The recent Game AI Meetup took place on 01 March 2023. Talks and presentation included:

Jakob Foerster (University of Oxford, UK): Opponent-Shaping and Interference in General-Sum Games

Original talk abstract: In general-sum games, the interaction of self-interested learning agents commonly leads to collectively worst-case outcomes, such as defect-defect in the iterated prisoner's dilemma (IPD). To overcome this, some methods, such as Learning with Opponent-Learning Awareness (LOLA), shape their opponents' learning process. However, these methods are myopic since only a small number of steps can be anticipated, are asymmetric since they treat other agents as naive learners, and require the use of higher-order derivatives, which are calculated through white-box access to an opponent's differentiable learning algorithm.

In this talk I will first introduce Model-Free Opponent Shaping (M-FOS), which overcomes all of these limitations. M-FOS learns in a meta-game in which each meta-step is an episode of the underlying (``inner'') game. The meta-state consists of the inner policies, and the meta-policy produces a new inner policy to be used in the next episode. M-FOS then uses generic model-free optimisation methods to learn meta-policies that accomplish long-horizon opponent shaping.

I will finish off the talk with our recent results for adversarial (or cooperative) cheap-talk: How can agents interfere with (or support) the learning process of other agents without being able to act in the environment?

Vanessa Volz (modl.ai): Establishing Trust in AI-based Tools for Game Development

Original talk abstract: AI-based tools to support the game development process have long been a topic in Game AI research, with popular publications in testing, churn prediction, asset, level and even game generation. However, the adaptation of these techniques from the games industry has been hesitant at best: The small-scale and simplified examples researchers use to demonstrate their work understandably only seldom convince the industry to risk investing in AI tools.

In this talk, I will speak about my experience establishing trust in AI-based tools to support creative processes in game development. Having worked on this topic in both industry and academia, I will address issues ranging from establishing a common language and explaining AI behaviour to issuing performance guarantees via benchmarking and theoretical analysis.

Mike Preuss (Leiden University, The Netherlands): In the eye of the storm? Where are we going with game AI?

Original talk abstract: Looking back at the last 10 years of research in Game AI we find that Big Tech research has shaken up things quite a lot. A number of challenges were resolved in record time (Go, StarCraft, etc) and AI algorithm development is probably still increasing in speed. However, it seems that the use of AI in game-making has not changed that much, and academic research often opts for "smaller problems", slowly turning towards Human-Centered AI as possibly most important general research direction. How can we approach the next leap predicted by Alex Champandard 10 years ago of really intelligent game AI? And where would we want that? Mike presents some inconclusive thoughts and ideas on future developments.

The Game AI Meetup takes place several times a year. To sign up and receive updates, please register/join here: https://www.meetup.com/game-ai-meetup-gaim-of-london/