Re: [sl4] Uploading (was : goals of AI)

From: Stathis Papaioannou (stathisp@gmail.com)
Date: Thu Dec 03 2009 - 04:54:06 MST


2009/12/3 Bradley Thomas <brad36@gmail.com>:

> The example of the Universe you give is interesting but I don't think it is
> analagous because by definition the Universe is not dependent on its
> environment. Every conscioussnes that we can conceive of, is. Even if we
> stretch our definition of consciousness to include these systems without
> input, no real system is immune to the effects of the environment... E.g.
> cosmic rays, quantum effects etc. There are going to be perturbations.
>
> But lets even assume that we can somehow prevent that (with shielding for
> example). You're saying that it is ok they run at different speeds. But if
> we have two essentially identical intelligent agents, lets call them Forex
> Trading Black Boxes A and B, similar in every way except that B is
> ever-so-slightly faster, clearly B has an advantage.
>
> Even if we assume that these two programs can eternally run the exact same
> code-sequence, they still differ in their respective performance. This is
> why again I make the point that I don't think it makes sense to say the
> identity resides purely in the code. The physical implementation makes a
> difference. To outside observers, these implementations display different
> behaviour. These agents have different personalities. If my brain ran at
> double speed, people would notice the difference. I'd be some kind of
> superhero or at least I would drive my wife nuts twice as often.

There wouldn't be a problem if the programs were an inputless virtual
environment with conscious inhabitants. You could run it on any
general purpose computer with enough time and memory and the result
would be just the same, independent of the actual hardware. If you are
living in such a virtual reality now there is no way you could know
anything about the details of the hardware you are running on, or
notice any change if the hardware is changed.

-- 
Stathis Papaioannou


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:05 MDT