28 June, 2022
Manuel Eberhardinger

Schloss Dagstuhl Seminar: Human-Game AI Interaction

Image by David Monniaux CC BY-SA 1.0

The Human-Game AI Interaction Seminar took place from 19th-24th of June 2022, at Schloss Dagstuhl, a Leibniz Center for Informatics with the goal of advancing research in computer science. Schloss Dagstuhl was founded in 1990 and is funded by the federal and state governments in Germany. At Dagstuhl, two seminars are held in parallel each week to advance research on a particular topic, such as Security of Machine Learning or Cognitive Robotics. Our PhD Student Manuel Eberhardinger was invited to participate in the seminar on Human-Game AI Interaction.

At Dagstuhl, researchers from industry and academia meet on this particular topic and try to find different areas to work on to find new ways to solve problems or advance research in this area. The following picture shows a list of all the proposed topics, such as GAN based procedural content generation or explainable AI for games. Every morning, all participants choose a topic that they work on in a working group throughout the day. At the end of each day, the working groups present their work and ideas to the other participants in a presentation.

The evenings were used to socialize or hear presentations on other research projects or accepted papers. One presentation introduced the Ludeme Project, a study of the world’s traditional strategy games throughout recorded human history, beginning with 8000 BC. The other talk was about Off-Belief Learning which proposed a method for better cooperation between humans and computers.

The last day was used to wrap up the entire week and suggestions for planning the next Dagstuhl seminar on games were also accepted.

In the end, this seminar was a fruitful week with a lot new contacts and also many new ideas which could be beneficial for my PhD process. I would like to thank everyone who made it possible for me to participate in this seminar and would recommend everyone to attend a Dagstuhl seminar if possible.