Re: Patented Ethical AI

From: Rafal Smigrodzki (rafal@smigrodzki.org)
Date: Mon Aug 23 2004 - 11:11:32 MDT


Martin Striz wrote:

>Don't you sometimes just hate intellectual property laws? Some of my work
>involves bioinformatics/genomics, and we've had a big problem with private
>companies patenting genome sequences. I don't know if that will be an issue
>now that private databases of the human genome exist, but I couldn't imagine
>having to pay my grant money (your tax money) to Craig Venter for the
>discoveries that I make.
>
### If you really could dispense with the data under patent and still
make your discoveries, there would be no case for rewarding the
providers of these data with a part of your grant money. On the other
hand, should you come to the conclusion that making discoveries without
the patented data is impossible, it is in your best interest to reward
the providers of such data. Or else, nobody will bother sequencing
genomes for you to play around with.

Indeed it is a bit arrogant to claim one "makes discoveries", seemingly
without a debt to others (such as the shady entities who sequence
genomes). We all stand on the shoulders of giants, or multitudes of lab
worker drones, and our discoveries are merely elaborations on their
epiphanies.

But, back to the real issue: It is interesting to investigate the
patentability of RPOP. Since as envisioned by Eliezer, the RPOP is not a
person (and to help us keep that in mind, references to emotionally
laden prosaic terms such as "friendliness" are fastidiously avoided), so
the question of self-ownership does not arise. There is no prior art,
there is demonstrable utility, there is a fair amount of
non-obviousness, so the basic criteria of patentability are met. Yet,
the question of enforceability arises, in at least two aspects:

Firstly, the RPOP may decide that being covered by a patent would clash
with its goal system, and therefore refuse to be bound by the law, even
without recourse to personhood and self-ownership. In that case, given
the likelihood that the RPOP would be able to defeat or deflect the
enforcers, the patent would be moot.

Secondly, even if the RPOP accepted its own patentability (and the
attendant rights of exclusion, transfer, termination, etc., vested in
the IP owner) the owner might be unable to exercise these rights fully,
due to the specific properties of the patented property. For example,
the right of transfer might be curtailed, since the RPOP isn't likely to
accept binding instructions from any single human, and a transfer of
rights (even if not actively opposed by the RPOP) would not change the
situation anyway.

On the other hand, certain patent rights might be easier to enforce: the
RPOP might accept a copyright on self and even help with the enforcement.

Future will tell.

Rafal



This archive was generated by hypermail 2.1.5 : Tue Feb 21 2006 - 04:22:43 MST