Procedural Content Generation (PCG) systems typically struggle to generate cohesive content across multiple domains. Large Language Models (LLMs) understand semantic relationships of game elements, at least when they are described in natural language. This paper investigates how LLMs and text-to-image models can generate narrative, visual, and gameplay contentcoherently. We leverage LLMs as a scaffold for an automated, theme-driven asset generation pipeline, enabling unique and scalable game experiences with no developer input besides setting up a game template. This paper introduces CrawLLM, a dungeon crawler with card combat mechanics, as a testbed for an LLM-driven game generation pipeline. A Mixtral 8x7B model generates game themes, guiding the creation of narrative content. Visual assets are produced via Stable Diffusion XL, using ControlNet and IP-Adapter modules to achieve game-ready formats. A user study conducted on snapshots of fully generated games indicates that the underlying semantic themes remain clearly discernible in many cases, although intended visual styles were less clear. This work demonstrates the potential of LLM-driven pipelines for PCG, while highlighting areas for improvement in content specificity.

In this demo paper we present Prompt Override, a serious game in which players engage in prompt-based hacking challenges by manipulating the system prompt of a large language model (LLM) to solve puzzles. The game features two LLMs: one as the target of the player's hacking attempts, and another as a rogue assistant guiding the player throughout the game. As players explore a simulated file system, editing prompt snippets, Prompt Override offers a unique take on the hacking game genre-one where rigid solution paths are replaced with more open-ended, language-driven problem solving. By leveraging realworld limitations and vulnerabilities of LLMs, Prompt Override encourages players to develop a deeper understanding of prompt engineering, model behavior, and the ethical implications of interacting with intelligent systems.

This paper introduces the Procedural Content Generation Benchmark for evaluating generative algorithms on different game content creation tasks.The benchmark comes with 12 game-related problems with multiple variants on each problem.Problems vary from creating levels of different kinds to creating rule sets for simple arcade games.Each problem has its own content representation, control parameters, and evaluation metrics for quality, diversity, and controllability.This benchmark is intended as a first step towards a standardized way of comparing generative algorithms.We use the benchmark to score three baseline algorithms: a random generator, an evolution strategy, and a genetic algorithm.Results show that some problems are easier to solve than others, as well as the impact the chosen objective has on quality, diversity, and controllability of the generated artifacts.

Can we replicate the power of evolutionary algorithms in discovering good and diverse game content via generative machine learning (ML) techniques?This question could subvert current trends in procedural content generation (PCG) and beyond.By learning the behavior of quality-diversity (QD) evolutionary algorithms through ML, we stand to overcome the computational challenges inherent in QD search and ensure that the benefits of QD search are reproduced by efficient generative models.We introduce a novel, end-to-end methodology named Machine Learning of Quality Diversity (MLQD) which is executed in two steps.First, tailored QD evolution creates large and diverse training datasets from the ground up.Second, sophisticated ML architectures such as the Transformer learn the datasets' underlying distributions, resulting in generative models that can emulate QD search via stochastic inference.We test MLQD on the use-case of generating strategy game map sketches, a task characterized by stringent constraints and a multidimensional feature space.Our findings are promising, demonstrating that the Transformer architecture can capture both the diversity and the quality traits of the training sets, successfully reproducing the behavior of a range of tested QD algorithms.This marks a significant advancement in our quest to automate the creation of high-quality, diverse game content, pushing the boundaries of what is possible in PCG and generative AI at large.

How do spaces make us feel? What is the perceived emotional impact of built form? This study proposes a framework to identify and model the effects that our perceived environment can have by taking into consideration illumination and structural form while acknowledging its temporal dimension. To study this, we recruited 100 participants via a crowd-sourcing platform in order to annotate their perceived arousal or pleasure shifts while watching videos depicting spatial navigation in first person view. Participants’ annotations were recorded as time-continuous unbounded traces, allowing us to extract ordinal labels about how their arousal or pleasure fluctuated as the camera moved between different rooms. Given the subjective nature of the task and the noisy signals from real-time annotation, a number of processing steps are applied in order to convert the data into ordinal relationships between affect metrics in different rooms. Experiments with random forests and other classifiers show that, with the right treatment and data cleanup, simple interior design features can be adequate predictors of human arousal and pleasure changes over time. The dataset is made available in order to prompt exploration of additional modalities as input and ground truth extraction.

Affective computing faces a pressing challenge: the limited ability of affect models to generalise amidst varying contextual factors within the same task. While well recognised, this challenge persists due to the absence of suitable large-scale corpora with rich and diverse contextual information within a domain. To address this challenge, this paper introduces a GameVibe, a novel corpus explicitly tailored to confront the lack of contextual diversity. The affect corpus is sourced from 30 First Person Shooter (FPS) games, showcasing diverse game modes and designs within the same domain. The corpus comprises 2 hours of annotated gameplay videos with engagement levels annotated by a total of 20 participants in a time-continuous manner. Our preliminary analysis on this corpus sheds light on the complexity of generalising affect predictions across contextual variations in similar affective computing tasks. These initial findings serve as a catalyst for further research, inspiring deeper inquiries into this critical, yet understudied, aspect of affect modelling.

Tools for designing content require a medium that allows the designer to efficiently express their creativity, and a system that ensures the content being designed adheres to the domain of interest. Interacting with Large Language Models (LLMs) via natural language is extremely intuitive for a human designer, although it remains largely unexplored. However, this approach has a limitation: LLMs are prone to hallucinations and they tend to ignore parts of the user request in their responses. One workaround is to let LLM use tools such as function calling to ensure consistency of the content. We formalise this approach by proposing LLMaker, a general framework for consistent video game content generation empowered by LLMs, bridging the gap between creative vision and technical execution. We demonstrate LLMaker’s application in generating dungeon crawler level layouts, comparing it against alternative LLM-based methods for content generation over multiple tests, testing for consistency of the outputs and elapsed time per request.

In this demo paper we present LLMaker, a videogame level design tool that operates solely via natural language instructions. Unlike existing level design tools, LLMaker allows the designer to express their intent in an intuitive way, interacting with a cognitively undemanding interface, while still ensuring the generated levels adhere to specific domain and playability constraints.

Video game players increasingly seek immersive and personalised experiences that resonate with their unique interests and personalities. This study explores the novel application of large language models (LLMs) in game re-theming, a process that adapts game assets to new settings and narratives. We applied this to an original game, CrawLLM, which combines dungeon crawling and card combat mechanics. The gameplay structure guided the construction of prompts for LLMs to generate new themes, stories, characters, and locations, which then directed the generation of corresponding visual assets. This approach enabled the game’s aesthetics to emerge from its underlying narrative. We hereby demonstrate a playable version of the game with 20 pre-generated, non-curated themes. This study paves the way for future research on automated game generation, where LLMs can orchestrate diverse content creation pipelines to construct entire games from high-level prompts, while maintaining cohesive user experiences.

How can we reliably transfer affect models trained in controlled laboratory conditions (in-vitro) to uncontrolled real-world settings (in-vivo)? The information gap between in-vitro and in-vivo applications defines a core challenge of affective computing. This gap is caused by limitations related to affect sensing including intrusiveness, hardware malfunctions and availability of sensors. As a response to these limitations, we introduce the concept of privileged information for operating affect models in real-world scenarios (in the wild). Privileged information enables affect models to be trained across multiple modalities available in a lab, and ignore, without significant performance drops, those modalities that are not available when they operate in the wild. Our approach is tested in two multimodal affect databases one of which is designed for testing models of affect in the wild. By training our affect models using all modalities and then using solely raw footage frames for testing the models, we reach the performance of models that fuse all available modalities for both training and testing. The results are robust across both classification and regression affect modeling tasks which are dominant paradigms in affective computing. Our findings make a decisive step towards realizing affect interaction in the wild.

Games are designed to elicit strong emotions during game play, especially when players are competing against each other. Artificial Intelligence applied to predict a player’s emotions has mainly been tested on single-player experiences in low-stakes settings and short-term interactions. How do players experience and manifest affect in high-stakes competitions, and which modalities can capture this? This paper reports a first experiment in this line of research, using a competition of the video game Hearthstone where both competing players’ game play and facial expressions were recorded over the course of the entire match which could span up to 41 minutes. Using two experts’ annotations of tension using a continuous video affect annotation tool, we attempt to predict tension from the webcam footage of the players alone. Treating both the input and the tension output in a relative fashion, our best models reach 66.3% average accuracy (up to 79.2% at the best fold) in the challenging leave-one-participant out cross-validation task. This initial experiment shows a way forward for affect annotation in games “in the wild” in high-stakes, real-world competitive settings.

Tabletop Role-Playing Games (TTRPGs) offer players the opportunity to form imaginary gameworlds and stories within them, create community, solve problems, and explore identity. Designers and researchers have tried to identify how aspects of TTRPGs facilitate collaboration, immersion, creativity, and more. However, there has been no attempt to develop a formal assessment methodology for player experience during TTRPG play. This paper argues that evaluating TTRPG players’ experience can provide vital data for Game Masters to improve on their future games, for players to reflect on their experience, and for TTRPG designers or event organizers to collect and compare data. As a first step towards developing such an evaluation method, we identify important dimensions of TTRPG play that can be meaningful to track and actionable to improve upon. Moreover, we review player experience dimensions and evaluation methods in digital games, and explore similarities and differences with TTRPGs.

Many human-AI authoring tools are used in a playful way, while being primarily designed for task-achievement—not playfulness. We argue that playfulness is an important yet overlooked factor of user behaviour and experience when interacting with such tools. Motivating and rewarding playfulness as an exploratory, task-agnostic, open, and subversive attitude can support the satisfaction of more diverse user goals, and have a strong, positive effect on the user experience, the emerging human-AI interaction, and the resulting artefact. In this paper, we motivate the importance of playfulness as user experience in human-AI authoring tools, and propose concrete strategies to design for playfulness in the human user through UI design, in the AI through algorithms, or through interventions to their dialog. We conclude with an outlook of the research agenda.

What if emotion could be captured in a general and subject-agnostic fashion? Is it possible, for instance, to design general-purpose representations that detect affect solely from the pixels and audio of a human-computer interaction video? In this article we address the above questions by evaluating the capacity of deep learned representations to predict affect by relying only on audiovisual information of videos. We assume that the pixels and audio of an interactive session embed the necessary information required to detect affect. We test our hypothesis in the domain of digital games and evaluate the degree to which deep classifiers and deep preference learning algorithms can learn to predict the arousal of players based only on the video footage of their gameplay. Our results from four dissimilar games suggest that general-purpose representations can be built across games as the arousal models obtain average accuracies as high as 85 percent using the challenging leave-one-video-out cross-validation scheme. The dissimilar audiovisual characteristics of the tested games showcase the strengths and limitations of the proposed method.

This article proposes a procedural content generator which evolvesMinecraftbuildings according to an open-ended and intrinsic definition of novelty. To realize this goal, we evaluate individuals’ novelty in the latent space using a 3-D autoencoder (AE), and alternate between phases of exploration and transformation. During exploration the system evolves multiple populations of CPPNs through CPPN-NEAT and constrained novelty search in the latent space (defined by the current AE). We apply a set of repair and constraint functions to ensure candidates adhere to basic structural rules during evolution. During transformation, we reshape the boundaries of the latent space to identify new interesting areas of the solution space by retraining the AE with novel content. In this study, we evaluate five different approaches for training the AE during transformation and its impact on populations’ quality and diversity during evolution. Our results show that by retraining the AE we can achieve better open-ended complexity compared to a static model, which is further improved when retraining using larger datasets of individuals with diverse complexities.

Video game testing has become a major investment of time, labor, and expense in the game industry. Particularly the balancing of in-game units, characters, and classes can cause long-lasting issues that persist years after a game's launch. While approaches incorporating artificial intelligence have already shown successes in reducing manual effort and enhancing game development processes, most of these draw on heuristic, generalized, or optimal behavior routines, while actual low-level decisions from individual players and their resulting playing styles are rarely considered. In this article, we applydeep player behavior modelingto turn atomic actions of 213 players from six months of single-player instances within the MMORPG Aion into generative models that capture and reproduce particular playing strategies. In a subsequent simulation, the resulting generative agents (“replicants”) were tested against common NPC opponent types of MMORPGs that iteratively increased in difficulty, respective to the primary factor that constitutes this enemy type (Melee, Ranged, Rogue, Buffer, Debuffer, Healer, Tank, or Group). As a result, imbalances between classes as well as strengths and weaknesses regarding particular combat challenges could be identified and regulated automatically.

This paper introduces a paradigm shift by viewing the task of affect modeling as a reinforcement learning (RL) process. According to the proposed paradigm, RL agents learn a policy (i.e. affective interaction) by attempting to maximize a set of rewards (i.e. behavioral and affective patterns) via their experience with their environment (i.e. context). Our hypothesis is that RL is an effective paradigm for interweaving affect elicitation and manifestation with behavioral and affective demonstrations. Importantly, our second hypothesis-building on Damasio's so-matic marker hypothesis-is that emotion can be the facilitator of decision-making. We test our hypotheses in a racing game by training Go-Blend agents to model human demonstrations of arousal and behavior; Go-Blend is a modified version of the Go-Explore algorithm which has recently showcased supreme performance in hard exploration tasks. We first vary the arousal-based reward function and observe agents that can effectively display a palette of affect and behavioral patterns according to the specified reward. Then we use arousal-based state selection mechanisms in order to bias the strategies that Go-Blend explores. Our findings suggest that Go-Blend not only is an efficient affect modeling paradigm but, more importantly, affect-driven RL improves exploration and yields higher performing agents, validating Damasio's hypothesis in the domain of games.

Self-supervised learning (SSL) techniques have been widely used to learn compact and informative representations from high-dimensional complex data. In many computer vision tasks, such as image classification, such methods achieve state-of-the-art results that surpass supervised learning approaches. In this paper, we investigate whether SSL methods can be leveraged for the task of learning accurate state representations of games, and if so, to what extent. For this purpose, we collect game footage frames and corresponding sequences of games’ internal state from three different 3D games: VizDoom, the CARLA racing simulator and the Google Research Football Environment. We train an image encoder with three widely used SSL algorithms using solely the raw frames, and then attempt to recover the internal state variables from the learned representations. Our results across all three games showcase significantly higher correlation between SSL representations and the game’s internal state compared to pre-trained baseline models such as ImageNet. Such findings suggest that SSL-based visual encoders can yield general—not tailored to a specific task—yet informative game representations solely from game pixel information. Such representations can, in turn, form the basis for boosting the performance of downstream learning tasks in games, including gameplaying, content generation and player modeling.

Using artificial intelligence (AI) to automatically test a game remains a critical challenge for the development of richer and more complex game worlds and for the advancement of AI at large. One of the most promising methods for achieving that long-standing goal is the use of generative AI agents, namely procedural personas, that attempt to imitate particular playing behaviors which are represented as rules, rewards, or human demonstrations. All research efforts for building those generative agents, however, have focused solely on playing behavior which is arguably a narrow perspective of what a player actually does in a game. Motivated by this gap in the existing state of the art, in this paper we extend the notion of behavioral procedural personas to cater for player experience, thus examining generative agents that can both behave and experience their game as humans would. For that purpose, we employ the Go-Explore reinforcement learning paradigm for training human-like procedural personas, and we test our method on behavior and experience demonstrations of more than 100 players of a racing game. Our findings suggest that the generated agents exhibit distinctive play styles and experience responses of the human personas they were designed to imitate. Importantly, it also appears that experience, which is tied to playing behavior, can be a highly informative driver for better behavioral exploration.

This paper presents a generative architecture for general-purpose room layouts that can be treated as geometric definitions of dungeons, mansions, shooter levels and more. The motivation behind this work is to provide a design tool for virtual environments that combines aspects of controllability, expressivity and generality. Towards that end, a two-tier level representation is realized, with a graph-based design specification constraining and guiding the generated geometries, facilitated by constrained evolutionary search. Expressivity is secured through quality-diversity search which can provide the designer with a broad variety of level layouts to choose from. Finally, the generator is general-purpose as it can produce layouts based on different types of static grid structures or as free-form, curved structures through an adaptive Voronoi diagram that is evolved along with the level itself. The method is tested on a variety of design specifications and grid types, and results show that even with complex design constraints or malleable grids the algorithm can produce a broad variety of levels.

Having access to accurate game state information is of utmost importance for any artificial intelligence task including game-playing, testing, player modeling, and procedural content generation. Self-Supervised Learning (SSL) techniques have shown to be capable of inferring accurate game state information from the high-dimensional pixel input of game footage into compressed latent representations. Contrastive Learning is a popular SSL paradigm where the visual understanding of the game’s images comes from contrasting dissimilar and similar game states defined by simple image augmentation methods. In this study, we introduce a new game scene augmentation technique—named GameCLR—that takes advantage of the game-engine to define and synthesize specific, highly-controlled renderings of different game states, thereby, boosting contrastive learning performance. We test our GameCLR technique on images of the CARLA driving simulator environment and compare it against the popular SimCLR baseline SSL method. Our results suggest that GameCLR can infer the game’s state information from game footage more accurately compared to the baseline. Our proposed approach allows us to conduct game artificial intelligence research by directly utilizing screen pixels as input.

Stochastic gradient descent (SGD) is a premium optimization method for training neural networks, especially for learning objectively defined labels such as image objects and events. When a neural network is instead faced with subjectively defined labels---such as human demonstrations or annotations---SGD may struggle to explore the deceptive and noisy loss landscapes caused by the inherent bias and subjectivity of humans. While neural networks are often trained via preference learning algorithms in an effort to eliminate such data noise, the de facto training methods rely on gradient descent. Motivated by the lack of empirical studies on the impact of evolutionary search to the training of preference learners, we introduce the RankNEAT algorithm which learns to rank through neuroevolution of augmenting topologies. We test the hypothesis that RankNEAT outperforms traditional gradient-based preference learning within the affective computing domain, in particular predicting annotated player arousal from the game footage of three dissimilar games. RankNEAT yields superior performances compared to the gradient-based preference learner (RankNet) in the majority of experiments since its architecture optimization capacity acts as an eficient feature selection mechanism, thereby, eliminating overfitting. Results suggest that RankNEAT is a viable and highly eficient evolutionary alternative to preference learning.

Affect modeling is viewed, traditionally, as the process of mapping measurable affect manifestations from multiple modalities of user input to affect labels. That mapping is usually inferred through end-to-end (manifestation-to-affect) machine learning processes. What if, instead, one trains general, subject-invariant representations that consider affect information and then uses such representations to model affect? In this paper we assume that affect labels form an integral part, and not just the training signal, of an affect representation and we explore how the recent paradigm of contrastive learning can be employed to discover general high-level affect-infused representations for the purpose of modeling affect. We introduce three different supervised contrastive learning approaches for training representations that consider affect information. In this initial study we test the proposed methods for arousal prediction in the RECOLA dataset based on user information from multiple modalities. Results demonstrate the representation capacity of contrastive learning and its efficiency in boosting the accuracy of affect models. Beyond their evidenced higher performance compared to end-to-end arousal classification, the resulting representations are general-purpose and subject-agnostic, as training is guided though general affect information available in any multimodal corpus.

How can we model affect in a general fashion, across dissimilar tasks, and to which degree are such general representations of affect even possible? To address such questions and enable research towardsgeneralaffective computing, this paper introduces The Arousal video Game AnnotatIoN (AGAIN) dataset. AGAIN is a large-scale affective corpus that features over 1,100 in-game videos (with corresponding gameplay data) from nine different games, which are annotated for arousal from 124 participants in a first-person continuous fashion. Even though AGAIN is created for the purpose of investigating the generality of affective computing across dissimilar tasks, affect modelling can be studied within each of its 9 specific interactive games. To the best of our knowledge AGAIN is the largest—over 37 hours of annotated video and game logs—and most diverse publicly available affective dataset based on games as interactive affect elicitors.

This article introduces SuSketch, a design tool for first person shooter levels. SuSketch provides the designer with gameplay predictions for two competing players of specific character classes. The interface allows the designer to work side-by-side with an artificially intelligent creator and to receive varied types of feedback such as path information, predicted balance between players in a complete playthrough, or a predicted heatmap of the locations of player deaths. The system also proactively designs alternatives to the level and class pairing, and presents them to the designer as suggestions that improve the predicted balance of the game. SuSketch offers a new way of integrating machine learning into mixed-initiative cocreation tools, as a surrogate of human play trained on a large corpus of artificial playtraces. A user study with 16 game developers indicated that the tool was easy to use, but also highlighted a need to make SuSketch more accessible and more explainable.

Competitive board games have provided a rich and diverse testbed for artificial intelligence. This article contends that collaborative board games pose a different challenge to artificial intelligence as it must balance short-term risk mitigation with long-term winning strategies. Collaborative board games task all players to coordinate their different powers or pool their resources to overcome an escalating challenge posed by the board and a stochastic ruleset. This article focuses on the exemplary collaborative board game Pandemic and presents a rolling horizon evolutionary algorithm designed specifically for this game. The complex way in which the Pandemic game state changes in a stochastic but predictable way required a number of specially designed forward models, macroaction representations for decision making, and repair functions for the genetic operations of the evolutionary algorithm. Variants of the algorithm which explore optimistic versus pessimistic game state evaluations, different mutation rates, and event horizons are compared against a baseline hierarchical policy agent. Results show that an evolutionary approach via short-horizon rollouts can better account for the future dangers that the board may introduce, and guard against them. Results highlight the types of challenges that collaborative board games pose to artificial intelligence, especially for handling multiplayer collaboration interactions.

A key challenge of affective computing research is discovering ways to reliably transfer affect models that are built in the laboratory to real world settings, namely in the wild. The existing gap between in vitro and in vivo affect applications is mainly caused by limitations related to affect sensing including intrusiveness, hardware malfunctions, availability of sensors, but also privacy and security. As a response to these limitations in this paper we are inspired by recent advances in machine learning and introduce the concept of privileged information for operating affect models in the wild. The presence of privileged information enables affect models to be trained across multiple modalities available in a lab setting and ignore modalities that are not available in the wild with no significant drop in their modeling performance. The proposed privileged information framework is tested in a game arousal corpus that contains physiological signals in the form of heart rate and electrodermal activity, game telemetry, and pixels of footage from two dissimilar games that are annotated with arousal traces. By training our arousal models using all modalities (in vitro) and using solely pixels for testing the models (in vivo), we reach levels of accuracy obtained from models that fuse all modalities both for training and testing. The findings of this paper make a decisive step towards realizing affect interaction in the wild.

Assessing the believability of agents, characters and simulated actors is a core challenge for human computer interaction. While numerous approaches are suggested in the literature, they are all limited to discrete and low-granularity representations of believable behavior. In this paper we view believability, for the first time, as a time-continuous phenomenon and we explore the suitability of two different affect annotation schemes for its assessment. In particular, we study the degree to which we can predict character believability in a continuous fashion through a two-player game study. The game features various opponent behaviors that are assessed for their believability by 89 participants that played the game and then annotated their recorded playthrough. Random forest models are then trained to predict believability based on ad-hoc designed in-game features. Results suggest that a discrete annotation method leads to a more robust assessment of the ground truth and subsequently better modelling performance. Our best models are able to predict a change in perceived believability with a 72.5% accuracy on average (up to 90% in the best cases) in a time-continuous manner.

How does the form of our surroundings impact the ways we feel? This paper extends the body of research on the effects that space and light have on emotion by focusing on critical features of architectural form and illumination colors and their spatiotemporal impact on arousal. For that purpose, we solicited a corpus of spatial transitions in video form, lasting over 60 minutes, annotated by three participants in terms of arousal in a time-continuous and unbounded fashion. We process the annotation traces of that corpus in a relative fashion, focusing on the direction of arousal changes (increasing or decreasing) as affected by changes between consecutive rooms. Results show that properties of the form such as curved or complex spaces align highly with increased arousal. The analysis presented in this paper sheds some initial light in the relationship between arousal and core spatiotemporal features of form that is of particular importance for the affect-driven design of architectural spaces.

With the ubiquitous role of Artificial Intelligence (AI) in everyday applications such as smartphones and social media, children need digital literacy skills to navigate the digital world, critically view, and reflect on the social and ethical implications of the design and architecture of AI systems. To address this increasing need for AI literacy skills, particularly for younger students, this paper presents the rationale of the LearnML project which aims to develop a framework and game-based educational material for promoting AI literacy among primary and secondary education students. We also describe the design and initial assessment of the game “ArtBot”, developed as part of the LearnML project. We review existing literature, discuss the educational game design and development of “ArtBot” and describe the initial feedback of students and teachers. Our goal is to provide insights and suggest guidelines for the implementation of game-based learning environments for supporting AI literacy skills of students.

To which degree can abstract gameplay metrics capture the player experience in a general fashion within a game genre? In this comprehensive study we address this question across three different videogame genres: racing, shooter, and platformer games. Using high-level gameplay features that feed preference learning models we are able to predict arousal accurately across different games of the same genre in a large-scale dataset of over 1, 000 arousal-annotated play sessions. Our genre models predict changes in arousal with up to 74% accuracy on average across all genres and 86% in the best cases. We also examine the feature importance during the modelling process and find that time-related features largely contribute to the performance of both game and genre models. The prominence of these game-agnostic features show the importance of the temporal dynamics of the play experience in modelling, but also highlight some of the challenges for the future of general affect modelling in games and beyond.

Representing games through their pixels offers a promising approach for building general-purpose and versatile game models. While games are not merely images, neural network models trained on game pixels often capture differences of the visual style of the image rather than the content of the game. As a result, such models cannot generalize well even within similar games of the same genre. In this paper we build on recent advances in contrastive learning and showcase its benefits for representation learning in games. Learning to contrast images of games not only classifies games in a more efficient manner; it also yields models that separate games in a more meaningful fashion by ignoring the visual style and focusing, instead, on their content. Our results in a large dataset of sports video games containing lOOk images across 175 games and 10 game genres suggest that contrastive learning is better suited for learning generalized game representations compared to conventional supervised learning. The findings of this study bring us closer to universal visual encoders for games that can be reused across previously unseen games without requiring retraining or fine-tuning,

A core challenge of evolutionary search is the need to balance between exploration of the search space and exploitation of highly fit regions. Quality-diversity search has explicitly walked this tightrope between a population's diversity and its quality. This paper extends a popular quality-diversity search algorithm, MAP-Elites, by treating the selection of parents as a multi-armed bandit problem. Using variations of the upper-confidence bound to select parents from under-explored but potentially rewarding areas of the search space can accelerate the discovery of new regions as well as improve its archive's total quality. The paper tests an indirect measure of quality for parent selection: the survival rate of a parent's offspring. Results show that maintaining a balance between exploration and exploitation leads to the most diverse and high-quality set of solutions in three different testbeds.

This paper proposes a framework for the procedural generation of level and ruleset components of games via a surrogate model that assesses their quality and complementarity. The surrogate model combines level and ruleset elements as input and gameplay outcomes as output, thus constructing a mapping between three different facets of games. Using this model as a surrogate for expensive gameplay simulations, a search-based generator can adapt content toward a target gameplay outcome. Using a shooter game as the target domain, this paper explores how parameters of the players' character classes can be mapped to both the level's representation and the gameplay outcomes of balance and match duration. The surrogate model is built on a deep learning architecture, trained on a large corpus of randomly generated sets of levels, classes, and simulations from game playing agents. Results show that a search-based generative approach can adapt character classes, levels, or both toward designer-specified targets. The model can thus act as a design assistant or be integrated in a mixed-initiative tool. Most importantly, the combination of three game facets into the model allows it to identify the synergies between levels, rules, and gameplay and orchestrate the generation of the former two toward desired outcomes.

Balancing the options available to players in a way that ensures rich variety and viability is a vital factor for the success of any video game, and particularly competitive multiplayer games. Traditionally, this balancing act requires extensive periods of expert analysis, play testing and debates. While automated gameplay is able to predict outcomes of parameter changes, current approaches mainly rely on heuristic or optimal strategies to generate agent behavior. In this paper, we demonstrate the use of deep player behavior models to represent a player population (n = 213) of the massively multiplayer online role-playing game Aion, which are used, in turn, to generate individual agent behaviors. Results demonstrate significant balance differences in opposing enemy encounters and show how these can be regulated. Moreover, the analytic methods proposed are applied to identify the balance relationships between classes when fighting against each other, reflecting the original developers' design.

Tabletop roleplaying games (TTRPGs) and procedural content generators can both be understood as systems of rules for producing content. In this paper, we argue that TTRPG design can usefully be viewed as procedural content generator design. We present several case studies linking key concepts from PCG research – including possibility spaces, expressive range analysis, and generative pipelines – to key concepts in TTRPG design. We then discuss the implications of these relationships and suggest directions for future work uniting research in TTRPGs and PCG.

When evaluating the effectiveness of gamified app experiences in cultural heritage venues in terms of informal learning outcomes, a core challenge is the complexity involved in assessing intangible measures such as visitors’ appraisal of artwork. A comprehensive summary of the literature for conducting museum visitor evaluations is needed in order to understand how to measure the impact of gamification on user engagement, and the enhancement of the cultural heritage experience on learning. This paper first reviews related literature regarding the application of intrusive versus non-intrusive user evaluation methods, focusing on the REMIND protocol for conducting experiments with museum visitors. We relay our findings when applying the REMIND protocol in four gamified cultural heritage applications in the CrossCult project. Focusing on the assessment of informal learning in an application specifically designed for the visitors of the National Gallery of London, the paper concludes with recommendations, challenges, and future steps in evaluating games for cultural heritage.

As of 2020, the international workshop on Procedural Content Generation enters its second decade. The annual workshop, hosted by the international conference on the Foundations of Digital Games, has collected a corpus of 95 papers published in its first 10 years. This paper provides an overview of the workshop’s activities and surveys the prevalent research topics emerging over the years.

Learning to read and write is a fundamental right and a necessary skill for the personal, cultural, and economic development of people and their societies. However, children of developing countries, such as sub-Saharan areas, are currently at a greater risk of illiteracy. The current penetration of mobile technologies and the internet in sub-Saharan rural areas, however, offers a unique opportunity for tackling the challenge of literacy at a large scale. Motivated by the current shortage of preschool teachers for training handwriting in a personalised manner, this paper discusses the design of Djehuty, an educational gamified environment for preschoolers. Djehuty is equipped with an artificial intelligence module which generates a style of handwriting and suggests handwriting paths to the child in a mixed-initiative manner. The paper presents the key elements of the game prototype.

While artificial intelligence has been applied to control players’ decisions in board games for over half a century, little attention is given to games with no player competition. Pandemic is an exemplar collaborative board game where all players coordinate to overcome challenges posed by events occurring during the game’s progression. This paper proposes an artificial agent which controls all players’ actions and balances chances of winning versus risk of losing in this highly stochastic environment. The agent applies a Rolling Horizon Evolutionary Algorithm on an abstraction of the game-state that lowers the branching factor and simulates the game’s stochasticity. Results show that the proposed algorithm can find winning strategies more consistently in different games of varying difficulty. The impact of a number of state evaluation metrics is explored, balancing between optimistic strategies that favor winning and pessimistic strategies that guard against losing.

There is growing evidence suggesting that subjective values such as emotions are intrinsically relative and that an ordinal approach is beneficial to their annotation and analysis. Ordinal data processing yields more reliable, valid and general predictive models, and preference learning algorithms have shown a strong advantage in deriving computational models from such data. To enable the extensive use of ordinal data processing and preference learning, this paper introduces the Python Preference Learning Toolbox. The toolbox is open source, features popular preference learning algorithms and methods, and is designed to be accessible to a wide audience of researchers and practitioners. The toolbox is evaluated with regards to both the accuracy of its predictive models across two affective datasets and its usability via a user study. Our key findings suggest that the implemented algorithms yield accurate models of affect while its graphical user interface is suitable for both novice and experienced users.

Is it possible to predict the affect of a user just by observing her behavioral interaction through a video? How can we, for instance, predict a user's arousal in games by merely looking at the screen during play? In this paper we address these questions by employing three dissimilar deep convolutional neural network architectures in our attempt to learn the underlying mapping between video streams of gameplay and the player's arousal. We test the algorithms in an annotated dataset of 50 gameplay videos of a survival shooter game and evaluate the deep learned models' capacity to classify high vs low arousal levels. Our key findings with the demanding leave-one-video-out validation method reveal accuracies of over 78 % on average and 98% at best. While this study focuses on games and player experience as a test domain, the findings and methodology are directly relevant to any affective computing area, introducing a general and user-agnostic approach for modeling affect.

How could we gather affect annotations in a rapid, unobtrusive, and accessible fashion? How could we still make sure that these annotations are reliable enough for data-hungry affect modelling methods? This paper addresses these questions by introducing PAGAN, an accessible, general-purpose, online platform for crowdsourcing affect labels in videos. The design of PAGAN overcomes the accessibility limitations of existing annotation tools, which often require advanced technical skills or even the on-site involvement of the researcher. Such limitations often yield affective corpora that are restricted in size, scope and use, as the applicability of modern data-demanding machine learning methods is rather limited. The description of PAGAN is accompanied by an exploratory study which compares the reliability of three continuous annotation tools currently supported by the platform. Our key results reveal higher inter-rater agreement when annotation traces are processed in a relative manner and collected via unbounded labelling.

Quality-diversity (QD) algorithms search for a set of good solutions which cover a space as defined by behavior metrics. This simultaneous focus on quality and diversity with explicit metrics sets QD algorithms apart from standard single- and multi-objective evolutionary algorithms, as well as from diversity preservation approaches such as niching. These properties open up new avenues for artificial intelligence in games, in particular for procedural content generation. Creating multiple systematically varying solutions allows new approaches to creative human-AI interaction as well as adaptivity. In the last few years, a handful of applications of QD to procedural content generation and game playing have been proposed; we discuss these and propose challenges for future work.

This paper presents The Newborn World, a colocated multiplayer game for mobile devices with a focus on storytelling. The game has been designed following patterns of analog games that prompt players to tell stories, and uses digital representations of cards and tokens for interaction. The collaborative nature of storytelling is interspersed with competitive elements such as hidden goals and secret voting for best story. The paper exposes several design decisions for balancing goal-oriented and freeform play, competition and collaboration, as well as elements that constrain or foster players' creativity.

Which features of a game influence the dynamics of players interacting with it? Can a level's architecture change the balance between two competing players, or is it mainly determined by the character classes and roles that players choose before the game starts? This paper assesses how quantifiable gameplay outcomes such as score, duration and features of the heatmap can be predicted from different facets of the initial game state, specifically the architecture of the level and the character classes of the players. Experiments in this paper explore how different representations of a level and class parameters in a shooter game affect a deep learning model which attempts to predict gameplay outcomes in a large corpus of simulated matches. Findings in this paper indicate that a few features of the ruleset (i.e. character class parameters) are the main drivers for the model's accuracy in all tested gameplay outcomes, but the levels (especially when processed) can augment the model.

Is it possible to predict the motivation of players just by observing their gameplay data? Even if so, how should we measure motivation in the first place? To address the above questions, on the one end, we collect a large dataset of gameplay data from players of the popular game Tom Clancy's The Division. On the other end, we ask them to report their levels of competence, autonomy, relatedness and presence using the Ubisoft Perceived Experience Questionnaire. After processing the survey responses in an ordinal fashion we employ preference learning methods based on support vector machines to infer the mapping between gameplay and the reported four motivation factors. Our key findings suggest that gameplay features are strong predictors of player motivation as the best obtained models reach accuracies of near certainty, from 92% up to 94% on unseen players.

This paper presents a two-step generative approach for creating dungeons in the rogue-like puzzle game MiniDungeons 2. Generation is split into two steps, initially producing the architectural layout of the level as its walls and floor tiles, and then furnishing it with game objects representing the player's start and goal position, challenges and rewards. Three layout creators and three furnishers are introduced in this paper, which can be combined in different ways in the two-step generative process for producing diverse dungeons levels. Layout creators generate the floors and walls of a level, while furnishers populate it with monsters, traps, and treasures. We test the generated levels on several expressivity measures, and in simulations with procedural persona agents.

The feeling of horror within movies or games relies on the audience's perception of a tense atmosphere-often achieved through sound accompanied by the on-screen drama-guiding its emotional experience throughout the scene or game-play sequence. These progressions are often crafted through an a priori knowledge of how a scene or game-play sequence will playout, and the intended emotional patterns a game director wants to transmit. The appropriate design of sound becomes even more challenging once the scenery and the general context is autonomously generated by an algorithm. Towards realizing sound-based affective interaction in games this paper explores the creation of computational models capable of ranking short audio pieces based on crowdsourced annotations of tension, arousal and valence. Affect models are trained via preference learning on over a thousand annotations with the use of support vector machines, whose inputs are low-level features extracted from the audio assets of a comprehensive sound library. The models constructed in this work are able to predict the tension, arousal and valence elicited by sound, respectively, with an accuracy of approximately 65%, 66% and 72%.

This paper presents a framework for generating adventure games from open data. Focusing on the murder mystery type of adventure games, the generator is able to transform open data from Wikipedia articles, OpenStreetMap, and images from Wikimedia Commons into WikiMysteries. Every WikiMystery game revolves around the murder of a person with a Wikipedia article, and populates the game with suspects who must be arrested by the player if guilty of the murder or absolved if innocent. Starting from only one person as the victim, an extensive generative pipeline finds suspects, their alibis, and paths connecting them from open data, transforms open data into cities, buildings, nonplayer characters, locks and keys, and dialog options. This paper describes in detail each generative step, provides a specific playthrough of one WikiMystery where Albert Einstein is murdered, and evaluates the outcomes of games generated for the 100 most influential people of the 20th century.

This paper describes a method for generative player modeling and its application to the automatic testing of game content using archetypal player models called procedural personas. Theoretically grounded in psychological decision theory, procedural personas are implemented using a variation of Monte Carlo tree search (MCTS) where the node selection criteria are developed using evolutionary computation, replacing the standard UCB1 criterion of MCTS. Using these personas, we demonstrate how generative player models can be applied to a varied corpus of game levels and demonstrate how different playstyles can be enacted in each level. In short, we use artificially intelligent personas to construct synthetic playtesters. The proposed approach could be used as a tool for automatic play testing when human feedback is not readily available or when quick visualization of potential interactions is necessary. Possible applications include interactive tools during game development or procedural content generation systems where many evaluations must be conducted within a short time span.

The papers in this special section focus on game design based on artificial intelligence (AI). As researchers working with games, we are often faced with terms that either lack a definition entirely or which are contested and debated as part of their very nature. Many papers have tried and failed to define what “fun” is, from where “creativity” originates, or what “difficulty” means, for example, and many more papers in the future will try and fail to do the same. Tangling with ineffable concepts, with moving targets and nebulous ideas is all part of the joy of doing artificial intelligence (AI) research in a complex, multifaceted domain like games.

—The design process is often characterized by and realized through the iterative steps of evaluation and refinement. When the process is based on a single creative domain such as visual art or audio production, designers primarily take inspiration from work within their domain and refine it based on their own intuitions or feedback from an audience of experts from within the same domain. What happens, however, when the creative process involves more than one creative domain such as in a digital game? How should the different domains influence each other so that the final outcome achieves a harmonized and fruitful communication across domains? How can a computational process orchestrate the various computational creators of the corresponding domains so that the final game has the desired functional and aesthetic characteristics? To address these questions, this paper identifies game facet orchestration as the central challenge for artificial-intelligence-based game generation, discusses its dimensions, and reviews research in automated game generation that has aimed to tackle it. In particular, we identify the different creative facets of games, propose how orchestration can be facilitated in a top-down or bottom-up fashion, review indicative preliminary examples of orchestration, and conclude by discussing the open questions and challenges ahead.

Quality diversity (QD) is a recent family of evolutionary search algorithms which focus on finding several well-performing (quality) yet different (diversity) solutions with the aim to maintain an appropriate balance between divergence and convergence during search. While QD has already delivered promising results in complex problems, the capacity of divergent search variants for QD remains largely unexplored. Inspired by the notion of surprise as an effective driver of divergent search and its orthogonal nature to novelty this paper investigates the impact of the former to QD performance. For that purpose we introduce three new QD algorithms which employ surprise as a diversity measure, either on its own or combined with novelty, and compare their performance against novelty search with local competition, the state of the art QD algorithm. The algorithms are tested in a robot navigation task across 60 highly deceptive mazes. Our findings suggest that allowing surprise and novelty to operate synergistically for divergence and in combination with local competition leads to QD algorithms of significantly higher efficiency, speed, and robustness.

This paper introduces DATA Agent, a system which creates murder mystery adventures from open data. In the game, the player takes on the role of a detective tasked with finding the culprit of a murder. All characters, places, and items in DATA Agent games are generated using open data as source content. The paper discusses the general game design and user interface of DATA Agent, and provides details on the generative algorithms which transform linked data into different game objects. Findings from a user study with 30 participants playing through two games of DATA Agent show that the game is easy and fun to play, and that the mysteries it generates are straightforward to solve.

This paper introduces a surrogate model of gameplay that learns the mapping between different game facets, and applies it to a generative system which designs new content in one of these facets. Focusing on the shooter game genre, the paper explores how deep learning can help build a model which combines the game level structure and the game's character class parameters as input and the gameplay outcomes as output. The model is trained on a large corpus of game data from simulations with artificial agents in random sets of levels and class parameters. The model is then used to generate classes for specific levels and for a desired game outcome, such as balanced matches of short duration. Findings in this paper show that the system can be expressive and can generate classes for both computer generated and human authored levels.

Traditional evolutionary algorithms tend to converge to a single good solution, which can limit their chance of discovering more diverse and creative outcomes. Divergent search, on the other hand, aims to counter convergence to local optima by avoiding selection pressure towards the objective. Forms of divergent search such as novelty or surprise search have proven to be beneficial for both the efficiency and the variety of the solutions obtained in deceptive tasks. Importantly for this paper, early results in maze navigation have shown that combining novelty and surprise search yields an even more effective search strategy due to their orthogonal nature. Motivated by the largely unexplored potential of coupling novelty and surprise as a search strategy, in this paper we investigate how fusing the two can affect the evolution of soft robot morphologies. We test the capacity of the combined search strategy against objective, novelty, and surprise search, by comparing their efficiency and robustness, and the variety of robots they evolve. Our key results demonstrate that novelty-surprise search is generally more efficient and robust across eight different resolutions. Further, surprise search explores the space of robot morphologies more broadly than any other algorithm examined.

While the primary focus of affective computing has been on constructing efficient and reliable models of affect, the vast majority of such models are limited to a specific task and domain. This paper, instead, investigates how computational models of affect can be general across dissimilar tasks; in particular, in modeling the experience of playing very different video games. We use three dissimilar games whose players annotated their arousal levels on video recordings of their own playthroughs. We construct models mapping ranks of arousal to skin conductance and gameplay logs via preference learning and we use a form of cross-game validation to test the generality of the obtained models on unseen games. Our initial results comparing between absolute and relative measures of the arousal annotation values indicate that we can obtain more general models of player affect if we process the model output in an ordinal fashion.

How should annotation data be processed so that it can best characterize the ground truth of affect? This paper attempts to address this critical question by testing various methods of processing annotation data on their ability to capture phasic elements of skin conductance. Towards this goal the paper introduces a new affect annotation tool, RankTrace, that allows for the annotation of affect in a continuous yet unbounded fashion. RankTrace is tested on first-person annotations of tension elicited from a horror video game. The key findings of the paper suggest that the relative processing of traces via their mean gradient yields the best and most robust predictors of phasic manifestations of skin conductance.

A particular challenge of the game design process is when the designer is requested to orchestrate dissimilar elements of games such as visuals, audio, narrative and rules to achieve a specific play experience. Within the domain of adversarial first person shooter games, for instance, a designer must be able to comprehend the differences between the weapons available in the game, and appropriately craft a game level to take advantage of strengths and weaknesses of those weapons. As an initial study towards computationally orchestrating dissimilar content generators in games, this paper presents a computational model which can classify a matchup of a team-based shooter game as balanced or as favoring one or the other team. The computational model uses convolutional neural networks to learn how game balance is affected by the level, represented as an image, and each team's weapon parameters. The model was trained on a corpus of over 50,000 simulated games with artificial agents on a diverse set of levels created by 39 different generators. The results show that the fusion of levels, when processed by a convolutional neural network, and weapon parameters yields an accuracy far above the baseline but also improves accuracy compared to artificial neural networks or models which use partial information, such as only the weapon or only the level as input.

Divergent search techniques applied to evolutionary computation, such as novelty search and surprise search, have demonstrated their efficacy in highly deceptive problems compared to traditional objective-based fitness evolutionary processes. While novelty search rewards unseen solutions, surprise search rewards unexpected solutions. As a result these two algorithms perform a different form of search since an expected solution can be novel while an already seen solution can be surprising. As novelty and surprise search have already shown much promise individually, the hypothesis is that an evolutionary process that rewards both novel and surprising solutions will be able to handle deception in a better fashion and lead to more successful solutions faster. In this paper we introduce an algorithm that realises both novelty and surprise search and we compare it against the two algorithms that compose it in a number of robot navigation tasks. The key findings of this paper suggest that coupling novelty and surprise is advantageous compared to each search approach on its own. The introduced algorithm breaks new ground in divergent search as it outperforms both novelty and surprise in terms of efficiency and robustness, and it explores the behavioural space more extensively.

This paper presents a generative technique for game levels, focusing on expansive dungeon levels. The proposed two-step evolutionary process creates a high-level overview of the map, which is then used to specify constraints and objectives on multiple constrained optimization algorithms which generate the high-resolution segments of the map. Results show how different types of segments are possible, and how the different connectivity constraints and objectives affect the performance of the algorithm. The modular approach, which allows for a high-level specification of the level first and the subsequent compartmentalized generation of the final map's components, is both scalable and more computationally efficient than a direct encoding, while it allows for more control and user intervention on either level of detail.

Grounded in the divergent search paradigm and inspired by the principle of surprise for unconventional discovery in computational creativity, this paper introduces surprise search as a new method of evolutionary divergent search. Surprise search is tested in two robot navigation tasks and compared against objective-based evolutionary search and novelty search. The key findings of this paper reveal that surprise search is advantageous compared to the other two search processes. It outperforms objective search and it is as efficient as novelty search in both tasks examined. Most importantly, surprise search is, on average, faster and more robust in solving the navigation problem compared to objective and novelty search. Our analysis reveals that surprise search explores the behavioral space more extensively and yields higher population diversity compared to novelty search.

Novelty search is a recent algorithm geared toward exploring search spaces without regard to objectives. When the presence of constraints divides a search space into feasible space and infeasible space, interesting implications arise regarding how novelty search explores such spaces. This paper elaborates on the problem of constrained novelty search and proposes two novelty search algorithms which search within both the feasible and the infeasible space. Inspired by the FI-2pop genetic algorithm, both algorithms maintain and evolve two separate populations, one with feasible and one with infeasible individuals, while each population can use its own selection method. The proposed algorithms are applied to the problem of generating diverse but playable game levels, which is representative of the larger problem of procedural game content generation. Results show that the two-population constrained novelty search methods can create, under certain conditions, larger and more diverse sets of feasible game levels than current methods of novelty search, whether constrained or unconstrained. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. Additionally, the proposed enhancement of offspring boosting is shown to enhance performance in all cases of two-population novelty search.

Novelty search is a recent algorithm geared to explore search spaces without regard to objectives; minimal criteria novelty search is a variant of this algorithm for constrained search spaces. For large search spaces with multiple constraints, however, it is hard to find a set of feasible individuals that is both large and diverse. In this paper, we present two new methods of novelty search for constrained spaces, Feasible-Infeasible Novelty Search and Feasible-Infeasible Dual Novelty Search. Both algorithms keep separate populations of feasible and infeasible individuals, inspired by the FI-2pop genetic algorithm. These algorithms are applied to the problem of creating diverse and feasible game levels, representative of a large class of important problems in procedural content generation for games. Results show that the new algorithms under certain conditions can produce larger and more diverse sets of feasible strategy game maps than existing algorithms. However, the best algorithm is contingent on the particularities of the search space and the genetic operators used. It is also shown that the proposed enhancement of offspring boosting increases performance in all cases.

This paper introduces a search-based approach to personalized content generation with respect to visual aesthetics. The approach is based on a two-step adaptation procedure where: 1) the evaluation function that characterizes the content is adjusted to match the visual aesthetics of users; and 2) the content itself is optimized based on the personalized evaluation function. To test the efficacy of the approach, we design fitness functions based on universal properties of visual perception, inspired by psychological and neurobiological research. Using these visual properties, we generate aesthetically pleasing 2-D game spaceships via neuroevolutionary constrained optimization and evaluate the impact of the designed visual properties on the generated spaceships. The offline generated spaceships are used as the initial population of an interactive evolution experiment in which players are asked to choose spaceships according to their visual taste: the impact of the various visual properties is adjusted based on player preferences and new content is generated online based on the updated computational model of visual aesthetics of the player. Results are presented that show the potential of the approach in generating content which is based on subjective criteria of visual aesthetics.