On 13 Feb 2013, at 20:44, Craig Weinberg wrote:
On Wednesday, February 13, 2013 12:46:23 PM UTC-5, Bruno Marchal
wrote:
On 13 Feb 2013, at 17:35, Craig Weinberg wrote:
Wouldn’t Simulated Intelligence be a more appropriate term than
Artificial Intelligence?
A better term would be "natural imagination". But terms are not
important.
Except that we already have natural imagination, so what would we be
developing? Replacing something with itself?
Yes. That's what life does all the time.
The distinction between artificial and natural is artificial. Human
made. And so it is also natural, as all creatures tend to do that by
developing their ego.
Machines are just a collateral branch of life. Cars and houses are not
less natural than ribosomes and mitochondria.
Thinking of it objectively, if we have a program which can model a
hurricane, we would call that hurricane a simulation, not an
‘artificial hurricane’. If we modeled any physical substance,
force, or field, we would similarly say that we had simulated
hydrogen or gravity or electromagnetism, not that we had created
artificial hydrogen, gravity, etc.
Assuming those things exist.
Whether they exist or not, the mathematically generated model of X
is simulated X. It could be artificial X as well, but whether X is
natural or artificial only tells us the nature of its immediate
developers.
It depends on how you defined Hurricane, and different definition will
make different sense in different theories.
By calling it artificial, we also emphasize a kind of obsolete
notion of natural vs man-made as categories of origin. If we used
simulated instead, the measure of intelligence would be framed more
modestly as the degree to which a system meets our expectations (or
what we think or assume are our expectations). Rather than assuming
a universal index of intelligent qualities which is independent
from our own human qualities, we could evaluate the success of a
particular Turing emulation purely on its merits as a convincing
reflection of intelligence rather than presuming to have replicated
an organic conscious experience mechanically.
Comp assumes we are Turing emulable,
Which is why Comp fails. Not only are we not emulable, emulation
itself is not primitively real - it is a subjective consensus of
expectations.
It is a well defined arithmetical notion, which comp assumes.
and in that case we can be emulated, trivially.
Comp can't define us,
That's correct.
so it can only emulate the postage stamp sized sampling of some of
our most exposed, and least meaningful surfaces.
You can't know this. We have to bet on some level, and cannot be sure
it is correct. But the consequences of comp are extracted from the
mere existence of the subst level, not from the (impossible) knowledge
of it.
Comp is a stencil or silhouette maker. No amount of silhouettes
pieced together and animated in a sequence can generate an interior
experience.
You can't say that publicly. You can't pretend to know that. It is
your non-comp *hypothesis*.
If it did, we would only have to draw a cartoon and it would come to
life on its own.
That's a non sense. Even for doing something as "simple" as Watson or
big blue, it takes a lot of work.
To assume this being not possible assume the existence of infinite
process playing relevant roles in the mind or in life. But it is up
to you to motivates for them. The problem, for you, is that you have
to speculate on something that we have not yet observed. You can't
say "consciousness", as this would just beg the question.
It is consciousness, and it is not begging the question, since all
possible questions supervene on consciousness. Not sure what you
mean about infinite processes or why they would mean that
simulations can become experiences on their own.
Because any processes finitely describable is trivially Turing emulable.
The cost of losing the promise of imminently mastering awareness
would, I think, be outweighed by the gain of a more scientifically
circumspect approach.
Invoking infinities is not so much circumspect, especially for
driving negative statement about the consciousness of possible
entities.
What infinities do you refer to?
The special one you need to make sense of non-comp.
Putting the Promethean dream on hold, we could guard against the
shadow of its confirmation bias. My concern is that without such a
precaution, the promise of machine intelligence as a stage 1
simulacrum (a faithful copy of an original, in Baudrillard’s
terms), will be diluted to a stage 3 simulacrum (a copy that masks
the absence of a profound reality, where the simulacrum pretends to
be a faithful copy.)
Assuming a non comp theory, like the quite speculative theory of
mind by Penrose. Your own proposl fits remarkably ith comp, and some
low level of substitution, it seems to me (we have already discussed
this).
Sense contains comp, by definition, but a comp world cannot
generate, support, or benefit by sense in any way as far as I can
tell.
There is no comp world. Comp is an assumption. It assumes that the
brain is Turing emulable, meaning that consciousness is maintained.
Then whatever reality is, it is NOT computational. Arithmetic, with
the big A, is already not computational, not axiomatizable, and quite
beyond us. And necessarily so when we assume comp.
Comp does not just contradict materialism. It contradicts the 19th
century conception of machines and numbers. Even just its "Church's
thesis" part does that. We don't know what numbers are, still less
what they are capable of, still less what they are not capable of.
Bruno
http://iridia.ulb.ac.be/~marchal/
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at http://groups.google.com/group/everything-list?hl=en.
For more options, visit https://groups.google.com/groups/opt_out.