From: Charlie Stross (charlie@antipope.org)
Date: Fri Dec 28 2001 - 06:19:42 MST
On Thu, Dec 27, 2001 at 06:40:48PM -0800, Spike Jones wrote:
>
> For instance, hard takeoff fans: perhaps there an unknown factor
> where friendly AI begins to develop its own next generation,
> but a sub-AI, not necessarily "unfriendly" but just misguided,
> attempts to halt or reverse the growth of AI. Could there be
> a machine level equivalent of luddites and reactionary religious
> notions? Is such a thing even imaginable? I can imagine it.
There's another type of entity we ought to consider in the context
of a singularity: the limited liability company.
A company is a legal entity that can own property and obey
instructions. It can also issue instructions to other companies that it
has shares in, in accordance with a set of pre-wired programming (its
charter). I know of companies that are wholly owned by other companies in
a cyclic graph structure; a friend of mine runs an off-the-shelf company
factory and his robot companies act as secretary and treasurer for the
new nestlings until it's time to sell them to their new owners.
Currently this is all paper-and-pen work, but is there any legal
reason why you couldn't, for example, embed a formal grammar in a
company constitution along with an injunction to execute instructions
received from authorised directors in the form of source code? If/when
the company administration mechanisms move to network-mediated exchanges
(probably using EDIFACT documents embedded in XML to exchange paperwork)
this all becomes somewhat plausible ...
Adds a whole new meaning to the term "shell company", doesn't it?
I think we're actually seeing a rather odd expansion in the scope of,
and use of, limited companies. I mean, until 1947 here in the UK it
took an Act of Parliament to establish a company. Today, it costs about
twenty pounds and requires filling out a couple of forms. A startling
proportion of the people I know own companies. The "let's buy a shell
company, just in case it comes in handy" meme seems to have been spreading
on the back of the "let's register a domain" meme, and lots of people who'd
never have looked at the concept before are jumping on board.
(Wouldn't it make sense -- from a rights point of view, at least until the
law catches up with reality -- to implement any conscious AI as software
running on and owned by a limited-liability company, just to protect it
from the risk of some irresponsible Joy-like person switching the power
off and saying "it was just a program, what's all the fuss about"?)
And if the first human-equivalent AIs show up already equipped with the
same legal rights as a company, what does this imply?
(Personally, I'm not sure I want to be sharing the planet with Redmond
when Microsoft, Inc. wakes up and starts taking a hard look around.)
-- Charlie
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 08:12:53 MST