Wednesday, January 18, 2017

On Rene Descartes’ Fifth Discourse on Method

In Descartes’ autobiographical work (or a philosophical work which has an autobiographical structure to it) he outlines why he thinks a machine cannot convincingly simulate the actions of a man while it easily could a non-human animal (beast)[1].  He surprisingly conjures the possibility of a machine which looks exactly like a man and even is programmed (my words, not his) to speak a given phrase or expression which would be appropriate and therefore convincing in one situation but would not be enough to become convincing as a human being generally.  And this is true.  We see this if we imagine Abraham Lincoln automatons outside Mt. Rushmore who stand up, tip their hat and say in an earthly tone, “I was born in a log cabin in meager conditions, yet by the age of…”
However, Descartes commits two faults here – one obvious one not so.  The obvious one is his idea that a machine cannot think.  AI is not an idea that strikes him which is understandable because of his historical limitations and therefore we can excuse this minor error of his definition and conception of ‘machine.’  His second error is in believing that an automaton could not be convincing in a litany of scenarios.
For just as a computer can simulate chess, so I believe I could be playing a player in China when in actuality no such player exists, a computer can simulate a litany of games.  And just as a computer can simulate a litany of games so a computer can replicate without thought a litany of responses to a plethora of different chunks of audio (meaningless noise to the machine but generating a response nonetheless – the same way the body automatically responds to varying temperature and atmospheric pressure) in a way found convincing. 
Like a chess game, the majority of conversations often follow a certain pattern.  Each line of dialogue from each speaker can be viewed as a “move” (though I’m not sure what taking a chess piece would be in this analogy).  Moving pawn to D3 is equivalent to “how are you” and so on.  A machine could even be designed to unthinkingly listen to vocal cues and tone from the speaker and respond based both on the content and the form the content takes.  For just as an animal can “unthinkingly” survive in a litany of situations without thinking the words, “oh shit, wilder beast,” so a machine can replicate thought without actual being sentient.



[1] I paused here in particular in order to show that, if there were such machines having the organs and the shape of a monkey or of some other animal that lacked reason, we would have no way of recognizing that they were not entirely of the same nature as thee animals; whereas, if there were any such machines that bore a resemblance to our bodies and imitated our actions as far as this is practically feasible, we would always have two very certain means of recognizing that they were not at all, for that reason, true men.

No comments:

Post a Comment