From: Robert J. Bradbury (bradbury@www.aeiveos.com)
Date: Tue Sep 07 1999 - 21:27:13 MDT
On Tue, 7 Sep 1999, den Otter wrote:
> ... commenting on Greg's comments on
> > -The Transhuman Beach Party- (a very short story)
>
> Afaik, the Singularity = superintelligence, and the consensus currently
> seems to be that AI will beat IA (and uploading) in the race for SI.
I'm not sure that this is the "consensus". If we got the exponential
growth of nanoassembly tomorrow, we would have *neither* (a) the designs
to build stuff, nor (b) AI to populate the nanocomputers.
It is distinctly possible, and IMO *probable* that we will get
nanoassembly and molecular nanocomputers without an abundance
of nanodesigns or any comprehension of how to make an AI.
The "art" of engineering atomic scale structures and AI science
would have to get on curves similar to that of DNA sequencing
of the last 5 years, for the nanoassemblers to have lots of
stuff to build or for us to have an AI to populate a nanocomputer.
I don't see that happening right now.
Eliezer may pull the AI rabbit out of his hat, and maybe that
will solve the nanoscale design problem but that involves
a sharp discontinuity from current trends. If we want to
discuss probabilities, it is more probable that AI, robotics,
intelligent agents, etc. will step by step reproduce functions
of the human mind (playing chess, driving a car, speech,
speach-to-text, speech command, speech comprehension,
OCR, OCR with document comprehension, etc.). [I'm using
"comprehension" loosely here, more in the sense of grammatical
comprehension, rather than concept comprehension.] Engineering
libraries of nanoscale designs will slowly be built up, verification
programs will be created, etc., just as has happened in the semiconductor
industry. These things *will not* happen overnight.
> This means, almost by definition, an extremely uneven distribution of
> power and very rapid change. How do you plan to deal with such a
> scenario?
If Zyvex gave you a nanoassembler tomorrow, there would be little
power shift and little rapid change. There would be a huge discussion
about what to build, who would do the designs, how to distribute the wealth,
etc. The wealth doesn't stay localized, as soon as there is "proof
of concept" that directed nanoassembly is possible, every country
in the world will devote huge teams of engineers to reproducing the
results. If you had to "opt-out" of the patent treaties I think that
there is a good chance that countries would probably do that.
>
> What would you do if the Singularity came *tomorrow*? What would
> *any* of us do, except holding our breath?
>
If I'm working in the lab, and I happen to see a clear path to a
nanoassembler and realize that it is going to make my "company" rich
but not me, am I going to keep working for the company? Unlikely. I'm
going to go to the nearest investment banker and say -- Here is a path to
building a nanoassembler, would you kindly give me $100M to build it?
You would rapidly get a Balkanization of development efforts which
is probably a very good thing. You can only argue for a centralization
of the wealth created by the singularity, *if* you can argue that there
is only a single path to nanoassembly and a single individual/group/
goverment can keep it under their control. The only way I see
that happening is, *if* someone gets a nanoassembler *and* has the designs
for the machinery to eliminate everyone else on the planet in
such a way that it was impossible for them to mount a reasonable
defence. Knowing what I know about the weaknesses in nanotechnology
and the fact that the knowledge of those weaknesses is fairly well
distributed at this time, I'm very doubtful that that could occur.
> Yes, wealth is very important. However, just being "well off",
> "moderately rich" or even "rich" won't probably be good enough.
Why? I've already pointed out that everyone becomes "effectively"
rich ~5 years after robust nanotech becomes available. When
nanotechnology becomes clear on the horizon, we are going to
undergo a very interesting shift in wealth. What is the worth
of a skyscraper when a large number of nanobots can build the
equivalent in a very short period of time (even *if* you could convince
some silly person to leave his mansion and actually go to the office)?
What is the wealth of a company that may find its market eliminated by
open source designs? What is the wealth of government bonds when every
individual in the country can "leave"? The problem with "wealth" is
that you have to keep it someplace! If the traditional wealth
holdings undergo a substantial deflation in value, those who see it
coming and position themselves properly *will* be riding the wave.
> To get a prime spot on the wave, you'll have to out-paddle a whole lot
> of other skillful surfers, some of which will be more than willing to
> push you under in order to get to the wave first.
This is where the wave analogy fails. There is no "prime spot"
on "the" wave. Nanotechnology enables an expansion of dimensionality
so that there are now *many* waves. The human mind does not have
the ability to focus on them all. Since the barriers to inventing
new waves have been lowered, we will have trouble keeping track
of them all. It will be like the growth of WWW today only it will
include both the physical and intellectual realms.
> Hard cash will buy you the bulging, rock-hard muscles you'll need to
> stay ahead.
Hard cash? You mean those pieces of paper backed by the U.S. government?
Hmmmm... I'll take a nanoassembler over a stack of those pieces of paper
any day. Probably because the nanoassembler lets me make as much of
the hard cash as I want... :-) Oh, you meant gold? That stuff
you can molecularly sort out of the oceans? Oh, I know, you really
meant diamonds... :-)?
> As you'll be competing with large companies, governments, organized crime,
> dictators, terrorist groups and other folks with no shortage of funding,
> you better get rich indeed.
Large companies? You mean those entities that are left after open
source designs eliminate most of the known markets? Governments?
You mean those empty buildings left left when everyone has moved
to Oceania? Organized crime? Those people who want to sell me
something illegal I can build myself? Dictators, terrorist groups?
Oh, those people who want to control the behavior of people who
can easily relocate themselves where they don't have any influence?
I would have to say that most of these concepts rest on ideas that
will be outmoded by the singularity. These organizations will be
most threatening only if they adopt a luddite stance due to the
potential for the singularity to eliminate them.
Also, companies, governments, mobs and terrorist groups will not
easily be able to turn themselves into a unified SI. The problem of how
to effectively *merge* minds will be one of the most difficult to solve
and so may not develop until many years after nanotechnology develops.
All you can effectively do in the beginning is upload separate minds,
creating some kind of hive mind, and then allow them to pool their
intelligence in the same way we pool ours here, though with a much
higher intercommunication bandwidth. But if their group can do it, it
is likely that your group can do it to. As I believe was pointed
out in another thread, we really don't know how to evolve or
dramatically enhance "intelligence" (as compared with say memory).
If its a trip, stumble & fall, pick yourself up and try again process
(as seems likely), then the landscape to be explored and large
number of dimensions may make this a relatively slow process,
though much faster than what we are currently used to.
It is worth remembering that it still takes hundreds of years
to restructure the solar system into an optimal SI computing
architecture. It doesn't happen overnight.
>
> That's why essentially some sort of transhuman cooperation effort is
> needed, as the chances that any individual could get very rich, and
> keep track of things at the same time, are very small indeed.
Agreed. As "we" have devoted more thought to the problems, we
(collectively) may have more insight into how to chart a path
through the swamp that will minimize people drowning.
> It's a good start, but if we really want to stand a fair chance, I'm
> afraid we'd need a much bigger and much better coordinated effort. Being
> aware of something is one thing, but actually having access to the
> technology is something rather different...
Gee, ABC discusses uploading, CBS discusses CR, cryonics & immortality.
Are you suggesting that the Seventh-day-Born-again-Extropians,
should be banging on peoples doors and preaching in the streets
about SI conversion? There is a little matter of inertia to
deal with. If we allow the ideas to slowly crystalize in
the minds of the person-in-the-street, and they can see that the
benefits (even thought they upset the status quo) clearly outweigh
the risks, then we will have an army of anti-luddites.
>
> Oh yes, me too. Nothing to lose, might as well give it a shot...
>
To me, the singularity looks slower to develop than most people
think and therefore I have time to wax my surfboard. The parts
of the singularity I can see involves people, and since I understand
people at least to some degree, I'm not afraid of it.
I am *much* more concerned about what I may not see -- that early on
someone creates a self-evolving amoral AI and that that single entity
somehow gains access to nanotech and turns itself into an amoral SI.
In that case it may well be bye-bye-birdie time. I think this is
a strong argument for widely distributed nanotech/AI research since
in that case the multitude of paths make it likely that the concepts
and technologies would exist to mount a defense against rogue-Is.
In the end, its not whether you win or lose, its whether you enjoyed
the ride.
Robert
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:05:04 MST