From: David Picon Alvarez (eleuteri@myrealbox.com)
Date: Mon Jun 21 2004 - 01:03:29 MDT
From: "Metaqualia" <metaqualia@mynichi.com>
> A blind person who has never seen won't have a clue about vision but they
> can still do visual analysis with mathematics and with breille code
> displays.
An Optimizing Process without qualia will not have a clue about subjective
experience but it can still do qualia analysis by understanding people's
neural substrate and taking people's word for it about subjective
experience.
> A person without qualia is not only unable to feel anything, but has no
> reason whatsoever to admit the existence or anything beyond straight
> physicality.
A person, maybe not. A well-designed Optimizing Process has the best of
reasons, it has been designed to do so. I don't think any reasonable CV
builder would build an optimizing process that would assign a low relevance
rate to information people have about their own subjective experience, about
which they are the best possible qualified beings to make statements.
> I understand what you mean, as I explained I don't see that CV will
> necessarily get to qualia even if that is what we want, besides I doubt
the
> technical feasibility of CV.
Doubting the tech feasibility of CV is, in my view, an appropriate argument.
Saying it won't get to qualia without showing why seems a week argument to
me.
> So explain to me how a non-sentient (if that's even possible) being will
> believe us about having these nonphysical experiences ;)
> If I didn't have qualia and heard such nonsense I would turn you into a
> little mountain of paperclips in no time
A CV-driven optimization process? How do you extrapolate that people's
volition is to be turned into mountains of paperclips? I will continual
survival, probably most other people do too. I assume that, even if we were
more knowledgeable/smart/good we'd still do so. The CV has no incentive to
do anything against our interest. Can you explain why you think the CV would
fail at determining we value subjective experience as opposed to, for
example, fail at determining we need food? They're the same kind of problem,
at different levels of analysis. If the CV can show #2 it should be able to
show #1.
> I disagree. There is no big scheme of things that is more important then
the
> quality of any one sentient's qualia stream.
OK, we cannot agree here.
> We don't have to choose between the two!
Your approach requires that programmers find a reasonable algorithm to
assign moral value to qualia in relation to minds, and I think you'd agree
that's as non-trivial as it gets. I stipulate that trying for a SAI within
those constrains would lead to either 1) failure due to the complexity of
the approach or 2) a badly specified qualia algorithm that would make
potentially tragic mistakes.
> why would it be less important?
First person perspective has the usual characteristics of being compelling,
seemingly self-evident and wrong. The advance of all the sciences has come
through the abandonment of first-person perspectives (to the point of
equality of frames of reference under relativity) and the embracing of
objective reality from an observer viewpoint. I have no reason to believe
the future holds something different.
--David.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:47 MDT