< Back iGGi Talk at Develop:Brighton - Dominik Jeurissen iGGi PG Researcher Dominik Jeurissen held a talk on " LLM Agents For QA - Potential & Limitations " at this year's Develop:Brighton conference. Abstract: With tight deadlines and a constantly evolving game, properly testing a game is challenging. Using AI agents to simplify this work sounds promising, but machine learning is often too slow, and manually implementing the agents takes time. As such, one particularly exciting application for QA is to use Large Language Models (LLMs) as zero-shot game-playing agents. LLM-based agents can play games without pre-training, making them a valuable asset to test a constantly changing game. But how well do they play games? What are their strengths, and what do they struggle with? In this session, we will review how to implement zero-shot agents with LLMs and show examples of existing LLM-based game-playing agents. We will also show that although these agents have many limitations, they have the potential to be a valuable tool for QA to automate many repetitive tasks. The objectives of Dominik's talk were to provide the audience with an overview of the cutting-edge research on LLM-based zero-shot game-playing agents show what these agents can do well and what their limitations are give practical tips on how to utilize LLM agents as QA tools Dominik's talk has been recorded and will be made available to Develop ticket holders. Please contact Dominik directly if you have any queries regarding the presentation. Previous 11 Jul 2024 Next

< Back Testing TileAttack with Three Key Audiences Link Author(s) C Madge, M Poesio, U Kruschwitz, J Chamberlain Abstract More info TBA Link

iGGi Partners We are excited to be collaborating with a number of industry partners. IGGI works with industry in some of the following ways: Student Industry Knowledge Transfer - this can take many forms, from what looks like a traditional placement, to a short term consultancy, to an ongoing relationship between the student and their industry partner. Student Sponsorship - for some of our students, their relationship with their industry partner is reinforced by sponsorship from the company. This is an excellent demonstration of the strength of the commitment and the success of the collaborations. In Kind Contributions - IGGI industry partners can contribute by attending and/or featuring in our annual conference, offering their time to give talks and masterclasses for our students, or even taking part in our annual game jam! There are many ways for our industry partners to work with IGGI. If you are interested in becoming involved, please do contact us so we can discuss what might be suitable for you. Splash Damage

< Back A local approach to forward model learning: Results on the game of life game Link Author(s) SM Lucas, A Dockhorn, V Volz, C Bamford, RD Gaina, I Bravi, ... Abstract More info TBA Link

< Back The AI art school: A speculative design workshop Link Author(s) D Barrios-O'Neill, L O'Boyle, N Binyamini Ben Meir Abstract More info TBA Link

< Back Bodystorming in SocialVR to Support Collaborative Embodied Ideation Link Author(s) C Gonzalez Diaz, R Fiebrink, P Perry, R Gibson, B Martelli, S Deterding, ... Abstract More info TBA Link

< Back GAIG Meetup 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/ Previous 1 Mar 2023 Next

< Back Rolling Horizon NEAT for General Video Game Playing Link Author(s) D Perez-Liebana, MS Alam, RD Gaina Abstract More info TBA Link

< Back An Exploratory Analysis of Student Experiences with Peer Evaluation in Group Game Development Projects Link Author(s) A Mitchell, M Scott, J Walton-Rivers, M Watkins, W New, D Brown Abstract More info TBA Link

< Back "I just wanted to get it over and done with": a grounded theory of psychological need frustration in video games Link Author(s) N Ballou, S Deterding Abstract More info TBA Link

< Back Making Connections: Neurodevelopmental Changes in Brain Connectivity after Adverse Experiences in Early Adolescence Link Author(s) A Pollmann, R Sasso, K Bates, D Fuhrmann Abstract More info TBA Link

< Back Bootstrap Your Own Teacher: Online Policy Distillation for Multi-Game Reinforcement Learning Link Author(s) DJ Byrne, M Tot, P Duckworth, C Bonnet, A Laterre, TD Barrett Abstract More info TBA Link