing something, and if some of them don’t work maybe the others will.” (qtd. in Gadgetoff Artificial Intelligence) What he means by this is, if researchers stop looking for the perfect way of doing something, and find more pretty good ways of doing something; even if some fail we will have others to back them up.

Also Minsky (Minsky) believes that in Alan Turing’s test Turing did not pay attention to words like "mind" or "think" or "intelligence," but found that when a person applied those words to a computer, that is what made the test important. Turing’s ideas on the Turing test were just to fool the interrogator to believe that it was human, and it does this by ignoring what the interrogator says, and then applies a statement that may seem life-like.

The Turing Test is what is impeding researchers from progressing to realize an intelligent machine. When a child is born it has to go through a learning progress to become an adult. When a researcher builds a machine that attempts to pass the Turing test, they are trying to build it to fool an adult. But giving a machine enough intelligence to make an adult believe that it is a human is impossible. It is impossible because the machine learns whatever the programmer knows, but has no room for it to improve. Turing did have an idea for a learning machine (Turing 454), and this idea is the best yet. But in order to learn you have to have experiences and this is where researchers need to start from scratch. This means to start with a machine that attains a childlike mind, and as it grows older, it will learn new things.

Since all aspects of research within the Artificial Intelligence field are all ‘merely experimental’ knowing the correct method to gain succession towards a working Intelligent Agent is largely impossible at this moment. But to find the right way may not be the path to finding true artificial intelligence. Researchers will continue to stretch towards new models to learn what the human brain consists of, so we can find the right pathway to the future. To surpass this great science, keeping the mind open to new models and ideas is vital to come to where we live in a world where humans and intelligent robots coexist and hopefully co-evolve.

Works Cited

Turing, Alan M... "MIND" Vol. LIX. No. 236... October, 1950

French, Robert M... "The Turing Test the First Fifty Years." In Trends in Cognitive Sciences

04.03.2000 pp. 115-121

Minsky, Marvin L... "Future Models of Mind Machines." Massachusetts Institute of

Technology. http://www.media.mit.edu/people/minsky/

Hayes and Ford (1995) Turing Test considered harmful. In Proceedings of the Fourteenth

International Joint Conference on Artificial Intelligence, vol. 1, 972-977

Cascio, Jamais. "Artificial Intelligence, Real Rights."

http://www.worldchanging.com/archives/001878.html January 4, 2005:

Minsky, Marvin (1967) Computation: Finite and Infinite Machines. Prentice-Hall, p.2

Kolata, G (1982) How can computers get common sense? Science, 217, p. 1237

Minsky, Marvin. “Turing Test.” E-mail to Robert Winter January 31, 2007

Minsky, Marvin “Gadget off” Massachusetts Institute of Technology: Artificial Intelligence

2007, http://www.gadgetoff.com/2007/video-minsky.html

Inside the Mind

2008-04-14T12:50:00.005-04:00

Really, to make an AI machine we are trying to replicate the process that the mind produces. So it is vital to dissect the human brain (not literally), to understand what is going on in the mind. But the trouble is finding which part of the brain does what functions. I like to picture the mind being broken up into 3 different parts.

1. Emotion - This part of your brain would react off your senses, or experiences. Your body needs emotions to survive, and to gain moral life skills. From the womb, your senses come to play, different emotions that children feel, tell them how they react.
2. Rational - This part of your brain strays away from emotions, and uses intellectual, more sound ideas to deal with a situation. The mind needs this part of your brain to take things that your emotions set out and make them rational.
3. Motor Skills - This part of your brain takes the Emotion and Rational parts of your brain and sends out messages to your body so that your body can move and make motions to react in its environment.

What is Intelligence and what does it mean to a machine?

2008-04-13T22:34:00.007-04:00

So first, before implying that a machine is intelligent, we have to understand what "intelligence" really means. Intelligence is being able to learn and be able to build new understanding within your surroundings. Then being able to apply these understandings in order to manipulate ones environment. So first off this clears up a great deal of what Artificial Intelligence will be.

Now we need a machine to mimic the thought process that the human mind would go through (uses intelligence), and this would be to learn new things in order to survive within ones surroundings. The question has been for a long while, How do we do this? Also, What are different ways that we can make a machine safe, to be able to "co-evolve" with us?

Is the Turing Test the first step?

2008-04-11T11:11:00.008-04:00

The Turing test was a significant turning point for Artificial Intelligence. In 1950 Alan M. Turing formed the idea that if we can build a machine that can convince an interrogator that it is human, then in theory, this would make the machine intelligent. But is this the first step to take? No. Making a machine a fully intelligent is illogical. Being able to create such a machine that would hold all such information would take up more space than we have particles in our universe.

Turing was not wrong with his thoughts, I just think he was jumping forward more than he was. But what are the advantages of building such a machine. Well, truly there are none. Turing did not look into the aspect that the machine would really be intelligent, but looked at it as humans are actually fooled and this is what makes them apply the word intelligent to the machine.