Sunday, May 13, 2012

Dagstuhl seminar on Artificial and Computatio​nal Intelligen​ce and Games

I spent last week at Schloss Dagstuhl together with some 40 other AI researchers, talking about AI and games. It was awesome. This is the doodle i made the last day, I think that can serve as an evaluation:

The seminar was started by a series of short talks on different topics that could be interesting to discuss at the seminar. Here is the agenda for talks:

Agenda talks monday

Then, we divided ourselves into groups. Michael Young, Ruth Aylett, Richard Evans, Paolo Burelli and myself formed a group to talk about computational narrative. We worked on this until Wednesday morning when the full group got together. Each group presented their work to the others.
Computational Narrative group presentation

Then it was time to form new groups. Michael Mateas, Richard Evans, Ruth Aylett, Mike Preuss, Ana Paiva, Elisabeth Andre and myself got together to talk about Social Simulations in games. We scored a room with a white board.

We had plenary sessions too. One was about grants, another about industry cooperation. We all assumed that it was in a sorry state, but then we realized that 20 out of 34 were cooperating with industry in projects, so it didn't seem so bad after all. The third plenary resulted in a flurry of activity to create aigameresearch.com and .org. We want to communicate research results by showing them in games that people can download. We also thought about having competitions (agents etc).

In the evenings, we played board games. There was also a quiz-night. There was a challenge on formulating the questions. We were seven groups, and each group formulated six questions. The challenge in formulating the questions was to aim for that 50% of the other groups would be able to answer the question.

The group-work will be written up, so that each groups discussion is summarised in a paper. The proceedings will probably be available in a few months. (Our deadline for the papers is 30 days from now.)

It was a truly inspiring week. I found myself making connections and associations in my thoughts that will most likely guide my future work. It is as if getting together like this suddenly makes one function better, as if the smartness of the group somehow enhances the individual. The organisation team, Simon Lucas, Michael Mateas, Mike Preuss, Pieter Spronck and Julian Togelius, did an amazing job, aided by the smooth machinery of Shloss Dagsthul. Thank you :)

More info:
The seminar info-page: http://www.dagstuhl.de/12191
Alex Champanard of AI Game Dev wrote a post on Monte Carlo tree searches one of the evenings in the game room, here it is: http://aigamedev.com/open/coverage/mcts-research/

Pictures:
The pictures I took at the Seminar: http://www.flickr.com/photos/reality/sets/72157629706011040/
Pictures from Dana Nau: https://plus.google.com/u/0/?tab=mX#photos/102912579368456322089/albums/5740552644445171025



The hive mind

No comments: