The online interactive magazine of the Association for the Advancement of Artificial Intelligence

By Shi Johnson-Bey, Michael Buro, Derek Martin, Joshua McCoy, Ben Samuel 

The Workshop Program of the Association of the Advancement of Artificial Intelligence’s 18th Conference on Artificial Intelligence and Interactive Digital Entertainment (AIIDE-22) was held in Cal Poly Pomona, California from October 24-28. There were four workshops in the program: Experimental AI in Games, Artificial Intelligence for Strategy Games, Esports Analytics, and Intelligent Narrative Technologies.  

EXPERIMENTAL AI IN GAMES (W1) 

The 9th AIIDE conference Experimental AI in Games Workshop (EXAG) is interested in innovative applications of AI that present interesting ideas for developers, critics, players, and designers in all aspects of games and game development. These applications may include new systems made possible by AI, traditional AI techniques applied in new ways, and AI in support of mixed-initiative co-creative play experiences. 

This year was the first in-person EXAG since the start of the COVID-19 pandemic. We did our best to support a hybrid event to accommodate international presenters. We had excellent attendance on both days, with five paper sessions and two demo sessions (one formal, one informal). Our presentations spanned the following themes: story world generation, level generation, pixel art generation, adaptive MCTS, open-endedness in games, level reachability testing, NPC behaviors, AI-driven sonification, unit generation for real-time strategy games, empathetic AI, expressive range visualization, emulator frameworks for MCTS, and reinforcement Learning for fighting game AI. This year, EXAG received 22 submissions (21 for the research track and 1 for the demo track). We accepted 15 from the research track and one from the demo track.  

We extend a shoutout to our formal demo submission on Generating Emergent NPC Behaviours With Symbolic Reasoning from Sylvain Lapeyrade, Ph.D. student in Game Artificial Intelligence at Université Clermont Auvergne (UCA) at the LIMOS research unit in Clermont-Ferrand, France. We also extend a thank you to all the brave folks who impromptu presented their current work.  

The 2022 Experiment AI in Games Workshop was co-chaired by Abdelrahman Madkour, Lucas N. Ferreira, and Shi Johnson-Bey.  

ARTIFICIAL INTELLIGENCE FOR STRATEGY GAMES (W2) AND ESPORTS ANALYTICS (W4) 

The tenth workshop on Artificial Intelligence for Strategy Games (SG) was held on October 25, 2022 as a whole-day event at the AIIDE 2022 conference. This year we joined forces with the Esports Analytics Workshop (EA). This report briefly describes the workshop format and its contributions, and provides a link to all workshop resources. 

The continuing goal of the SG workshop is to bring together AI researchers and game AI programmers from industry, who are interested in strategic game AI, to present and exchange ideas on the subject, and to discuss how academia and game companies can work together to improve the state-of-the-art in AI for games. 

The goal of the EA part of the workshop was to begin to address the gap between esports’ popularity and esports’ research scarcity, and to help raise awareness and begin to build a community. 

This year’s workshop attracted 26 attendees across both SG and EA. During the one-day event, three peer-reviewed SG papers were presented, one invited industry talk was given on how esports analytics tools can help video game players, six research projects were briefly described in a series of “show-and-tell” sessions, results of this year’s AIIDE StarCraft AI competition were presented, and future research directions were discussed. We split the research contributions into three themes: 

  • Our first theme this year was research on Programmatic Strategies. One accepted paper and two show-and-tell sessions discussed effective methods for searching in program space to improve human playing strength using explainable programs and finding best-response strategies. 
  • The second theme was Esports Analytics. The invited industry talk and a show-and-tell session focused on how we can improve human game-playing and viewing experiences by analyzing game data. 
  • Our third theme was AI for Strategic Games. Here, two accepted papers and two show-and-tell contributions discussed improving agents for the game “Baba is You” and for RTS games by reinforcement learning, the use of transformers, by combining local and global search for better pathfinding, and applying standard single-agent search algorithms as opposed to learning playing policies. 

The workshop organizers (Derek Martin (SG), Ben Watson (EA), Erika Kleinman (EA), and Michael Buro (SG)) would like to thank all presenters and attendees for making the workshop a successful event. We are looking forward to next year’s sequel. 

More details about the workshop including a more detailed summary are available at www.cs.ualberta.ca/˜mburo/aiide22ws . 

INTELLIGENT NARRATIVE TECHNOLOGIES (W3) 

The Intelligent Narrative Technologies (INT) workshop series aims to advance research in artificial intelligence for the computational understanding and expression of narrative. Recent years have witnessed significant advances in the technical, creative, and aesthetic interpretation of narratives with digital media, including games, simulations, interactive fiction, and electronic literature. Our goal is to contribute to this forward momentum by gathering a multidisciplinary group of researchers and practitioners to share their latest work at the intersection of narrative and technology. 

After two years, the Intelligent Narrative Technologies Workshop returned for its 13th meeting featuring five papers and a community and panel discussion. There were fifteen registered attendees, with approximately fifteen additional interested onlookers. There were five accepted papers. Each paper was accompanied by an oral presentation and discussion period to focus on its contributions. The topics of the papers included augmented reality agents with emotional planning, quest generation, novelist methods for structuring stories, search for agent behavior, and deep learning for mixed-initiative story-image generation. Techniques from the planning community were prominently featured and included the A Behavior Language reactive planner, the Sabre narrative planner, and Monte Carlo tree search for next-step narrative planning. 

 Following the paper presentations and discussions, the workshop discussed the history and future of Intelligent Narrative Technologies in a panel format. Prominent topics included the relevance of the long-held distinction between author-based and character-based approaches and the role of machine learning in our subfield. Nearly half of the panel discussion focused on the importance of cultural, racial, and gender diversity within our technical system. The need for this inclusivity applies both to those who are developing these systems, as well as the narrative artifacts these systems produce. 

Workshop co-chair Joshua McCoy presented a summary report which included links to all previous iterations of the workshop. It is to be noted that several of these website links are no longer active, and can only be discovered via web archives. The workshop organizers recommend, for the future longevity of the workshop’s proceedings, for there to be a central Intelligent Narrative Technology website that is maintained and utilized from year to year, as opposed to the decentralized bespoke workshop websites that are prone to atrophy as years pass. 

Joshua McCoy, Mike Treanor, and Ben Samuel served as cochairs of this workshop.  

AUTHORS 

Michael Buro is a professor in the Computing Science Department at the University of Alberta in Edmonton, Canada. 

Derek Martin is a Ph.D. student at North Carolina State University, Raleigh, NC, USA. 

Joshua McCoy is an assistant professor in the departments of Computer Science and Cinema and Digital Media at UC Davis.  

Ben Samuel is an assistant professor in the department of Computer Science at the University of New Orleans.