Thursday, February 09, 2006

Knowing the Consciousness of a thing

People tend to have a mystical notion of consciousness, but let's get real-- consciousness is a psychological phenomenon like memory retrieval or planning. People like to talk to me about whether or not computers can be conscious. I think that someday they will, if they're not already to some degree, but the problem is we understand so little about what consciousness really is.

We don't know to what degree we are conscious, or of what exactly. Are you conscious of what someone says while you're engrossed in a TV show, and 10 seconds later you answer the question? Who knows? Are dogs conscious? Chickens? I'm sure you have an opinion, but I know it's not based on anything scientific, because scientists don't know what consciousness is well enough to say.

But let's assume that whether some being is conscious of some process is an empirical matter. That means there is some objective way, in principle, to determine if a being is conscious of something. This is reasonable, unless you're a mystic-- it's some high-level brain activity.

If this is true, the the consciousness of a particular AI program is also an empirical issue. Because let's be fair. Whatever methods we use to determine consciousness in a mink are those methods which should be used on computer programs or systems. This brings to mind the question of what those empirical methods are. The truth is we don't know. There are ideas out there, but there's nowhere near agreement on this. However the lack of empirical method does not mean that it is not an empirical issue. Whether a given computer program or system is conscious or not is a matter of fact, not of definition. Or at least it's primarily a matter of fact, just as the fact that I am tall is a matter of fact, and only in some sense a matter of definition. My height is objective. I think the degree of consciousness will be like height.

No comments: