Okay, here’s what I came up with to prove that my Logic course instructor is a computer program and not a human. We were supposed to prove it logically, which I’m afraid I didn’t lay out that well and then specify fallacies we used, which I’m sure there are a couple, but ran out of time to enumerate them. Hopefully it is effective enough as it stands to count for a good grade. At 5 in the morning, I’m going to bed cause I have class tomorrow and I still should write up an article review.
I have become convinced that our course instructor is actually an AI and not a human being. An Artificial Intelligence is something which has been continually improving over the years. And now, in 2006, computer systems are powerful enough to enable an AI with sufficient capability that it can teach convincingly. We will look at how to test if an AI has intelligence, aspects of what an AI is and show why an AI is a good choice to teach this class.
The Turing test is a test to decide whether a computer is intelligent by having the computer chat with a human. If the human believes he is talking to another human when he is really talking to a computer, then the computer has passed the Turing test and is deemed intelligent. To be determined to be human (or at least have human-like intelligence), the instructor would have to pass the Turing test. A sufficiently modern AI would be smart enough and random enough to be able to pass or nearly pass the Turing test. Since the discussion format of our online class is not real-time conversation, it would be difficult to test the instructor with a Turing test. Conducting a Turing test would be possible in the chat room but has not been done so there is not conclusive evidence.
One type of AI is called the Expert System. An Expert System has a database of facts and rules of how to apply them. It uses an inference logic engine to process questions asked by humans, and its answers are all based on its prior knowledge in its database. An Expert System is an expert only in its field of knowledge and thus is highly domain specific and unable to help much outside of those topics. This could explain our instructorâ€™s propensity to repeat the same question to different people in discussion. The answer given by the student triggers a particular response to ask a specific question.
Because this class is an online class, we never get to meet the instructor in person. Indeed, we donâ€™t even expect to meet him. Since this is the case, it would be easy for the instructor to be a computer AI as it would not require a body. So, just install the AI instructor in a server someplace and give him some classes to teach and youâ€™ll be all set. Advantages of using an AI over a human include cost, accuracy, and replacement. Once initially built and set up, an AI would be basically free where the human requires an ongoing and increasingly expensive salary. AI does not age or need to be retrained as a human would. Additionally, an AI can teach many more classes at once than its human counterpart could. This would even explain why logic was not even offered as an on-ground class: there is no physical teacher to instruct the class in the classroom.
Factasia, a site which hopes to design a vision for the future, suggests that the core of a successful AI system is an inference logic processor (http://www.rbjones.com/rbjpub/cs/ai022.htm). It would be mostly concerned with logical truth rather than being worried about content. Since inference logic is what this class is all about, an AI built to the Factasia vision would be a perfect candidate to teach a logic class. Its qualifications would be stellar and AIO could easily justify purchasing or developing an AI to teach the class.
So, because of the overwhelming advantages of using an AI to teach the Logic class and the lack of proof that our instructor is definitely a human, we must conclude that we have been listening to an AI for the last five and a half weeks.
(I have this feeling that I did use a fallacy or two, but not intentionally)