From xdrudis@tinet.org  Wed Dec 13 07:57:40 2000
Date: Wed, 13 Dec 2000 08:57:40 +0100
From: Xavier Drudis Ferran xdrudis@tinet.org
Subject: [Patents] Re: computer doing the inventing

Luuk van Dijk va escriure el Tue, 12 Dec 2000 18:25:16 +0100  :
>Xavier Drudis Ferran wrote:
>> If you can invent it just by having a program generate
>> random circuits and SPICE analyze them to see if they work, then that
>> means the patent owner should be the random circuit generator, not
>> you, and since computer software is not considered people in Europe,
>> your random circuit generator couldn't be the patent owner,
>> the patent wouldn't be possible.
>> 
>
>Actually this is exactly (well, nearly) what has been done: 
>	http://www.genetic-programming.com/johnkoza.html
>
>
If I was in a comic I would jump backward and fall on my head. 
Is like saying to someone "when pink elephants fly..." and 
he answering "well, just the other day I had to duck because one 
came flying too low...". I'll have a look, thanks for the URL.

>Koza is one of the leading figures in the field of evolutionary computing,
>and uses the fact that he can automatically generate electronic circuits
>and chemical cycles that *have been patented* to illustrate that Genetic
>Programming solves the ambition of artificial intelligence: having computers
>solve problems without telling them step by step how to do it.
>
Well, I find it interesting in what he does it for, research on AI. 
Although with the current examination process, I think he hasn't proved that 
he achieves artficial intelligence (unless he has other algorithms for 
generating lawspeak for the patent applications). I feel the intelligent
part of getting a patent nowadays is not inventing something but writting the 
claims and such so that it seems new and complicated.

But I still think the patents on a program output, however intelligent the 
program, should not be patentable. It is bad enough having people taking away
freedom from other people to start thinking of artificial intelligent agents 
taking away freedoms from people.

>They actually used SPICE to simulate the circuits.
>Guess what. Koza holds some patents on the method of Genetic Programming.
>
Well, I don't like those patents then because they are software patents, 
but if software patents were admissible, then at least the patent holder 
and inventor would be a human being.

Pray, where were all those patents granted?. In USA, in Europe?. 

>This is not some remote weird fruitcake wannabe-science, It is a blossoming field with 2-3 large annual conferences, and very many publications over the past decade.
>(eg. http://ls11-www.informatik.uni-dortmund.de/people/banzhaf/cfp.html)
>
>
>Thought this migt be relevant.
>
It is interesting, thanks for telling us.


--
Xavier Drudis Ferran
xdrudis@tinet.org







