Ethical Injunctions

From: H C (
Date: Thu Dec 22 2005 - 18:55:50 MST

As I understand it, ethical injunctions are not implicit within causal
validity semantics, as described in CFAI (and other places). If so, is this
additional functionality necessary for Friendliness, or is it just a
security precaution? Is it plausible that an ethical injunction can
unnecessarily cause unforeseen harm (either in the implicit function of the
system or in some explicit effect in reality)? How can we be sure the
ethical injunctions we create directly cohere to true and meaningful content
within the general intelligent function, rather than accidentally adding a
bias to the process that is completely alien to any human bias?


This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:54 MDT