Developing the ability to think critically about AI and interpret its outputs requires an understanding of AI bias, a key skill for both AI users and future developers. While some initiatives have introduced teens to algorithmic bias, few have engaged them in actively identifying and quantifying bias in real-world generative AI systems. This paper presents BiasViz, an interactive tool that leverages project-based and narrative-centered learning to help middle school students (11-14 year old) analyze AI bias in large language models. We conducted a study of 28 students’ interactions with BiasViz to evaluate its efficacy in fostering critical thinking about AI bias. Our findings suggest that BiasViz successfully introduced most students to AI bias, and some used the tool to explore personally relevant biases. We identify opportunities for the tool’s iteration and associated curriculum to promote learning and share insights for designing learning environments that foster youth’s critical thinking about AI.

Museums play a critical role in promoting public understanding of emerging technologies like artificial intelligence (AI), but it is unclear what design features lead to learning about AI in museums. We contribute a design research exploration of how embodiment and creativity foster AI literacy in museum exhibits. We present design prototypes of three museum exhibits—DataBites, Knowledge Net, and LuminAIx—that aim to teach middle schoolers about AI. We present results from a qualitative analysis of an in-museum study in which we examined participants’ understanding of and interest in AI through interviews and video recordings. Our findings illuminate how creativity fosters interest in AI and how different forms of embodiment contribute to learning about AI. We recommend that AI museum exhibits utilize creative and personally relevant activities to engage middle schoolers, support hybrid conceptualizations of AI, and leverage tangible interaction to make AI concepts approachable.

The Model AI Assignments session seeks to gather and disseminate the best assignment designs of the Artificial Intelligence (AI) Education community. Recognizing that assignments form the core of student learning experience, we here present abstracts of thirteen AI assignments from the 2025 session that are easily adoptable, playfully engaging, and flexible for a variety of instructor needs. Assignment specifications and supporting resources may be found at http://modelai.gettysburg.edu

Literary approaches to design fictions, though previously theorized to be diverse in form and content, often fall within narrow stylistic and content boundaries such as speculative abstracts, memos, and studies. By drawing on a rich history of science fiction criticism, we advocate for literary design fictions that diverge from what is commonplace in HCI and design research. We foreground our paper with a discussion of the poetics of science fiction, and their relationship to current design fiction practices. Specifically, we highlight how the poetics of a design fiction can work to familiarize or defamiliarize readers from the imagined world presented. We thus argue that considerations of poetics, specifically how they work to (de)familiarize readers of design fictions, enrich understanding of design fictions as a research method. We then provide and discuss three design fictions in the forms of poetry and flash fiction, which fictionalize anthropomorphism in AI and AI explainability, AI assistants and AI privacy, and the relationship between AI and human autonomy. This paper makes two contributions: 1) a poetics-based framework that broadens current understandings of written design fictions and 2) three design fictions that speculate on the future of human-AI interaction.

As artificial intelligence (AI) increasingly permeates everyday life, there is a growing need for public understanding of AI’s underlying principles. Existing educational interventions and explainable AI (XAI) tools cater mainly to children or adult experts. In this paper, we present three interactive web-based tools to foster AI learning among adults without technical backgrounds. Designed according to learning sciences and user-centered design principles, these tools simplify complex AI concepts like edge detection, confidence thresholds, and sensitivity, making AI more understandable for beginners and facilitating reflection on ethical issues. We present results from a mixed-methods evaluation of the tools with 42 participants. Results show heightened familiarity and confidence in AI concepts. Our qualitative analysis additionally reveals common interaction patterns amongst participants. This paper offers both a design contribution to the AI education and XAI communities and emergent interaction patterns to support the design of transparent and learner-centered AI for adult novices.

It is essential to increase children’s understanding of artificial intelligence and machine learning as they encounter it through their daily activities. We have developed DataBites, a museum exhibit aimed at fostering middle-school-age children’s understanding of supervised machine learning. DataBites engages visitors in learning about the steps and practices of supervised machine learning, using three guiding design principles: embodied interaction, creativity, and collaboration. Our design allows learners to use tangible pieces to collaboratively create their own labeled examples of pizzas and sandwiches to include in a training dataset for an image-based machine-learning pizza/sandwich classification algorithm. The algorithm can classify sandwiches and pizzas by learning patterns from people’s examples. Learners can view the results and self-evaluate how well their dataset did at enabling the algorithm to distinguish between the two items. This poster paper contributes a novel design and approach to engaging children in learning about AI in museum settings.

As young people increasingly use AI in their daily lives, it is imperative to foster these learners’ AI literacy. We present Knowledge Net, a collaborative tangible tabletop museum exhibit aimed at teaching users about knowledge representations, which are central to understanding AI and understudied in AI education research. In this exhibit, we center creative making and embodied interaction by allowing learners to craft the appearance, behaviors, and traits of characters in a virtual world by manipulating semantic networks. Our poster features the exhibit design and corresponding rationale, and this paper contributes an exploration of how creative making and embodied interaction can be utilized to teach young learners about knowledge representations–and AI more broadly–in informal learning environments.

As generative AI rapidly enters everyday life, educational interventions for teaching about AI need to cater to how young people, in particular middle schoolers who are at a critical age for reasoning skills and identity formation, conceptualize and interact with AI. We conducted nine focus groups with 24 middle school students to elicit their interests, conceptions of, and approaches to a popular generative AI tool, ChatGPT. We highlight a) personally and culturally-relevant topics to this population, b) three distinct approaches in students’ open-ended interactions with ChatGPT: AI testing-oriented, AI socializing-oriented, and content exploring-oriented, and 3) an improved understanding of youths’ conceptions and misconceptions of generative AI. While misconceptions highlight gaps in understanding what generative AI is and how it works, most learners show interest in learning about what AI is and what it can do. We discuss the implications of these conceptions for designing AI literacy interventions in museums.

Fostering learners’ interest remains an important challenge in computer science (CS) education. In this paper, we explore how creative music-making, tangible interfaces, and embodiment can be used toward this end. The primary contribution of this paper is Xylocode, a novel exhibit that introduces middle school age learners to computing concepts and fosters interest in CS via a tangible playspace for making music using an embodied simulation. We additionally present an in-museum evaluation of Xylocode with 29 middle school age children. Our results indicate that the exhibit fosters situational interest in computer science and leads to recognition of certain computing concepts, including arrays and global variables. Future research is needed to assess whether the exhibit leads to longer-term learning and/or interest gains and to explore why other computing concepts were not recognized by as many learners. We identify several implications and directions for future work based on our findings.

Co-creation in embodied contexts is central to the human experience but is often lacking in our interactions with computers. We seek to develop a better understanding of embodied human co-creativity to inform the human-centered design of machines that can co-create with us. In this paper, we ask: What characterizes dancers’ experiences of embodied dyadic interaction in movement improvisation? To answer this, we ran focus groups with 24 university dance students and conducted a thematic analysis of their responses. We synthesize our findings in an Interconnected Model of Improvisational Dance Inputs, where movement choices are shaped by the interplay between in-the-moment influences between the self, partner, and the environment, a set of generative strategies, and heuristics for a successful collaboration. We present a set of design recommendations for LuminAI, a co-creative AI dance partner. Our contributions can inform the design of AI in embodied co-creative domains.

Fostering young learners’ literacy surrounding AI technologies is becoming increasingly important as AI is becoming integrated in many aspects of our lives and is having far-reaching impacts on society. We have developed Knowledge Net and Creature Features, two activity boxes for family groups to engage with in their homes that communicate AI literacy competencies such as understanding knowledge representations, the steps of machine learning, and AI ethics. Our current work is exploring how to transform these activity boxes into museum exhibits for middle-school age learners, focusing on three key considerations: centering learner interests, generating personally meaningful outputs, and incorporating embodiment and collaboration on a larger scale. Our demonstration will feature the existing Knowledge Net and Creature Features activity boxes alongside early-stage prototypes adapting these activities into larger-scale museum exhibits. This paper contributes an exploration into how to design AI literacy learning interventions for varied informal learning contexts.

What generative AI futures do we want—and what futures do we not want? To imagine what might exist in the future, we apply speculative design to explore plausible scenarios for generative AI and human coexistence. In this paper, we present gAIrden and Onion AI: two in-progress speculative concepts of future generative AI tools, their use cases, and the systems in which they exist. We analyze the designs through lenses of Environment, Data Privacy, Embodiment, and Play. This trip into the future is driven by the research question: how might generative AI tools change how we produce creativity and culture? When we return to the present, we ask ourselves, how might generative AI support positive outcomes for individuals and communities? Can we predict (and potentially mitigate) negative consequences of generative AI tools? The speculative designs purposefully engage viewers in futures thinking to reclaim conversation around the future of technology.

The unique role that AI plays in making decisions that affect human lives creates a need to foster improved public understanding of AI systems. Informal learning spaces are particularly important contexts for fostering AI literacy, as they have the potential to reach a broader audience and provide spaces for children and parents to learn together. This paper explores 1) what types of dialogue family groups engage in when learning about AI in an at-home learning environment in order to inform our understanding of 2) how to design AI literacy activities for informal learning contexts. We present an analysis of family group dialogue surrounding three different AI education activities and use our findings to reflect on, update, and add to existing principles for designing AI literacy educational interventions.

This article presents Active Prolonged Engagement eXpanded (APEX), a framework and toolkit for informing evidence-based decisions about the iterative design of embodied, collaborative museum exhibits. We provide an overview of APEX, a framework that builds on both prior work and experimentally derived data to provide an understanding of how visitors' physical, social, emotional, and intellectual engagement transform during the course of their interaction with an exhibit. We present two case studies demonstrating how to apply APEX in practice, analyzing video recordings of participant interactions with different design iterations of TuneTable-an interactive exhibit for co-creative computational music-making-at both a macro- and micro-level. In the case studies, we explore how APEX reveals important features of participant interaction that suggest implications and directions for design. Finally, we present a toolkit of resources to aid researchers in operationalizing APEX as a framework for video analysis, in-situ observation, and iterative design and evaluation.

The Model AI Assignments session seeks to gather and disseminate the best assignment designs of the Artificial Intelligence (AI) Education community. Recognizing that assignments form the core of student learning experience, we here present abstracts of six AI assignments from the 2021 session that are easily adoptable, playfully engaging, and flexible for a variety of instructor needs. Assignment specifications and supporting resources may be found at http://modelai.gettysburg.edu.

Fostering public AI literacy (i.e. a high-level understanding of artificial intelligence (AI) that allows individuals to critically and effectively use AI technologies) is increasingly important as AI is integrated into individuals’ everyday lives and as concerns about AI grow. This paper investigates how to design collaborative, creative, and embodied interactions that foster AI learning and interest development. We designed three prototypes of collaborative, creative, and/or embodied learning experiences that aim to communicate AI literacy competencies. We present the design of these prototypes as well as the results from a user study that we conducted with 14 family groups (38 participants). Our data analysis explores how collaboration, creativity, and embodiment contributed to AI learning and interest development across the three prototypes. The main contributions of this paper are: 1) three designs of AI literacy learning activities and 2) insights into the role creativity, collaboration, and embodiment play in AI learning experiences.

AI is becoming increasingly integrated in common technologies, which suggests that learning experiences for audiences seeking a "casual" understanding of AI-i.e. understanding how a search engine works, not necessarily understanding how to program one-is an increasingly important design space. Informal learning spaces like museums are particularly well-suited for such public science communication efforts, but there is little research investigating how to design AI learning experiences for these spaces. This paper explores how to design museum experiences that communicate key concepts about AI, using collaboration, creativity, and embodiment as inspirations for design. We present the design of five low-fidelity AI literacy exhibit prototypes and results from a thematic analysis of participant interactions during a co-design workshop in which family groups interacted with the prototypes and designed exhibits of their own. Our findings suggest new topics and design considerations for AI-related exhibits and directions for future research.

Museum visitors often come into the museum space receptive to exploring new ideas, and this may encourage members of visitor groups to be supportive and cooperative when engaging together with exhibits. However, as participant groups explore the concepts of the exhibit, interruptions, conflicts, or disagreements may result. We collectively label this social tension as discord. This paper studies discord among family groups interacting with TuneTable, a museum exhibit designed to promote middle school students' interest in and learning of basic computing concepts (e.g. loops, conditionals) through music programming. We analyzed video recordings of each participant group and found that discord often appears alongside three markers of high engagement: a) complex physical manipulation of exhibit components; b) conversation demonstrating an in-depth understanding of how the exhibit works; and c) instances of collaboration between group members. Our findings suggest that certain types of discord could potentially be indicators of productive learning experiences at museum exhibits related to computing. In addition, when designing informal learning experiences for computing education, our findings suggest that discord is a potential trigger for deeper engagement that warrants further exploration.

Artificial intelligence (AI) is becoming increasingly integrated in user-facing technology, but public understanding of these technologies is often limited. There is a need for additional HCI research investigating a) what competencies users need in order to effectively interact with and critically evaluate AI and b) how to design learner-centered AI technologies that foster increased user understanding of AI. This paper takes a step towards realizing both of these goals by providing a concrete definition of AI literacy based on existing research. We synthesize a variety of interdisciplinary literature into a set of core competencies of AI literacy and suggest several design considerations to support AI developers and educators in creating learner-centered AI. These competencies and design considerations are organized in a conceptual framework thematically derived from the literature. This paper's contributions can be used to start a conversation about and guide future research on AI literacy within the HCI community.

Artificial intelligence (AI) is becoming increasingly pervasive in our everyday lives. There are consequently many common misconceptions about what AI is, what it is capable of, and how it works. Compounding the issue, opportunities to learn about AI are often limited to audiences who already have access to and knowledge about technology. Increasing access to AI in public spaces has the potential to broaden public AI literacy, and experiences involving co-creative (i.e. collaboratively creative) AI are particularly well-suited for engaging a broad range of participants. This paper explores how to design co-creative AI for public interaction spaces, drawing both on existing literature and our own experiences designing co-creative AI for public venues. It presents a set of design principles that can aid others in the development of co-creative AI for public spaces as well as guide future research agendas.

Co-creative (i.e. collaboratively creative) activities involving physical interaction are becoming more prevalent in museums as a way of promoting opportunities for exploratory learning-through-doing. However, there is still a need for new techniques for understanding how physical interaction relates to engagement and creative expression in order to both evaluate exhibits and iterate on their design. This article reports on a study of how family groups physically interact in a museum environment with a specific co-creative exhibit--TuneTable. We relate observable markers of physical interaction with stages of engagement/expression based in the literature and identify several different trajectories of participant engagement and creative expression as they navigate the exhibit. We explore what these trajectories tell us about the types of inquiry and experimentation that TuneTable supports and discuss design implications. This paper's main contribution is a deep study of how physical markers reveal trajectories of creative engagement within a specific co-creative installation.

Play-in particular adult-child play---is an important component of child development. This article investigates how to design for adult-child play through Sound Happening, an interactive installation for musical play. We present a preliminary analysis of participant interactions with Sound Happening at The Children's Museum of Pittsburgh, where a total of 112 children and 53 adults interacted with the exhibit. The data from these interactions indicates that Sound Happening can facilitate both verbal parental engagement and partnered adult-child play. We highlight several features of Sound Happening that can be used as design principles for adult-child play environments. These include incorporating embodied interfaces, designing for multiple levels of engagement, and utilizing culturally recognizable interaction modalities.

This work addresses difficulty in web-supported programming. We conducted a lab study in which participants completed a programming task involving the use of the Java Swing/AWT API. We found that information about participant web accesses offered additional insight into the types of difficulties faced and how they could be detected. Difficulties that were not completely solved through web searches involved finding information on AWT/Swing tutorials, 2-D Graphics, Components, and Events, with 2-D Graphics causing the most problems. An existing algorithm to predict difficulty that mined various aspects of programming-environment actions detected more difficulties when it used an additional feature derived from the times when web pages were visited. This result is consistent with our observation that during certain difficulties, subjects had little interaction with the programming environment, they made more web visits during difficulty periods, and the new feature added information not available from features of the modified existing algorithm. The vast majority of difficulties, however, involved no web interaction and the new feature resulted in higher number of false positives, which is consistent with the high variance in web accesses during both non-difficulty and difficulty periods.

Automatic detection of programmer difficulty can help programmers receive timely assistance. Aggregate statistics are often used to evaluate difficulty detection algorithms, but this paper demonstrates that a more human-centered analysis can lead to additional insights. We have developed a novel visualization tool designed to assist researchers in improving difficulty detection algorithms. Assuming that data exists from a study in which both predicted programmer difficulties and ground truth were recorded while running an online algorithm for detecting difficulties, the tool allows researchers to interactively travel through a timeline showing the correlation between values of the features used to make predictions, difficulty predictions made by the online algorithm, and ground truth. We used the tool to improve an existing online algorithm based on a study involving the development of a GUI in Java. Episodes of difficulty predicted by the previously developed algorithm were correlated with features extracted from participant logs of interaction with the programming environment and web browser. The visualizations produced from the tool contribute to a better understanding of programmer actions during periods of difficulty, help to identify specific issues with the previous prediction algorithm, and suggest potential solutions to these issues. Thus, the information gained using this novel tool can be used to improve algorithms that help developers receive assistance at appropriate times.

This paper reports on the design and evaluation of LuminAI, a socially interactive art installation in which participants can engage in collaborative movement improvisation with virtual agents and other humans. LuminAI was used as a technical probe to study social interaction within interactive art at a local art gala during which over 100 participants interacted with the system. Video and interview data was gathered during the event and analyzed using thematic analysis to develop a taxonomy to guide the design of socially interactive systems involving humans and artificial agents. This taxonomy helped us identify areas where LuminAI was successful, where it needs improvement, and conceptual spaces we have yet to explore.