Re: Educating an AI.

From: Stephen Reed (
Date: Tue Jul 09 2002 - 14:43:52 MDT

> I consider much of our current IP law to be grossly Un-Friendly
> and a serious handicap to [even] human advancement. It is even
> technically illegal for me to digitize my own library and make
> it available wherever I have a net connection. Most parts of it
> I can only [legally] access by carrying around dumb slabs of
> dead trees! It is certainly illegal for me to share it [in
> electronic form] although I can loan out individual, original
> hard-copy books.
> On the other hand, having worked inside many corporate firewalls
> in my life, I question that very much of great value to our AI
> is inside many of them. A standing joke of mine in many
> software companies I have been in is that the best way to bog
> down our competition hopelessly would be to leak our proprietary
> source code to them. So much is locked up and considered
> precious which would be dismissed as trivial and ingrown if it
> was aired publicly.


A thought exercise carried out from time to time among my co-workers at
Cycorp is to imagine an AI that can read. Consider how and if the
vast IP tied up in text can be protected from a non-human yet
intelligent reader. Discard for a moment the difficulties of scanning
and OCR - assume them solved.

1. Does copyright keep an AI from perceiving, understanding and making
use of the knowledge content of a book? We think not but have not reached the point
where we need a lawyer or court test.

2. Would a knowledge economy replace or supplement the bookselling
business? The reduction in friction would address the point of your post.


Stephen L. Reed                  phone:  512.342.4036
Cycorp, Suite 100                  fax:  512.342.4040
3721 Executive Center Drive      email:
Austin, TX 78731                   web:
         download OpenCyc at

This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:40 MDT