Adaptation, or ability and willingness to consider an alternative approach, is a critical component of learning through reflection, especially in educational games, where there are often multiple avenues to success. As a domain, educational games have shown increased interest in using retrospective visualizations to promote and support reflection. Such visualizations, which can facilitate comparison with peer data, may also have an impact on adaptation in educational games. This has, however, not been empirically examined within the domain. In this work, we examine how comparison with other players’ data influenced adaptation, a part of reflection, in the context of a game that teaches parallel programming. Our results indicate that comparison with peers does significantly impact willingness to try a different approach, but suggest that there may also be other ways. We discuss what these results mean for future use of retrospective visualizations in educational games and present opportunities for future work.

Esports play can cultivate real world skills. However, the path to mastery is not easy, and difficulty progressing can result in discontinuation. In the absence of a human coach, computational tools may provide much-needed guidance. However, the specific improvement activities that players engage in and the exact challenges they face are not well defined in the context of computational support. As such, most tools can only support players based on a high level understanding of their practices. We present the results of an interview study (n=17) that identified four improvement activities: practicing, leveraging the knowledge of others, tracking performance, and reflecting on gameplay and setting goals for the future, and four challenges: coordinating and collaborating with teammates, knowing what to do next, tracking game state, and tracking skill and improvement. We discuss six implications for future design and development based on these results.

Predictive models are increasingly used to make various consequential decisions in high-stakes domains such as healthcare, finance, and policy. It becomes critical to ensure that these models make accurate predictions, are robust to shifts in the data, do not rely on spurious features, and do not unduly discriminate against minority groups. To this end, several approaches spanning various areas such as explainability, fairness, and robustness have been proposed in recent literature. Such approaches need to be human-centered as they cater to the understanding of the models to their users. However, there is little to no research on understanding the needs and challenges in monitoring deployed machine learning (ML) models from a human-centric perspective. To address this gap, we conducted semi-structured interviews with 13 practitioners who are experienced with deploying ML models and engaging with customers spanning domains such as financial services, healthcare, hiring, online retail, computational advertising, and conversational assistants. We identified various human-centric challenges and requirements for model monitoring in real-world applications. Specifically, we found that relevant stakeholders would want model monitoring systems to provide clear, unambiguous, and easy-to-understand insights that are readily actionable. Furthermore, our study also revealed that stakeholders desire customization of model monitoring systems to cater to domain-specific use cases.