The Conscious Internet - Part V: Deciphering the Logic

A while back a wrote a short article titled "the Conscious Internet" concerning the development of AI and computer technology in regards to the Internet. The article is written with a very philosophical approach to the subject, but handles real life facts. It has long been my intention to publish it here on the blog, but I just haven't gotten around to doing so. Until now ...

Here's part 5 of 8. You can find the previous chapter here. Happy reading, and please comment below.

Deciphering the Logic
When discussing cognition and artificial intelligence, one does not get far before encountering one of two major theses regulating the subjects; the Turing Test and the Chinese Room Theory. Computer experts around the world usually adhere to one of the two models, spending countless hours trying to prove its principles. 

The Turing Test was formulated in 1950 by the British mathematician Alan M. Turing, as a way of proving or disproving whether a machine was intelligent or not. The test places three different people; a man, a woman, and a interrogator, in separate rooms, with no contact with each other except for a link through which they can send messages - for instance via using a computer. The interrogator then asks the other two questions in order to determine which recipient is who. 

At an unknown point during the test, one of the two being interviewed is replaced by a very clever machine, programmed to answer the interrogator’s questions in the best of human behavior. The question Turing was asking himself was if a machine could be so smart enough that the interrogator would not be able to tell that a switch has been made.

Turing pointed out that the test alone was not created to, nor was it even suitable for, determining whether a program was intelligent or not. He saw the possibilities of a brilliant machine failing the test if its own intelligence differed too much from human intelligence. Or how an ingeniously programmed, but in essence fairly dumb machine, would be able to trick the interrogator into thinking it is still talking to a human counter-part. 

So far no machine has been able to pass the test. The best programs last for about a minute until they finally trip over themselves and repeat a sentence word by word. The program does not only need to be as quick and witty as its flesh and blood counterpart, it also needs to be as dumb and irregular. Some programs have also failed due to the fact that they were too quick, and gave back to accurate information, making the interrogator suspicious. Artificial intelligence also have to balance artificial unpredictability, and artificial stupidity, in order to be truly intelligent.

The Chinese Room Theory formulated in 1980 by John Searle, a professor in philosophy at UC Berkeley (University of California - Berkley), is based on almost the same thinking as the Turing Test. One test subject, who doesn’t know Chinese, is placed alone in a room with only a Chinese instructions manual to keep him company. Messages in Chinese would then be sent into the room, and the test subject is then asked to answer the message using the manual. 

Every message sent into the room would have a corresponding answer, given by the manual, which the test subject then would pass on. Searle argued that regardless of which messages was sent in and out of the room, the person decoding would never gain any knowledge of what was relayed simply because he did not know Chinese. 

The relationship with artificial intelligence is here quite obvious. A computer does not understand the code that is being fed into it, nor does it know how to decipher it, the computer simply knows how to react to the programming. It will therefor never, as a cognitive being would, be able to reserve itself against code it did not like, regardless of how illogical it may be. The computer would in other words jump of the bridge, if only it was programmed to do so.

... continues in Part VI: The Bright Side of Life