[Sorry about the lack of snips... I want to preserve the context here.
This is a fascinating discussion]

SMChristenson <smchris@visi.com> wrote in message news:<pan.2004.04.09.02.08.14.381998@visi.com>...

On Thu, 08 Apr 2004 13:28:14 -0700, Andrew Lias wrote:

You're bolder than I am. Two and a half centuries is just too far out
for me to make any sort of guess. One thing that I have noticed about
previous predictions is that they are always a combination of
predictions that are too meek as well as too optimistic. History is
messy and, beyond a very short horizon, very difficult to predict. There
is no science of futurology.

Oh, I hear you "Vinge-wise". [BTW, he was the first author I "heard" in
an online panel before hearing him at a live panel. :) A bit of a "True
Names" experience there.] But I feel pretty confident in that
prognostication. Two reasons:

1) I'm only setting a minimum. Who knows what the maximum might be? We
could be in for another Dark Ages.

2) In truth, I know I'm no expert, but I _do_ have a psych major, make my
living in computer databases and such and have a foreign grad honours
degree in epistomology where weird speculative ruminations like, "What is
it like to be a bat?" count as serious journal articles.

No, It'll be a LOOONGG time. I'm futurist enough to think philosophers
like John Searle who claim you just can't make an artificial consciousness
are old fuddy-duddys but philosopher enough to recognize that he has a
point that people grossly underestimate the challenge.

And you can't just map 3-d electrical impulses and move them into silicon
and "see what happens". You basically have to know how to create an
artificial person _first_ in order to know how to move a consciousness
onto silicon or whatever substrate. I figure it'll get perfected about the
same time controlled fusion is relegated to a high school physics
experiment.

My gut suggests that you're right. However, I've looked at history
and 250 years actually is a *long* time in terms of technological
progress. Long enough that it's impossible to know what breakthroughs
will occur between now and then.

I am, however, in perfect agreement that the AI community regularly
underplays the challenge. Beyond the points you've already mentioned
is the fact that neurons aren't simple threshhold gates. Individual
neurons are complex entities in and of themselves. Likewise,
neurology, alone, provides an insufficient model of cognition. A bit
part of the cognitive process is the bath of chemicals that our brain
finds itself in. The brain is not easily sepparable from the body.

Of course, this is making the assumption that AI will following a
biological model for cognition. Although I doubt it, there may be
other, more efficient ways to achieve human equivilent intelligences.
Again, I'm doubtful but, honestly, who knows?

Suffice it to say that the task is utterly daunting and I wouldn't be
surprised if we might even be looking at something like a 500 year
timeline but, on the flip side, I find myself looking at some of the
problems which were considered intractable two hundred and fifty years
ago.

My general view is that any predicition of the future further than
fifty years out is, in fact, useless.

I agree with that. I particularly like the notion of Intelligence
Amplification as a good step towards the Singularity rather than the
wholesale construction of greater-than-human AI. You don't even need to

Yes, problems there. I'm not an entertainment snob. I think Terminator
and HAL are easily as plausible as Commander Data in terms of prototype
sentience.

I'm thinking more along the lines of machine/human interfaces of some
sort. The brain is already an exceedingly sophisticated computer.
Why not take advantage of that fact rather than reinventing the wheel?

get into the messy debate about genetic engineering. As Vinge has
pointed out, simply improving our ability to retrieve and organize
information has the practical effect of increasing our intelligence.
From that perspective, even Google is an IA device.

Yes and no. In some ways, it's like longevity. We can extend mean
longevity through public health administration and perhaps a society can
increase mean practical intelligence through shared knowledgebases and
decisionmaking. But one is still bumping up against a maximum, don't you
think without going transgenic?

I believe that you are correct but I'm not sure where, exactly, that
limit is. There's a lot of power in being able to effectively
distribute and collate knowledge and I don't think that we're anywhere
near the limits of efficiency and effectiveness. It may not even be
necessary to go transgenic at all if our agents become sufficiently
advanced as to allow us to get past those AI bottlenecks that we were
discussing.

Bruce Sterling probably had an insight in writing Schizmatrix. When we
start colonizing the Solar System, populations will be isolated enough
that they can artificially evolve without a lot of social disruption
within groups.

Well, yes except that I'm not optimistic that that's going to be
happening anytime, soon. Solar colonization isn't going to happen on
the basis of good intentions. There needs to be a strong economic
motive to doing so. Given that very steep expense of getting into
space, that puts a very high bar on such efforts, especially since
most of what you can do out there can be done better or more cheaply
by robots (particularly since robots don't need life insurance). As
good as dispersal would be for humanity in the long run, the short run
bottlenecks are daunting, to say the least. At least until someone
comes up with a cheap way to get people into orbit and a good reason
to keep them there (and let's not bring up beanstalks, yet... that's
another long term investment that's battered by the short term
expenditures).

--
Andrew Lias
http://andrewlias.blogspot.com

[Sorry about the lack of snips... I want to preserve the context here.
This is a fascinating discussion]

My gut suggests that you're right. However, I've looked at history and
250 years actually is a *long* time in terms of technological progress.

I am, however, in perfect agreement that the AI community regularly
underplays the challenge. Beyond the points you've already mentioned is
the fact that neurons aren't simple threshhold gates. Individual

Of course, this is making the assumption that AI will following a
biological model for cognition. Although I doubt it, there may be

I'm thinking more along the lines of machine/human interfaces of some

limit is. There's a lot of power in being able to effectively
distribute and collate knowledge and I don't think that we're anywhere

near the limits of efficiency and effectiveness. It may not even be
necessary to go transgenic at all if our agents become sufficiently

Bruce Sterling probably had an insight in writing Schizmatrix. When we
start colonizing the Solar System, populations will be isolated enough
that they can artificially evolve without a lot of social disruption
within groups.

Well, yes except that I'm not optimistic that that's going to be
happening anytime, soon. Solar colonization isn't going to happen on
the basis of good intentions. There needs to be a strong economic

space, that puts a very high bar on such efforts, especially since most
of what you can do out there can be done better or more cheaply by
robots (particularly since robots don't need life insurance). As good

comes up with a cheap way to get people into orbit and a good reason to
keep them there (and let's not bring up beanstalks, yet... that's
another long term investment that's battered by the short term
expenditures).

I don't discount the possibility of an educating class surviving in the
U.S. though -- with perhaps an ugly connotation on "surviving". We'll
just lose the initiative and have to travel to Taiwan for our enhancements.

The birthrate of the educated (high IQ) class is below replacement,
while the stupid are proliferating. And America's major immigrants are