Chinese Room

Comments:
http://jngonzales-chi2011.blogspot.com/2011/01/chinese-room-blog.html
http://isthishci.blogspot.com/2011/01/blog-entry-1-minds-brains-and-programs.html


Reference:
Minds, Brains, and Programs
By John Searle
Venue: Behavioral and Brain Sciences journal

Summary:

Searle disputes the level of understanding in this entry.  He acknowledges that a computer program can generate the appropriate responses to a question about a  story, but not that the program actually understands the story.  To demonstrate this, he uses an analogy where he is the program and the story is Chinese symbols slid under a door in the room he is in.  He has no idea what the Chinese symbols mean.  The computer gives him instructions on how to deal with the symbols, and he slides them back under the door with the correct responses.  In this example, he generates responses that will pass a Turing test, but he actually does not understand anything he did.    He views "weak AI" as him generating the correct responses and believes that is in place.  He does not think "strong AI" is plausible, meaning the computer actually understands the answers before outputting them.  Searle uses the term "intentionality" as a simile to understanding.

Discussion:

I must say that I agree with Searle.  The computer doesn't truly intend to answer, it just follows what is programmed to do.  Even with machine learning, we are just giving more "rules" to the program itself.  The program cannot make a mistake and learn from it.  Until a programmer fixes the bug, the program will continue to produce the same results.  I don't think computers will ever possess "common sense," which to me is a key component separating a computer from the cognitive skills of a mind.  One could write a program that would symbolically make the computer willingly commit suicide. I don't think someone could manipulate a human so that they would willingly destroy themselves.  We may be able to create stronger and stronger AI, but I don't think we'll ever reach the level that Searle describes as "strong AI."