Wednesday, April 29, 2009

The worlds inside my head

Maybe it's the weather, but I'm not really in any mood to write coherently. Nonetheless, there's something I want to say, so I'll just go with the flow and hope you can forgive things like rambling, poor organization, and general incoherence. Let's start off with a little question: what's real? There are various different answers, of course. I don't know whether there's such a thing as objective reality -- in fact, it might well be impossible to know whether there is. I take it as an axiom that everything I know (or think I know) is based on my perceptions -- I don't believe in a priori knowledge. So, not "I think, therefore I am", but rather "I can hear myself think, therefore I am".

Then let's forget about objective reality for the moment. Subjectively speaking, what is real? I've already begged the question -- I'm assuming that my subjective reality is determined by my perceptions. Now, of course, I have a problem, because I perceive things all the time that are, intuitively at least, not real. I might think I heard someone call my name, or misread a word, or dream. What then? The first answer that comes to mind is that, for the moment, even those illusions are real. It's not until I realize I was wrong, until I wake up, that I'm able to perceive the difference between what I thought was true and what was actually true, and until I perceive that difference, it doesn't exist. If I start hallucinating and never stop, that hallucination is my new reality.

The upshot of this is that I believe it's reasonable to say that a completely convincing imaginary world is no more or less real than what we think of as actual reality. For instance, suppose that telepresence* technology advances to the point where I can't tell the difference between meeting someone in person and meeting them through teleconferencing. Then it doesn't seem all that crazy to say that I've met someone "in real life", even if I haven't actually physically been in the same room with them. In mathematics, this is called extensionality -- two functions are extensionally equal if when given the same input they give the same output. I'm sure there's a nice philosophical term for it too, but I don't know what it is.

If you guessed that all of this was just an excuse to fantasize about future technology, you'd be more than half right. From this perspective, an AI capable of passing the Turing Test is basically human, and sufficiently realistic augmented or virtual reality is just as good as actual reality. Bringing it back to the present day, consensual hallucinations like the Internet actually exist, not just as side effects of networks and displays, but as parallel worlds generated by the belief of their users. Less convincing illusions like the worlds inside of novels and movies are slightly less real, little toy worlds in our heads that we can start, stop, rewind, and reshape to some extent. The more the world seems to have an existence of its own, independent of the will of the perceiver, the stronger the illusion -- and thus the reality -- of reality. There's no magical line past which a virtual world suddenly pops into existence; rather, it was there all the time, slowly growing more real as it became more convincing. No matter its substrate -- atoms, words on a page, bits in a computer's memory -- if it seems real to me, then it is real to me.

I've run out of steam for the moment, but I greatly enjoy thinking about this topic, so I'm sure you'll hear about it again if you stay tuned. Again, my apologies for the incoherence.

*Telepresence, as sci-fi as it sounds, is just a fancy word for real-time communication methods such as telephone, video conferencing, instant messaging, etc., that allow people to give the impression of "being present from a distance".


  1. Looking forward to the next entry Wil. Got a couple questions though.

    What does it mean to be human? And since our perception of reality is subjective to our cultural upbringing, so what kind of reality will an AI perceive?

  2. Ooh, good questions!

    Subjectively, to be human is to act in a way that one believes a human would act. In other words, if I talk to an AI and I can't tell that it's not human, then it's human. Of course, that's all based on my own ideas of the ways in which humans would and wouldn't talk.

    And... would an AI perceive? Would it really think, or just fool me into thinking it thinks? These are sufficiently complicated questions that I think I'll save them for another post. Look for it in the near future!

  3. Gonna be keeping watch here for sure :) I always like reading other people' ideas.

    I find it interesting that you consider what determines a being human as how they act and not what makes them function (culture vs. biology). Most people I talk to always bring in the biological side of things, but of course, that can be related to the peeps I hang with.