Re: [SL4] washington post article

From: Eliezer S. Yudkowsky (
Date: Sun Apr 02 2000 - 14:30:15 MDT

Greg A wrote:
> I have an answer. How about:
> Use MPC technology, such as that being developed by Fact Technologies, to
> extend humanity's control of the universe through successive legions of
> Vinge-style physical, digital, and social automation,

Now, there goes someone who really loves relational databases. If I
ever need a relational database built, this is probably the person to ask.

I don't know about a Singularity or an AI, though.

SQL doesn't exactly capture the full complexity of physics, life, mind,
and culture, except in the Physicist's Paradigm sense that SQL is Turing-complete.

Of course, this isn't really what you're talking about, according to
FactTechnologies's video whitepaper. You're talking about using
real-time enterprise computing systems (which are usually built on
relational databases, albeit not real-time ones) to add another layer of
organization to the company; that is, by automating the flows of
information that result in certain types of responses - even
thermostat-like feedback - you hope that certain types of higher-level
phenomena will emerge, just as a capitalist economy emerges from
selfishness, Gaia effects from DaisyWorld, brains from selfish-neuron
wiring algorithms, and so on. Personally, I'm not sure how much
higher-level phenomena you're going to get, but it does seem worth a
shot. An organization with real-time feedback mechanisms should at
least be more efficient than one without them, maybe substantially more
efficient. I've had my own ideas along those lines. But I don't think
it's enough to save the world.

Even with real-time feedback mechanisms and emergent phenomena, a
corporation composed of human elements plus automation does not have the
complexity of a single human brain. If you have an automated
corporation with hundreds of non-genius humans, they will not be able to
outthink a genius human. Thoughts are enormous structures that exist in
a brain with a hundred billion neurons and a hundred trillion synapses;
the bandwidth between humans, in a corporation, isn't enough to expand
the class of things that are Obvious, and while it does expand the class
that's Inventable, it does so in a very limited, airy way.

Maybe two or three or sixty-four nongeniuses with BCI telepathy links
(BCI: Brain-Computer Interface) could outthink a genius, if the
interconnection bandwidth was high enough. (Certainly, both
corporations and telepaths can do things that a genius can't; the
question I'm asking is whether they can do genius-things - invent
Newtonian physics, predict General and Special Relativity, design an AI...)

Groups of humans can spread out a thought-structure over multiple
individuals, enabling the construction of mental structures far larger
than any single human is capable of. But the elements of that structure
have to be joined by links limited to the bandwidth of human language
(admittedly with a shared reference base). Human genius is
characterized by very dense, tightly-interconnected chains of reasoning.
 Those genius thoughts might be Understandable by a corporation - they
are, for that matter, understandable by individuals; GEB, Q.E.D. - but
the thought processes that Invent them cannot be spread across corporations.

A group of humans is simply capable of holding a larger overall
structure. Each individual link, each individual insight, is the
product of a single individual, because the cognitive abilities that
produce those insights are neural modules and neural processes with
internal variables that just don't get shared by language.

Imagine a thousand Gary Kasparovs trying to beat Deep Thought. If they
were all in the same room thinking about it out loud, that's analogous
to a modern corporation. If a computer system projects a couple of
moves ahead and assigns various positions to each separate Kasparov,
then that's a FactSystem corporation - or at least, that's what they're
aiming for. Even so, though, the Kasparovs won't be able to share
insights, chess-perceptions, about each of the hundred boards assigned.
They'll need to have the same insights over and over. They won't be
able to take an insight on one board and apply it to a second, or invent
forks that they'd have to look at two boards at once to see. And when
it comes time to chose the best move, each one will have to give some
kind of numeric rating to the chessboard in the absence of any knowledge
about what the other possibilities are like. They won't be able to
match a BCI-telepathic team, and a BCI group can't match a
general-intelligent AI with a chess cortex.

This is essentially the same objection I have to Marc Stiegler's
characterization of the Earthweb as a superintelligence, or Robin
Hanson's characterization of modern corporations as being in the same
league as hardware intelligence enhancement. (Greg A, if you
haven't read a book called _Earthweb_ by Marc Stiegler, I think you'd
really enjoy it.)

If humans are Legos, then corporations represent the class of structures
that can be built with Legos; automated corporations are the class of
structures that can be built with motorized Legos; hardware intelligence
enhancement is the class of structures that can be built with modern
manufacturing; and AI is nanotechnology.

> while attempting to
> restrain that growth to such an extent that it doesn't dilute our essential
> humanity?

Just thinking about the Singularity for a few years has severely diluted
my "essential humanity", or at least what most people would cite as
"essential humanity" - my fear of nonexistence, my emotional attachment
to various parts of my personality and cognitive architecture, my
allegiance to the human race (though I'm still fond of it), that sort of thing.

Can't we just admit that we don't even know which parts of ourselves are
valuable, and keep an open mind about which parts we might want to throw
away? Right now, it's all theory. Once we're faced with the reality,
our perspectives will change.

I certainly don't have the moral (or engineering) authority to chide
anyone about remaining attached to their humanity, but the correct goal
from that perspective is to ensure that each individual has the option
of remaining human. Not to ensure that everyone is similarly restrained.

--      Eliezer S. Yudkowsky

                 Member, Extropy Institute
           Senior Associate, Foresight Institute
GET A NEXTCARD VISA, in 30 seconds!  Get rates as low as 2.9%
Intro or 9.9% Fixed APR and no hidden fees.  Apply NOW!

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:07 MDT