[SLED] Avatars as historical figures

Kenneth Schweller SCHWELLER at bvu.edu
Fri Mar 9 08:02:03 PST 2007

I agree with Kim that no existing AIs can yet 'accomodate genuine open conversation' and that we are no where near Turing's 1950 prediction that ""in about 50 years' time it will be possible to program computers...to make them play the imitation game so well that an average interrogator will not have more than 70 per cent chance of making the right identification after five minutes of questioning."  Nevertheless I believe that chat bots can make a significant, if limited, contribution to the atmospherics of a virtual world. 
     I am currently teaching AI classes on SL and one of our first assignments was for each student to make a chatbot with a distinct personality.  Then, in a  'chatbot fest' we paired the bots  with one another to see how they would carry on.  The results were quite amusing as you can imagine.  I have been hesitant to release the script for these bots because of the potential for spam, abuse and annoyance but I am more than willing to contribute the code for use in the virtual historical recreation projects I have been hearing about.   The bots connect to an external 'pandora bot' server and are easily programmed off-world through the pandora bot web site.  To protect against infinite looping and spam the bots only 'whisper' and after 100 responses they 'blow a fuse' and have to be manually reset.  They also have a limited knowledge of their SL context and who they are talking to.
      If anyone is interested just let me know.
 Cheers, Ken Schweller ( Ken Caliber, SL)
Dr. Ken Schweller
Professor of Computer Science and Psychology
Buena Vista University
Storm Lake, IA 50588
schweller at bvu.edu 
http://schweller.bvu.edu ( http://schweller.bvu.edu/ )
712 749-1819

>>> "Kim FLINTOFF" <kimbo2 at iinet.net.au> 3/9/2007 2:14 AM >>>

Hi Martin,
Welcome aboard the good ship SLED.
I?ve yet to see AI that is sufficiently developed that it can accommodate genuine open conversation.  Engaging students in role and then using an elaborate chatbot is probably only suitable if you are going down a very linear quest driven learning path. As we?ve all encountered in RPGs ? the NPCs are nearly always quite limited in their responses precisely to keep the play focussed on the predetermined narrative? in my field, drama education, such linear sequential approaches are considered counterproductive? they inhibit rich engagement, play and exploration?  we want to foster ambiguity, uncertainty, unpredictability.. we want student interest to drive an inquiry ? we also want creative, artistic responses that might be more aesthetically pleasing than factually accurate?.  I?ve yet to find an AI model that works as well as a live facilitator?
And I wonder if AI is capable of dealing with ?metaxis? (the holding of two or more realities in concurrent experience/awareness) ? students engaged in role work best when they can shift between positions of engagement in role and a more detached reflexive position?  the cueing of shifts between realities will present some real challenges to any intelligent agent..  
Having said all that, I think the use of intelligent agents is going to be a very exciting development in MUVEs.
Keep us posted on your progress?

Kim FlintoffB.A., Grad. Dip. Ed., M.Ed.
PhD Candidate
QUT Creative Industries: Performance Studies
Sessional Lecturer/Tutor
ECU School of Education and Arts: Contemporary Performance / Drama Education
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.secondlife.com/pipermail/educators/attachments/20070309/a9365f08/attachment.htm

More information about the Educators mailing list