In the Beginning was the Blog

Can Computers Have Knowledge?

I just handed in my final essay for philosophy looking at something close to my geek heart. I decided to approach the topic of artificial intelligence (a deprecated term now but still has meaning in the public mind at least) from an epistemological angle. Epistemology? That is the study of knowledge: what is it and how do we get it?

I can’t give you the whole paper until the results are in but I thought I’d run through some of the interesting stuff I came across during the research here. Having read through some of the introductory and historical stuff, I found that at least a couple of paper referencing a guy named Searle who wrote about a thing called the Chinese Room.

To understand the Chinese Room, you have to go back a bit into sci-fi land and the excitement that was being generated as the first large scale integration of electronics and general purpose computers was getting momentum. At that stage, scientists who were brought up on fantastic novels about robots and machines building the future started to think that maybe we had arrived. Research into AI was getting some promising results where computers were able to make decisions and even diagnose blood diseases. It was thought at the time that if we could sort out all the rules of decision making and thought that our own brains work with and then program them into a computer, we might find ourselves talking to Baum’s Tin Woodsman (Wizard of Oz).

Hence the Chinese Room argument in which Searle sought to argue that the kind of AI envisioned at that time (1980) could never really think the way we do. The argument is based on the Turing test which goes back to early AI research in the 50s. In the Turing test, typed messages are exchanged between two rooms. The human in one room must guess is the typist in the other room is a computer or human by typing messages and examining the responses. If the person cannot tell the computer from a human, then we can say the computer is intelligent.

In the Chinese Room experiment, a giant phrase book is generated in chinese with millions of responses to various Chinese phrases. I guess Chinese is chosen because it is meaningless to most English speakers. On one side is a native Chinese speaker. On the other side, there is either the computer with all the rules programmed in or a human with the rule book. Searle’s argument is that just as the human doesn’t learn Chinese in this experiment, so the computer doesn’t think or know Chinese.

So that’s the Chinese Room in a nutshell and next time I’ll talk about how we might move beyond this kind of “propositional” thinking because let’s face it. A world without Skynet would be a bit boring.

Some links of interest:
The cybernetic imagination in science fiction / Patricia S. Warrick. looks at the history of AI and robots in science fiction.