From: Wei Dai (weidai@weidai.com)
Date: Mon Jun 28 2004 - 19:05:16 MDT
Could those who advocate a qualia-based morality please propose an
algorithm, which given two sets of qualia, will output which set is
better?
For example, consider dream qualia, which I'll define as qualia with these
properties:
1. occurs when long-term memory formation is suppressed, and
2. reality discrimination is suppressed, and
3. is of much lower fidelity than normal qualia, but still passes the
relaxed reality-discrimination tests so that you don't notice
anything is wrong (and if you occasionally do notice, you quickly forget)
Today of course we mostly have dream qualia in our dreams. But an AI that
is trying to maximize positive qualia would have to consider whether to
accomplish that using dream qualia instead of normal qualia, for example
by putting everyone into low-fidelity VR and modifying their brains to
suppress memory formation and reality discrimination. The advantage of
dream qualia is the much lower computational requirements, both memory and
processing power, so that the system could accomodate many more people
than with normal qualia. How should it evaluate whether these advantages
outweigh the disadvantages (if any)?
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:00:47 MDT