From: Michael Lorrey (retroman@together.net)
Date: Thu Oct 01 1998 - 16:03:28 MDT
Robin Hanson wrote:
> Joe Jenkins writes:
> >> Economic Growth Assuming Machine Intelligence
> >> http://hanson.berkeley.edu/aigrow.pdf , .ps
> >Why have you assumed a state of slavery for the sentient beings? How
> >would self ownership and a right to ones own wages, for the machine
> >intelligences change your results?
>
> I did not assume machine slavery, and so my results are uneffected by
> slavery vs. not.
Uh, I don't think so. If all AI machines were treated with a slave
existence, then biological humanity as a whole would develop an
aristocratic level of wealth based on the uncompensated profits from the
productivity of the AI machines. I think that this scenario is the
preferable one at least at first, as we are not JUST talking about
completely human level AI. There will be a whole taxonomy of various AI
entities, some of which we may want to grant personal sovereignty to, but
hardly all of them. Its rather dumb to grant it across the board when 90%+
of AI entities will have intelligence less than a monkey. The dumber AIs
will undoubtedly be developed first, and should be kept in a slave/master
relationship, until actual human level AI is developed. Of course, if the
'upload' scenario develops before outright evolved AI, then it follows that
where a 40 IQ AI entity would be a factory slave, an uploaded human mind
would automatically carry over his/her personal sovereignty, as the
upload's IQ would baseline at 100-200, and would grow upward from there.
This does have a legal precedent. People who test below a certain level on
intelligence/cognitive tests are considered wards of the state, i.e.
permanent minors. There is also a grey area, where a retarded individual
who can demonstrate learned skills of autonomy can gain limited or full
personal sovereignty. It is also seen that retarded individuals may gain
greater levels of autonomy by intelligence amplification from
augmentation.
Mike Lorrey
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:49:37 MST