Re: Hyper-AI's vs Transhumans
Dan Clemmensen (Dan@Clemmensen.ShireNet.com)
Wed, 06 May 1998 20:11:21 -0400
Paul Hughes wrote:
>
> Dan Clemmensen wrote:
>
> > IMO an SI will very likely start as an augmented human,
> > a collaboration between one human and a computer. This SI can
> > rapidly augment the computer part of itself, so there is no
> > reason to assume that a second SI will come into existence prior
> > to the first SI taking over the world, or that a de novo AI will
> > ever come into existence. The SI may choose to decouple from its
> > human component, but such a decision, and indeed any decision made
> > by the SI, is beyond analysis by humans.
> > Please see: http://www.shirenet.com/~dgc/singularity/singularity.htm
>
> I've read over your essay a couple of times now. Compelling.
>
> Are you suggesting that the singularity will essentially consist of only
> one intelligence (i.e. one entity, one ego) with, if we are lucky, an
> arbitrary number of uploaded humans as cells in its brain? A borganism?
>
This is my opinion. However, I realize that it is only one of a wide range
of possibilities. There are a whole bunch of interesting things for use to
consider, and in retrospect I feel that dispite my own admonishments to the
contrary I have focused in on a particular point within the space of possible
singularity initiators: I've picked the "single human" spot on the
"initiator size" axis, and the "instantaneous" point on the "rate" axis.
I've also picked the "single entity" on the "post-human population size"
axis. All of these choices are in my little paper, but none of them are
sufficiently explicit and I spend little time on explaining the alternatives.
I rejected the "zero humans" point on the initiator axis because think AI
(i.e., no human component of the SI) is harder to achieve than a collaboration.
I rejected the larger numbers of humans because I guess I thought a multi-
human collaborative component was harder than a single-human component, but
I'd now guess that this is a weaker argument. Now, I would still assign
a near-zero probablity to the pure AI and a large probability to an initiator
size of one human, but I think a small group (say a workgroup at a lab) is
possible, and a larger colloboration {say an internet mailing list :-) }
is also possible.
I picked the "instant" point on the rate axis for the reasons stated in the
paper. Of course, I started with Vernor Vinge's "singularity" as part
of the title of my paper, so I guess I'm biased. This choice is easy to
attack on all kinds of perfectly reasonable grounds, mostly because the
term "singluarity" isn't the correct mathematical/physical analogy. It's
better to use a phase change as the analogy: The liquid-to-solid transition
of a supercooled liguid when a seed crystal is added. The analogy is, I think,
clear: the raw available computing power on the internet represents the
supercooled liguid and some as-yet-undeveloped software represents the
crystal.
It might be nice if some of the participants from the prior round of the
">Web" discussions would add any new thought they may have had on this matter.
As to uploads: the SI may or may not choose to permit or compel uploads, and
may or may not choose to permit the continued existence of humans in any form.