Search Results

< Back Dr Martin Balla Queen Mary University of London iGGi Alum Before starting his PhD Martin studied Computer Science at the University of Essex. His main interest is artificial intelligence and its application to all sort of problems ranging from computer vision to game AI. He likes spending his spare time with various activities which mainly involves reading, playing video games and skateboarding. Martin's PhD thesis focuses on Reinforcement Learning agents that can adapt to changes in the reward function and/or changes in the environment. His work investigates how agents can transfer their knowledge to changes in the environment, such as new rewards, levels or visuals. Outside of his main research direction, Martin is involved with the Tabletop games framework (TAG), which is a collection of various tabletop games implemented with a common API with a focus on various game-playing agents (including RL). TAG brings various challenges to RL agents compared to search-based agents, such as complex action spaces, unique observation spaces (various embeddings), multi-agent dynamics with competitive and collaborative aspects, and lots of hidden information and stochasticity. Email m.balla@qmul.ac.uk Website LinkedIn Mastodon BlueSky GitHub Other Link Supervisors: Dr Diego Pérez-Liébana Prof. Simon Lucas Featured Publication(s): Multi-task Reinforcement Learning for Adaptive Agents Program Committee and Subreviewers Tiny Moves: Game-based Hypothesis Refinement PyTAG: Tabletop Games for Multi-Agent Reinforcement Learning PyTAG: Challenges and Opportunities for Reinforcement Learning in Tabletop Games TAG: Pandemic Competition Task Relabelling for Multi-task Transfer using Successor Features TAG: A tabletop games framework Design and implementation of TAG: a tabletop games framework Evaluating generalisation in general video game playing Evaluating Generalization in General Video Game Playing Analysis of statistical forward planning methods in Pommerman Themes Game AI - Previous Next

< Back Dr Oliver Scholten University of York iGGi Alum Oliver Scholten is working on understanding the use of cryptocurrency technologies for gambling and gaming. His work provides researchers with the tools and context needed to understand player behaviours in these technologically advanced domains. He is the creator of gamba - a python library designed to enable quick replication of existing player behaviour tracking studies. He has also published several peer reviewed articles, and had written evidence published by the UK House of Lords which describes the mechanics behind decentralised gambling applications. As a PhD student, his thesis focuses on decoding and analysing cryptocurrency gambling and cryptocurrency gaming transactions. These transactions offer a more granular insight for researchers into both gambling and gaming than has been historically possible, this work therefore lays the foundations for explorations across different schools of research, and more specifically, advanced player transaction analytics. Please note: Updating of profile text in progress Email oliver@gamba.dev Website LinkedIn Mastodon BlueSky GitHub Other Link Featured Publication(s): On the Evaluation of Procedural Level Generation Systems On the Behavioural Profiling of Gamblers Using Cryptocurrency Transaction Data Inside the decentralised casino: A longitudinal study of actual cryptocurrency gambling transactions Decentralised Gambling Overview Decentralised Gambling: The York Combined Transaction Set Unconventional Exchange: Methods for Statistical Analysis of Virtual Goods Utilising VIPER for Parameter Space Exploration in Agent Based Wealth Distribution Models Ethereum Crypto-Games: Mechanics, Prevalence, and Gambling Similarities Themes Game Data - Previous Next

< Back Author's declaration Link Author(s) D Ratcliffe Abstract More info TBA Link

< Back Incremental game mechanics applied to text annotation Link Author(s) C Madge, R Bartle, J Chamberlain, U Kruschwitz, M Poesio Abstract More info TBA Link

< Back Examining the effects of video game difficulty adaptation on performance and player experience Link Author(s) M Frister, P Cairns, F McNab Abstract More info TBA Link

< Back iGGi Research Retreat "Unconference" Group Outcomes The Future of AI The "Problem" We discussed what the "future of AI" might look like, how it will change us as a society (for better and worse) and what possibilities it would create in the future. What we did As you can imagine, the "future of AI" is somewhat of a broad and undirected topic. Therefore in the morning we allowed free flowing conversations to see where it went and then towards the end tried to join up the threads into the things that we thought were the most worthy of further analysis and thought. In the afternoon we tackled the specifics of how to approach a game with emergent characters and stories, a topic oft dreamed of by game designers, but hitherto unattainable. The "Outcome" The morning discussions: Our discussions were wide-ranging, but the opening question captured the essence of our inquiry. From personalised AI assistants we quickly moved to the broader economics of AI — circling back again and again to two themes: whether AI can ever replicate the human experience, and the friction between the utopian ideals we project onto it and the gravitational pull of capitalism. I have sought to recount our discussion on these themes as accurately as memory allows, adding only modest(?) embellishment where it aids narrative coherence. On AI and human interaction: As someone quoted: “Do you know what it smells like in the Sistine Chapel?” (Good Will Hunting). The line reminds us that knowledge can be learned, but wisdom must be lived. Does an AI know what it is to be human? Can it ever truly understand the human experience? We are not just a collection of data points; we are a tapestry of emotions, experiences, and connections. AI can analyse patterns, but can it ever grasp the essence of what it means to be human? Human art matters because it exposes something fragile. To create is to risk oneself: to bleed, to reveal, to offer a fragment of the human condition for others to recognise. AI can imitate the form, but form without risk is mimicry. Can imitation ever supply the soul? Perhaps all we truly crave, as a species, is to be seen — to connect with each other. Yet history suggests authenticity is not always required. Chess engines long ago surpassed every human master, yet millions still prefer to watch people play. Calculators did not end mathematics; they expanded it, making it more ambitious and more accessible. Technology rarely erases human practice — but it does reframe it. The question is not only whether we can connect with AI as with another human, but whether we will still insist on doing so. AI companions and “digital girlfriends” already suggest that some are content with machine-mediated intimacy. The unsettling prospect is not that AI lacks a soul, but that we may cease to care. What happens when a generation grows up regarding “connection enough” as something delivered by code? If we defer not only thought but also empathy, attention, and intimacy to our machines, what remains distinctly human? We shape our tools, and thereafter our tools shape us (Culkin). On AI and economics: AI will not escape the gravitational pull of commerce. As today’s internet is financed by advertising, it is inevitable that AI systems will be bent to the same imperatives — nudging our choices, steering our attention, and monetising our interactions. Already we see the first signs: an Alexa Show inserting shopping prompts directly into the home. For all our talk of AI safety and ethics, it is commerce that drives development. But once human labour itself is displaced, what then? A utopia of leisure where we are free to follow our passions? Or a dystopia in which a minority, owning the means of cognition itself, consign the rest of us to redundancy? Could a society without labour even cohere — or would it demand a wholesale reinvention of politics, economy, and democracy itself? As we tried to weave our threads together into something coherent, this was the question that seemed most fertile for further debate. What is the politics of AI? What would a political and economic system look like that could accommodate these changes? How do we ensure that the benefits of AI are shared equitably, rather than concentrating power and wealth in the hands of a few? Our discussions were rich, unsettling, and illuminating, tracing both promise and peril. Yet the future owes no loyalty to our prophecies. The greater danger is not that AI will fail to know us, but that, in its shadow, we will lose the thread of ourselves. In the afternoon: Having solved the future of AI before lunch, we turned in the afternoon to the far more mundane task of reinventing the games industry. One of the industries evergreen obsessions is how to make characters and narratives more believable. As games grow ever more immersive, the hunger for deeper storytelling only intensifies. Yet, despite extraordinary technical progress, this is the frontier we keep failing to cross — because the obstacle is less technical, and more human. The costs of creating content are crushing. Large language models offer a tantalising shortcut: a machine that might spin endless dialogue, branching quests, even whole worlds on demand. But this promise comes bound up with limitations so profound they may be unsolvable with current methods. My own curiosity lies in a hybrid approach: using generative AI not as an all-purpose author, but as a tool to help construct traditional symbolic systems — frameworks that could give structure and coherence while still leaving room for human craft. If it worked, it might nudge us forward in this arena without the problems that come with the other approaches. But this is not a conversation one can leap into lightly. It demands deep knowledge of how games are actually made, tested, and sold, as well as a sober reckoning with both the failures and the potential of LLMs. Much of our discussion was spent simply reaching the starting line. The groups diversity produced fresh perspectives, but the depth of the subject meant we could not advance far in the time available. What did become clear was this: the problem is not just technical, it is communicative. If this debate is to progress, the challenge must be articulated in a way that is accessible beyond a narrow circle of experts. That, I realised, is the real work still ahead. Post Script: We live in a time of unprecedented change (at least in living memory). The world has enjoyed relative peace up until about 2020 but it feels like the geopolitical sands are shifting in a once-in-a-century phenomenon and such change has wide-ranging global political and economic implications for modern society broadly, but specifically within technology. This scenario is quite relevant to some of our discussions; the place and role of AI in our society is hard to gauge when our society is going through some fairly tectonic shifts. I think it will be a job for historians in the future to determine whether the emergence of advanced AI and these changes are correlation or coincidence, but it is clear that we cannot evaluate and analyse AI within a vacuum. The wider context is key here and that context is both nebulous and shifting all the time. Perhaps AI can support such a perspective in ways that one (or many) human minds cannot comprehend at once? Previous Next Previous Next

< Back Cooperative games with partial observability Link Author(s) PR Williams, D Perez-Liebana, SM Lucas 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. Forma

< Back Better Dead than a Damsel: Gender Representation and Player Churn Link Author(s) Lauren Winter, Sarah Masters Abstract More info TBA Link

< Back Evaluating virtual reality experiences through participant choices Link Author(s) M Murcia-López, T Collingwoode-Williams, W Steptoe, R Schwartz, ... Abstract More info TBA Link

< Back VERTIGØ: visualisation of rolling horizon evolutionary algorithms in GVGAI Link Author(s) R Gaina, S Lucas, D Perez-Liebana Abstract More info TBA Link

< Back UCL+ Sheffield at SemEval-2016 Task 8: Imitation learning for AMR parsing with an alpha-bound Link Author(s) J Goodman, A Vlachos, J Naradowsky Abstract More info TBA Link