COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	haugel[s82,jmc]		Notes on Haugeland's "Artificial Intelligence - the very idea"
C00004 ENDMK
C⊗;
haugel[s82,jmc]		Notes on Haugeland's "Artificial Intelligence - the very idea"

draft received from Stanton

1. If intelligence is characterized in a problem oriented way,
then our idea of it needn't be anthropomorphic.  Thus we could
imagine it solving mathematical problems, which I would claim would
be much the same for any intelligence.

2. Turing didn't propose the test as a definition - but as a sufficient
condition, so that he could decline to follow further the arguments
of philosophers who would call a machine unintelligent no matter
what capabilities it had.  The sufficient condition needs to be
tied down, because it has turned out that many people are easily
fooled, so the person must be an expert - perhaps with advice.

Actually, because intelligence can be analyzed into intellectual
mechanisms, the Turing test needn`t be actually used.  Instead we
ask what mechanisms a specific program embodies.

3. Creativity can be easy according to my technical definition.
My program will do nothing I couldn't do if I lived a million years.

4.