Creative learning has enriched on-site museum education by fostering engagement, exploration, and active participation. However, the structured integration of creative learning processes into digital museum education remains relatively underexplored. Although Generative Artificial Intelligence (GenAI) presents considerable potential to support creative learning, its comprehensive application across all stages in non-formal learning environments, such as museums, remains limited. To investigate this potential, we conducted a formative study employing a previously developed prototype with seven learners and four senior experts to identify learners’ specific needs and challenges. Informed by the findings, we designed and developed MuseForge, a platform integrating GenAI with the five stages of the iterative creative development path to support a personalized and dynamic creative learning experience. A between-subjects study with 32 participants demonstrated that learners using MuseForge achieved significantly higher learning motivation, engagement, and learning gain in creative self-efficacy, highlighting its effectiveness in supporting creative learning in digital museum environments.
Dialogue systems have long supported learner reflections, with theoretically grounded, rule-based designs offering structured scaffolding but often struggling to respond to shifts in engagement. Large Language Models (LLMs), in contrast, can generate context-sensitive responses but are not informed by decades of research on how learning interactions should be structured, raising questions about their alignment with pedagogical theories. This paper presents a hybrid dialogue system that embeds LLM responsiveness within a theory-aligned, rule-based framework to support learner reflections in a culturally responsive robotics summer camp. The rule-based structure grounds dialogue in self-regulated learning theory, while the LLM decides when and how to prompt deeper reflections, responding to evolving conversation context. We analyze themes across dialogues to explore how our hybrid system shaped learner reflections. Our findings indicate that LLM-embedded dialogues supported richer learner reflections on goals and activities, but also introduced challenges due to repetitiveness and misalignment in prompts, reducing engagement.
Educational technology (EdTech) solutions have shown promise for disseminating educational opportunities to last-mile learners, particularly in the Global South. Low-infrastructure EdTech as a digital learning resource is especially critical to understand in remote contexts where educational opportunities and resources are limited. Our work investigated insights from 81 learners who engaged with a remote course that provided engineering education through radios and mobile phones in rural Uganda. Findings revealed that the course facilitated goal-driven and practical motivations in safe, adaptable environments. Our work goes beyond the idea that low-infrastructure EdTech can easily facilitate learning, highlighting diverse learner experiences navigating radio and phone use and presenting novel findings on community skepticism towards the course. Our research extends the EdTech and HCI literature by bringing light to the underrepresented voices of last-mile learners, sharing their insights on interacting with low-infrastructure EdTech, and how these insights can guide the design of contextually aligned EdTech.
Despite growing interest in using LLMs to generate feedback on students’ writing, little is known about how students respond to AI-mediated versus human-provided feedback. We address this gap through a randomized controlled trial in a large introductory economics course (N=354), where we introduce and deploy FeedbackWriter—a system that generates AI suggestions to teaching assistants (TAs) while they provide feedback on students’ knowledge-intensive essays. TAs have the full capacity to adopt, edit, or dismiss the suggestions. Students were randomly assigned to receive either handwritten feedback from TAs (baseline) or AI-mediated feedback where TAs received suggestions from FeedbackWriter. Students revise their drafts based on the feedback, which is further graded. In total, 1,366 essays were graded using the system. We found that students receiving AI-mediated feedback produced significantly higher-quality revisions, with gains increasing as TAs adopted more AI suggestions. TAs found the AI suggestions useful for spotting gaps and clarifying rubrics.
Immersive learning environments such as virtual classrooms in Virtual Reality (VR) offer learners unique learning experiences, yet providing effective learner support remains a challenge. While prior HCI research has explored in-lecture support for immersive learning, little research has been conducted to provide post-lecture support, despite being critical for sustained motivation, engagement, and learning outcomes. To address this, we present AttentiveLearn, a learning ecosystem that generates personalized quizzes on a mobile learning assistant based on learners’ attention distribution inferred using eye-tracking in VR lectures. We evaluated the system in a four-week field study with 36 university students attending lectures on Bayesian data analysis. AttentiveLearn improved learners’ reported motivation and engagement, without conclusive evidence of learning gains. Meanwhile, anecdotal evidence suggested improvements in attention for certain participants over time. Based on our findings of the field study, we provide empirical insights and design implications for personalized post-lecture support for immersive learning systems.
Applying the keyword method for vocabulary memorization remains a significant challenge for L1 Chinese–L2 English learners. They frequently struggle to generate phonologically appropriate keywords, construct coherent associations, and create vivid mental imagery to aid long-term retention. Existing approaches, including fully automated keyword generation and outcome-oriented mnemonic aids, either compromise learner engagement or lack adequate process-oriented guidance. To address these limitations, we conducted a formative study with L1 Chinese-L2 English learners and educators (N=18), which revealed key difficulties and requirements in applying the keyword method to vocabulary learning. Building on these insights, we introduce WordCraft, a learner-centered interactive tool powered by Multimodal Large Language Models (MLLMs). WordCraft scaffolds the keyword method by guiding learners through keyword selection, association construction, and image formation, thereby enhancing the effectiveness of vocabulary memorization. Two user studies demonstrate that WordCraft not only preserves the generation effect but also achieves high levels of effectiveness and usability.
LLMs promise to democratize technical work in complex domains like programmatic data analysis, but not everyone benefits equally. We study how students with varied experiences use LLMs to complete Python-based data analysis in computational notebooks in a graduate course. Drawing on homework logs, recordings, and surveys from 36 students, we ask: Which experience matters most, and how does it shape AI use? Our mixed-methods analysis shows that technical experience – not AI familiarity or communication skills – remains a significant predictor of success. Students also vary widely in how they leverage LLMs, struggling at stages of forming intent, expressing inputs, interpreting outputs, and assessing results. We identify success and failure behaviors, such as providing context or decomposing prompts, that distinguish effective use. These findings inform AI literacy interventions, highlighting that lightweight demonstrations improve surface fluency but are insufficient; deeper training and scaffolds are needed to cultivate resilient AI use skills.