Re: No AI for Nano/No Nano for copyloads

From: Robert J. Bradbury (
Date: Thu Jul 13 2000 - 10:51:47 MDT

On Thu, 13 Jul 2000, Robin Hanson wrote:

> I think it likely that we will have the ability to create a single
> expensive "upload" (for less than $1B) well before advanced nanotech,
> and given the vast amounts of money to be made from one this is what
> will happen.

Hmmmm, I see several questions/problems:
  (1) who "owns" the hardware the upload runs on?;
  (2) does the upload have the "classically" attributed human
      rights or is it a simple slave mind?

Unless you can take the consciousness out of the machine there
are going to be a lot of people arguing you can't use it as
a slave. If it is a free agent, then a lot of legal groundwork
will need to be laid for it to act as an economic agent (have
a S.S. number, pay taxes, etc.). Then if it doesn't own its
hardware it could end up paying hefty lease fees to IBM or whomever
does own it. Looks like a swamp to me.

> Then its a matter of filling market niches and business
> competition that I think leads to domination of the upload economy by
> the first few thousand, and perhaps the first few ten, uploads.

If this strategy is followed before nanotech, then you still have
the problems that the hardware is going to fill a room and will
produce a very hefty annual electricity bill. You are likely to
have 10-15 years before the hardware gets cheap enough that the cost
of running an upload falls below the annual food budget of a real
person. It may be an interesting research tool but I doubt it
will quickly break out in a way that changes the general economy.

> By the time your "evoloads" show up, the game is basically over.

Depends. There is likely to be a lot of human prejudice against
uploads. I could see trade wars and sactions developing between
the nations that allow or use them and those that don't.

Even if that doesn't happen a general awareness of the economic
threat uploads represent could drive a Manhatten project for
evoloads. If AI keeps getting more and more sophisticated
(becoming non-AI in the process), then the economic advantages
of the uploads may have been trumped by mass-produced hardware/
software agents.

If you assume the upload is a full copy of a human you have
an interesting problem initially of the upload having capacities
and capabilities for an economy that doesn't exist yet. It also
has the standard pattern for human needs and desires. It might
get paid megabucks to write some fancy software and then spend
it all for a loopback interface that makes it feel like it is
getting a nice massage.


This archive was generated by hypermail 2b29 : Mon Oct 02 2000 - 17:34:33 MDT