From: Michael Wilson (email@example.com)
Date: Mon Jul 18 2005 - 07:06:10 MDT
Ben Goertzel wrote:
>> While the situation is perhaps rather better in the overall
>> space of AGI projects, it is unfortunate that while some have
>> realised the importance of working implementations, virtually
>> no-one has been working on the problem of objective criteria
>> for evaluating AGI theories.
> Hmmm... Do you have any specific thoughts in this regard, Michael?
I have spent a good portion of my time on this area, but it would be
impractical to discuss my results here for all the usual reasons.
However it does surprise me that no-one has put serious effort into
the more straightforward task of constructing an 'IQ test for AIs'.
Over the last couple of years I have built up a collection of test
problems culled from assorted literature, but I have yet to encounter
any centralised effort to identify all the key competencies a general
intelligence should exhibit, and then supply a set of tests in
ascending difficultly for each competency. Ideally this would be
realised as software and easily interfacable with any AGI project.
There are some 'AI world' simulation frameworks available (and some
people on this list are working on their own private versions of the
same), so a simple subset could probably be constructed just as a set
of world files and goal descriptions for such a world.
I regret that this reads as a 'wouldn't it be nice if someone...' post
as I do not currently have time to work on such a project myself. I
will try and make my own test problem library publicly available as
and when it becomes practical to do so, and I would be in favour of the
SIAI working on such an 'AGI benchmark' if the organisation had a
sufficient excess of funds and staff.
* Michael Wilson
How much free photo storage do you get? Store your holiday
snaps for FREE with Yahoo! Photos http://uk.photos.yahoo.com
This archive was generated by hypermail 2.1.5 : Wed Jun 19 2013 - 04:01:09 MDT