From: Arnt Richard Johansen (arntrich@gmail.com)
Date: Fri Jan 13 2006 - 15:30:57 MST
First of all, it seems to be very likely that artificial minds above a
certain level of complexity are going to have qualia. The thought
experiment of replacing one neuron at a time ensures that it feels just as
real to be a simulated brain as a biological brain. When it comes to
artificial minds that are built on different systems than that of a
biological neural network, the image is less certain, but it seems
reasonable to suppose that qualia arises from simple properties that are
common to most system that we would describe as "thinking" instead of
merely "calculating", rather than a more complex set of properties that is
unique to a neural network. At least the possibility of artificial minds
with qualia cannot be ruled out!
If artificial minds have qualia, it should be obvious that we cannot treat
them the way we would any other piece of machinery. At the very least,
they should not be subjected to torture. (This is, unless I am not
mistaken, similar to the level of protection lab animals have in most
countries.) Also, at least in the case of human-level AI, they should not
be destroyed (ie. killed).
Scientists have to file an application with an ethics committee if they
want to use animals in their research. If they simulate brains, they can
do whatever they want. This is happening today, although at the time being
the simulations are either not real-time, or only of a small portion of
the brain. See [1], [2].
I think it is very important that artificial minds be given these rights
as soon as possible. Preferably *before* the appearance of the first
real-time human-level brain simulation, or any other human-level AI.
How can we protect the earliest simulated minds against termination or
torture? Legislation is an option, but it is hard to gain consensus to
award rights to a class of entities that does not yet exist, and may not
ever come to exist. Also, we have the major challenge on how to get SL<1
legislators to understand that numbers inside a computer can have feelings
too.
[1] http://neurodudes.com/2005/12/05/a-simulation-the-size-of-the-entire-human-brain
[2] http://vesicle.nsi.edu/users/izhikevich/interest/index.htm
-- Arnt Richard Johansen Mail sent to this address will not be read, because it is archived on the web in clear-text, and is thus likely to be picked up by spambots. My domain is nvg.org, and my username is three letters long and spelt alfa romeo juliet.
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:55 MDT