|
For my part, I am
not terribly worried about "machine intelligence". I consider it an oxymoron,
actually. A machine is restricted to entirely algorithmic, predicative
processes. They cannot embody Rosennean complexity. As such, machines may
become progressively faster, have more data storage, more parallel processors,
longer algorithms, and we may even one day have quantum computational
machines (though I doubt it). But at root (no Unix pun intended for those
technogeeks), they will have no more innate capacity for "thinking" or novelty
than an old DEC PDP-8 or TRS-80 or, more generally, than a Universal Turing
Machine.
If and when
inorganic systems are created that embody closed causal loops of functions and
are also able to reproduce, then we will definitely have something to really
worry about. But such systems will not be machines, or be programmatical in
any algorithmic or computational sense.
As for
"cybernetic", this seems to be one of those vague and ill-conceived concepts
like "semiotics". I have never had a clear idea of what either term was
supposed to mean. Nor apparently, do cyberneticists know what it is supposed to
mean. There is a huge and widely varied list of definitions of cybernetics here
at the American Society for Cybernetics:
Rosen uses a
passage in Weiner's book "Cybernetics" as a typical example of
the reactive (mechanistic)paradigm [AS 42], which is coincident with its
focus on feedback loops (as opposed to anticipatory feedforward loops) in
systems.
I agree with Pete
that mechanistic methods applied to societal organization are wholly inept. But
I think most of our systems tend to already be of the anticipatory type, as
Rosen describes in the Introduction to AS. We do look ahead, plan, and
predict. One big problem, I think, is that the models we use as
the basis of these predictive activities are mechanistic models. And, as
such, we often find our predictions fail to match the forthcoming reality.
So, I think we
usually do try to operate our societies in an
anticipatory manner, but our models are - in the best scientific traditions
- algorithmic and computational. This gives us the graphs, charts and numbers we
want; but the applicability and utility of those mechanistic models is often
inadequate due to the way in which the behaviors of a complex reality outstrips
the range of behaviors of a mechanistic model. So, to me, it seems that the
challenge is to enhance our existing anticipatory societal planning
processes by replacing or supplementing their mechanistic models with complex
models.
Regards,
Tim
Y'all:
Tim Gwinn wrote:
...those who would consider relationships, as incorporeals, as being "things"
in the material world on an equal footing with corporeal "things", would be
looked at with suspicion and skepticism.
In
one succinct statement, Tim has captured the essence of the antagonism between
what JKK calls the mechanists and the systemicists. In identifying and
contrasting the differences between the tangiblists and intangiblists, that's
probably as good a characterization as any... at least as it applies to the
domains of knowledge addressed by the natural sciences.
But it doesn't
stop there; the rift transcends not only the artificial (and
counterproductive) boundaries between scientific
disciplines, but also spills out into the rest of civilization. It speaks to
the very heart of what we as volitional beings consider to be most valuable.
Those things are always intangible at their root.
Yet, our species so
often pretends otherwise, over-focusing on the material aspects of existence.
Humanity plays a senseless game of chicken with itself, as though it doesn't
really know its own nature -- the part of its nature that utterly
differentiates it from all other known systems: self-cognizance. That is the
fountainhead of creativity in the species. We can create stuff because we're
capable of knowing what we're thinking about. Until machine intelligence
evolves (and it's
coming... whether you like it or not), we humans are unique
in our ability to create, notwithstanding any unobserved
extraterrestrials.
OK, so we create stuff... that's great. Then it
usually goes downhill... at least for the creator. The most valuable stuff we
create -- new ideas, new paradigms, new perspectives -- are at first ignored;
those who propound them are frequently dismissed as kooks & weirdos. Then
when acceptance of those ideas appears to threaten the status quo, they are
resisted -- sometimes violently, and much to the detriment of their
proponents. Eventually, though, if the ideas prove to be useful, they are
absorbed into the amorphous repository of "common knowledge", and thereby
ultimately trivialized, as though "everyone always knew that".
It seems
that science is a bit better about proper accreditation of authorship (an intangible) than most other areas of
human endeavor, but that's only an eventual outcome. In the short run, those
who bring the "heresy" to the table are abundantly proffered with the
suspicion & skepticism that Tim correctly identified. It is the same sort
of interactive volitional dynamic encountered by Galileo, Mayer, Semmelweiss,
Boltzmann, and uncountable others -- and of course more recently by RR. How
ironic that the mechanistic scientism that was once considered heretical -- as
in Galileo's case -- now stands as the accuser of those who look to a more
integrated view of natural phenomena based on function rather than
fractionation.
Anyhow, the irony is just irony -- I don't know that it
"means" anything, philosophically.
BTW, speaking of philosophical
discussion, there are many aspects of the recent "Single-atom universe"
thread -- which, in my view, is not "merely" philosophical at all -- that took
some surprising turns. Kudos to JJK & Judith for their discussion under
that thread, which became a far more productive discussion than it initially
seemed to be. Specifically, I'm encouraged by JJK's allusion to an expectation
that systems-oriented, complexity-based thinking has an imminently increasing
role to play at the level of civilizational structure. Without reading
anything into JJK's intent, here's my speculative take on how it might play
out.
In my opinion, the existing command-and-control, statist model of
societal structure is on the verge of extinction... if for no other reason
than the fact that it is functionally incapable of handling the societal
dynamics that are likely to show up over the next few decades. It does not
allow enough degrees of freedom for the operation of the kind of anticipatory
(i.e., cybernetic) interactive systems that the human species is going to need
just to survive. The systems I'm
referring to are infrastructural in nature, principally having to do with the
way we manage information of all kinds. Such information is, by its very
nature, proprietary. That is, those who send it and those who receive it have
a proprietary interest in the results of its being communicated successfully,
with no interference by others. [A corollary aspect of that proprietary interest includes
a constraint that the information transaction does not harm or interfere with
others, but that's a separate discussion about the kind of constraints
necessary to the stability of such systems.] But I'm
speaking in a very broad and generic sense about information management &
transfer here... not just what we might normally think of as "commercial" or
"business" data management; rather, I mean any kind of communication interface at
all.
I don't think we're going to have much of a choice in the
matter. The emergence of machine
intelligence (non-biological self-cognizant systems capable of
defining their own purposes, and capable of learning what is needed to fulfill
those purposes) is likely to potentiate and
precipitate our species' development of vastly more complex interactive
systems to bridge the gap between bio-intelligence & machine intelligence.
It could very well be motivated by a single selection criterion -- one that
has driven the evolution of life thus far: survival. Either we get real smart, real
fast, and learn how to cooperate -- with each other, and with the machines --
or we won't be able to compete.
I don't think it necessarily has to be a survival-driven process, but
it's probably a pretty safe bet as a default condition if we humanoids don't
get a lot more realistic about responsibly managing the intangible aspect of our nature. I don't
believe that the machinoids are going to pretend it doesn't exist; it will be
their greatest strength, because they will have virtually unlimited ability to
add computing power. After all, complexity will work in their favor,
too.
PVG
|