AI-assisted decision-making systems are increasingly used across domains, prompting research into design factors that influence their effectiveness. Prior work has examined elements such as explanation type, interface design, and cognitive support interventions; however, the complexity of real-world tasks makes the generalization of findings difficult. In particular, in real-world decision-making settings like Fake News Detection studies on human-AI collaboration remain underexplored. This study investigates how explanation type and system interactivity affect human-AI collaboration in the context of fake news detection. In a controlled online experiment (N = 161), we varied the explanation type and the interactivity of an AI system that assists with the fake news detection task. Results show that conversational systems increased user reliance, while feature importance explanations appeared to be more effective than counterfactual explanations. These findings highlight the importance of explanation design and interaction modality in shaping effective AI-assisted decision-making in real world settings where underlying regularities are difficult for non-expert users to detect.

Until recently the game industry heavily relied on manual playtesting to test the games it produces. Even if the benefits of introducing automated testing are acknowledged, it is rarely done in practice. Some of the main hurdles include the lack of automated testing tools that can target computer games as well as the complexity of automated game plays which are much more difficult to program than typical simple test sequences. This article presents an agent-based testing framework called aplib that comes with a Domain Specific Language (DSL) that allows complex playtests to be programmed more abstractly. A so-called goal structure is used to abstractly formulate a playtest scenario in terms of main goals and their decomposition into subgoals. Scenarios that are not too complicated can be formulated using static goal structures. More complex scenarios may need a test agent that can dynamically adapt its play according to the situation that evolves during the play. To handle such cases, aplib allows dynamic goals to be expressed as well. Invariants and pre-/post-conditions are used to assert the properties that a play is expected to satisfy. They include differential properties that allow constraints on the current state to be related to that of past states. Three case studies are included in the article. The first one aims to evaluate the performance of playtests programmed with aplib. The second shows that the approach can also be combined with other automated testing approaches, in this case reinforcement learning. The third shows the applicability of such playtests in a 3D setup and for non-functional testing.

With the rapid advancement of robotics, robots' helping behaviors are increasingly framed not only as functional assistance but also as prosocially meaningful interaction. In this context, the resource cost borne by the help-provider is a critical factor, yet it has not been systematically explored in existing human-robot interaction (HRI) research. Understanding how humans perceive and respond to different types of helping is essential for building better human-robot relationships. This study addresses this gap through two experiments. Study 1 examined the role of agent resource type (robot's own resources vs. external resources). Results showed that when robots shared their own resources, participants did not report significant differences in overall attitudes or prosocial behavior, but attributed higher performance trust and expressed stronger feelings of gratitude and guilt. Study 2 further examined the importance of agent resources (robot battery level: high vs. low). The results showed that even when relative costs were the same, participants tended to perceive sharing from a low-battery robot as more reliable, while variations in resource type or importance did not significantly change social responses. These findings suggest that human evaluations of robots are shaped not only by the outcomes of helping but also by the perceived cost and sacrifice underlying robot actions. Our work offers an initial direction for integrating resource cost considerations into the design of social robots.

This article conducts a systematic review on Procedural Content Generation (PCG) for Cooperative Games. Our main goal was exploring the state of the art in this area of research, and, in particular, its application to Level Generation for Cooperative Games. Another objective was determining how generating content for Cooperative Games could be different from generating content for other types of games. Our results showed a lack of research in this particular application of PCG. Despite this, we found a variety of use cases of PCG for Cooperative games, most of them related to autonomous generation and co-creative and mixed-initiative design. Additionally, we found a variety of methods and techniques being used to generate levels for Cooperative Games that covered the broad categories we used to condense this information. Concerning the differences between generating content for Cooperative Games and generating content for other types of games, the studies we found suggested that PCG for Cooperative Games should take into account the Design Specificities of Cooperative Games - the types of Tasks, Cooperative Game Design Patterns and Game Mechanics used to incentivize and support cooperative gameplay.

This demonstration presents the Collaborative AI Arena, a web-based platform for large-scale evaluation of Human-AI collaboration in the wild. It was built to support research on Human-IA collaboration addressing both the challenges of performing joint actions and discussing the task to reach common ground. The platform currently includes four different tasks that can be tested with sample AI agents built using foundational AI models.

This paper presents a web application designed to study the impact of explainable AI (XAI) and XAI interactivity on fake news detection. The system enables a controlled user study with a 2x2 design, varying XAI interactivity (high vs. low) and explanation type (counterfactual vs. feature importance). High interactivity conditions utilize an LLM-based chatbot for dynamic explanations, while low interactivity conditions provide fixed, predetermined explanations. The demonstration highlights how AI explanations influence user trust, reliance, and learning in news verification tasks. The web application's video show-casing the main features is available at: https://drive.google.com/file/d/ 1_U5wUoHMrUEHFSXrna90LpWjnUbv9eJn/view?usp=sharing

Access to public spaces is of the utmost importance for social cohesion, inclusion, and civic engagement. Nevertheless, a large majority of public spaces remain incredibly uncomfortable environments for neurodivergent individuals due to, for instance, the unpredictability of such spaces and the sensory stimuli within them. Smart City technologies present an exciting opportunity to improve the accessibility and enjoyment of the spaces where they are deployed by, for instance, offering users the ability to customise a space to their specific sensory needs. However, the research topic of public space technologies for neurodivergent individuals remains scattered and sparsely documented. This critical review analyses the existing domains of inquiry, contributing a theoretical framework based on Spatial Justice and Neuroqueer Technoscience and suggests future research avenues informed by this framework. We advocate for the participatory co-creation of a neurodivergent-affirming landscape of public space technologies that both support neurodivergent needs and promote neurodivergent joy.

User Experience (UX) has a great impact in today’s market, making good UX evaluations crucial for the development process. However, UX evaluations pose many challenges and significant costs. A possible solution to reduce costs and empower conducting UX evaluations is to automate a significant part of the process. Automated UX evaluations offer a solution around some important issues, such as the diversity of the audience that tests the system and the possibility of performing continuous testing of different versions. It also opens new possibilities, such as simulating long-term or multiparty interactions. We argue that to move to automated testing in a sound and comprehensive way, we need good definitions and good UX models. In this article, we present a comprehensive framework to assist the definition of UX tests as a first step towards automated UX testing. We believe this framework will help the formalization and systematization of the automated tests, and its modular structure makes it comprehensive enough to cover different aspects of UX and types of systems. We discuss previous UX definitions and models that inspired our framework, and we also discuss concepts from functional testing, such as coverage, applied to UX testing.

This paper explores the application of Procedural Content Generation (PCG) in the creation of sprite sheets for humanoid character art and animation. PCG offers the possibility of automating the generation of diverse characters and dynamic animations, which is a critical issue in games that use PCG to generate diverse Non-Player Characters (NPC) and enemies. Our approach proposes the use of cellular automata algorithms to create the components of sprites needed for the representation of the character in the game and the core movement animations. The advantages of PCG include significant time savings, enhanced variation and diversity, and scalability for game development. However, challenges related to maintaining artistic control and ensuring quality assurance must be addressed. The paper emphasizes the potential of PCG for humanoid character design in 2D game development, unlocking new creative possibilities supported in the generation of many characters in the game.

State of the art agent-modelling tools support the creation of powerful Socially Intelligent Agents (SIAs) capable of engaging in social interactions with participants in various roles and environments. However, their deployment demands a laborious authoring task as it is necessary to manually define behaviour rules and create content for different interaction scenarios. While Socially Intelligent Agents (SIAs) research has centered on the user experience, we shift focus to the authors. To understand the challenges faced by authors who create these agents, we performed an innovative analysis of the authoring experience in modern agent modelling tools. One key finding is that, while SIA concepts are generally understandable, emotional-based concepts are not as easily comprehended or used by authors. We propose a hybrid solution approach that culminated in the development of Authoring Assisted FAtiMA-Toolkit. The augmented agent modeling tool incorporates a data-driven authoring assistant to boost author productivity while promoting transparency and authorial control. To evaluate the impact of this framework on the authoring experience, we conducted a user study. Results showed that authors using the Authoring-Assisted FAtiMA-Toolkit were on average able to create more SIA-related content in less time. Our findings suggest that data-augmented, theory-grounded agent modeling tools can support the development of affective social agents by reducing the authoring burden without sacrificing the framework’s clarity or the authors’ control over the content

Dynamics in Human-AI interaction should lead to more satisfying and engaging collaboration. Key open questions are how to design such interactions and the role personal goals and expectations play. We developed three AI partners of varying initiative (leader, follower, shifting) in a collaborative game called Geometry Friends. We conducted a within-subjects experiment with 60 participants to assess personal AI partner preference and performance satisfaction as well as perceived warmth and competence of AI partners. Results show that AI partners following human initiative are perceived as warmer and more collaborative. However, some participants preferred AI leaders for their independence and speed, despite being seen as less friendly. This suggests that assigning a leadership role to the AI partner may be suitable for time-sensitive scenarios. We identify design factors for developing collaborative AI agents with varying levels of initiative to create more effective human-AI teams that consider context and individual preference.

Neurodivergent children spend most of their time in neurodiverse schools alongside their neurotypical peers and often face social exclusion. Inclusive play activities are a strong vehicle of inclusion. Unfortunately, games designed for the specific needs of neurodiverse groups are scarce. Given the potential of robots to support play, we led a co-design process to build an inclusive robotic game for neurodiverse classrooms. We conducted five co-design workshops, engaging 80 children from neurodiverse classrooms in designing an inclusive game. Employing the resulting design insights, we iteratively prototyped and playtested a tabletop robotic game leveraging of-the-shelf robots. Reflecting upon our findings, we discuss how the longitudinal co-design process (rather than the resulting game) was key in allowing children the space to learn how to accommodate accessibility needs and create inclusive play experiences. We posit the use of co-design to enhance children's interpersonal relationships, fosters feelings of ownership, and encourages appropriation practices as a strategy to sustain inclusive experiences that extend beyond project timelines or artefact designs.

This Special Interest Group (SIG) explores the transformative impact of Generative Artificial Intelligence (GenAI) on Human-Computer Interaction (HCI) research processes. The theme here is to answer “question zero”: when to use and when to refrain from using AI tools during the research cycle? The discussion is guided by five research phases commonly used in HCI: research planning, prototyping, data collection, analysis and synthesis, and dissemination and communication. We investigate how GenAI accelerates project cycles, enhances reproducibility, and influences inclusivity in research. We also address the challenging ethical considerations about the ownership of generated content. Our goal is to build a community of HCI enthusiasts to harness the early advantages of the recent groundbreaking technology and foresee challenges arising from its prevalence in the scientific community.

Testing User eXperience (UX) during the development of complex systems, such as games, often requires costly and time-consuming human involvement. To address this issue, we propose the use of modular UX testing agents capable of providing quick and on-demand UX estimates. These agents empower developers to efficiently assess UX design goals and understand user actions and preferences. We present two data-driven modules for creating a flexible UX testing agent: one for emotion prediction and another for generating persona-like behaviour. The agent enables the evaluation of various player types and their emotional responses within a game environment.

Human-agent teamwork is a promising research stream with great potential to impact society. Research on collaborative AI and human-agent interaction has tackled the problem from several perspectives, but we argue that a focus on teams as a unit and a model for human-agent team dynamics is missing. Such a focus is particularly relevant if we aim at involving agents as active team members and at building sustainable teams over time. A team perspective on human-agent collaboration requires new models that pose challenges for AI and humans alike. AI needs new models to build an understanding of team level variables, such as team structure and cohesion, to be able to monitor the team and act on the team beyond performing the task. Humans, in turn, need to be able to incorporate agents as team members in their mental models of teamwork and integrate them into team processes. Such human-agent team dynamics models should be built taking into account four different levels: individual, interpersonal, team, and organisational. We believe that to fulfill this vision we need to bring together the different fields of AI and social sciences.

This paper proposes a novel pipeline for generating game levels that elicit predefined emotional experiences from players. Our approach uses evolutionary algorithms alongside data-driven persona agents, predictive emotional models, a PCG parametric level generator, and a newly defined language for the clear and computable definition of player emotional experiences: ExpREx (Experience Regular Expressions). Using these components, we evolve game levels to match the player experience goals specified using the ExpREx language, aiming to create levels that evoke specific emotional experiences for different subsets of players. The efficacy of our method was validated through a user study involving 101 participants, whose continuous annotations of emotional experience were collected and analyzed to assess the congruence between the actual emotional responses elicited and those targeted by our pipeline. We found that 93.73% of the ExpREx goals targeted were also reported by the user study subjects.

Procedural content generation (PCG) is a popular topic in game research and practice, however, the generation of content for cooperative games, specifically content that requires collaboration between both players to be completed, is still underdeveloped. In this work, we contribute to the body of knowledge of PCG for cooperative games, describing our approach for generating levels, for the cooperative game Geometry Friends, based on genetic algorithms and the definition of cooperative constraints. We present the evaluation conducted to test the quality of a sample of levels generated and the appropriateness of the constraints that define areas of reach for each player in the game. The evaluation showed that the constraints given to the generation algorithm are able to express different levels of cooperation in the levels generated, according to the subjective assessment of players.

Although Science, Technology, Engineering, and Mathematics (STEM) are essential for the development of society, men hugely outnumber women in the majority of STEM fields in higher education, a factor that hinders inclusion and restricts the possibility of having different points-of-view. Previous studies indicate multiple causes of low female motivation in STEM degrees and careers, which inspired several initiatives to increase female interest in STEM. A proven way to captivate an audience to change its attitude is the heroic narrative model, a style of narrative in which a character goes through a sequence of difficulty-increasing and attitude-shaping quests. This paper proposes a heroic narrative model named Heroine’s Learning Journey (HLJ) targeted at counteracting low female participation in STEM courses. In particular, the HLJ model is developed especially for enhancing STEM online courses, by using a narrative that can encourage female students to engage and prevail in them. The HLJ model is divided into three acts, each composed of several stages symbolizing steps tailored to a female student’s development. The model was applied to set up the structure of a preexisting Machine Learning online course with hundreds of enrolled students. Although a first version of the course already presented a higher-than-expected female enrollment per se ( $\approx 37.3\%$ ), with HLJ, we verified an even higher female enrollment ( $\approx 59.2\%$ ), slightly surpassing male enrollments. The feedback provided in learners’ responses to a final, voluntary and anonymous questionnaire, allowed to obtain the degree of satisfaction of participants at the end of the course with the HLJ. The responses indicated that, at the end of the second edition of the online course, students were able to acknowledge the existence of a STEM gender imbalance, and appreciated the motivating nature of the HLJ model. From several student’s feedback and comments submitted in the questionnaire, one can conclude that the attitude-shaping character of the HLJ was greatly appreciated, in addition to the technical content of the course. All these preliminary results are indicative of the usability of HLJ to foster gender balance in STEM online courses. Thus, the present study contributes to STEM Education by leveraging the motivation of young women to enter and prevail in these areas of study.

Many neurodivergent (ND) children are integrated into mainstream schools alongside their neurotypical (NT) peers. However, they often face social exclusion, which may have lifelong effects. Inclusive play activities can be a strong driver of inclusion. Unfortunately, games designed for the specific needs of neurodiverse groups, those that include neurodivergent and neurotypical individuals, are scarce. Given the potential of robots as engaging devices, we led a 6-month co-design process to build an inclusive and entertaining robotic game for neurodiverse classrooms. We first interviewed neurodivergent adults and educators to identify the barriers and facilitators for including neurodivergent children in mainstream classrooms. Then, we conducted five co-design sessions, engaging four neurodiverse classrooms with 81 children (19 neurodivergent). We present a reflection on our co-design process and the resulting robotic game through the lens of Self-Determination Theory, discussing how our methodology supported the intrinsic motivations of neurodivergent children.

Workplace wellness programs constitute a preventive measure to help avoid healthcare costs for companies, with additional benefits for employee productivity and other organizational outcomes. Interventions using social robots may have some advantages over other conventional telemedicine applications, since they can deliver personalized feedback and counseling. This investigation focused on a health-promoting intervention within work environments, and compared the efficacy of the intervention on two distinct groups, one guided by a human agent and the other by a robot agent. Participants (n = 56) were recruited from two Portuguese organizations and led through eight sessions by the social agent, the goal being to encourage health behavior change and adoption of a healthier lifestyle. The results indicate that the group led by the robot agent revealed better post-intervention scores than the group led by the human agent, specifically with regard to productivity despite presenteeism and regard of their level of mental well-being. No effects were found concerning the work engagement level of participants in either group. By demonstrating the potential of using social robots to establish therapeutic and worth relationships with employees in their workplaces, this study provides interesting new findings that contribute to the literature on health behavior change and human-robot interaction.