Spatially distributed citizen-science systems depend on where participants contribute, leaving some places well documented and others overlooked. Gamified rewards often promote contribution volume rather than spatial coverage, while fixed rewards may reinforce existing gaps. We extend GAME (Goals And Motivation Engine), a previously developed gamification engine, with an adaptive spatial scoring mechanism that varies points across a map. Under-covered areas retain higher rewards, whereas recently reported locations lose part of their value, encouraging exploration without restricting where participants contribute.We examine GAME in an exploratory field deployment in Barrio Brasil, Santiago, Chile. During a two-hour illegal-dumping mission, 69 enrolled undergraduate students used the system and 11 submitted geolocated reports. Movement traces, system logs, and a standardized gameful-experience questionnaire (GAMEFULQUEST) were used to characterize behavioral and spatial patterns. Reporting was only partly associated with movement. The spatial distribution of delivered reward value, computed using the same proximity-weighted adaptive rule for every valid submission, differed descriptively from the movement distribution and allocated some reward value to under-visited areas. Self-reported gameful experience was associated with overall activity but not with reporting efficiency.Because the deployment lacked a control condition, included few active contributors, and had no independent measure of dumping prevalence, the findings are preliminary and do not demonstrate that adaptive scoring outperforms simpler designs. The study contributes a human–computer interaction account of adaptive spatial gamification in citizen science, a combined behavioral–spatial evaluation approach, and design implications for supporting balanced spatial coverage in equity-relevant civic technologies.
Intelligent projection plays an important role in stage performances, yet in-situ projection authoring remains time-consuming and difficult to adapt to dynamic physical conditions. We use projection authoring to refer to the workflow of translating creative intent into visual content, grounding it in the physical scene, inspecting projected outcomes, and refining the result; image generation is only one step in this workflow. Generative Artificial Intelligence (GenAI) offers opportunities to support this workflow, but applying general-purpose models in situ raises interaction challenges of robustness, alignment, and control. We examine speech-driven in-situ projection authoring as an interaction design problem and present Speech2Pro, a prototype that integrates speech input, generative image models, and depth-aware scene sensing to mediate between user intent, generated content, and physical context. Its enabling mechanism, Contrastive Learning Integrated with Perception Data Fusion (CLIPF), combines contrastive training with depth-aware inference to reduce breakdowns such as ambiguous prompts and projection misalignment. Algorithm-level analysis and a user-centered study indicate improved text–image alignment and spatial precision, and support for iterative creative workflows in stage projection contexts.
Older adults are motivated to develop AI literacy and consider it important. Prior research suggests that conversational agents can support older adults’ AI literacy learning by simplifying AI concepts and providing interactive learning experiences. However, little is known about how such agents should be designed to address the learning challenges older adults face. To address this, we interviewed nine older adults to explore their challenges and identify design opportunities. Findings revealed affective barriers related to confidence and engagement, emphasizing the importance of peer-like characteristics in agents, such as reciprocal knowledge sharing. Guided by these insights, we designed a peer-like agent and compared it with a website and an assistant agent in a between-subjects study with 45 older adults learning about deepfakes. Results showed that the peer agent fostered higher engagement and supported confidence more than the others. We conclude with design recommendations for conversational agents to support older adults’ AI learning.
Fostering students' empathy and creativity toward marginalized user groups is essential to achieving human-centered design in design education. Although experiential learning has gained increasing attention in recent years, there remains a lack of operable and perceptually engaging tools to support empathy instruction. To this end, this study developed a hand pain simulation glove and incorporated it into design instruction. Two quantitative experiments and one qualitative study were conducted to evaluate the glove's pedagogical potential. In the initial experiment, 30 students were subjected to three distinct pedagogical interventions: textual experience, video experience, and embodied experience, with the objective of comparing their respective impacts on empathy scores. A subsequent experiment engaged another cohort of 30 students in product design tasks under different experiential conditions, and their creative outcomes were evaluated by expert raters. Results showed that embodied simulation produced greater gains in empathy and design performance than text- and video-based instruction, particularly in innovative thinking and technical functionality. By translating otherwise invisible and action-dependent pain-related constraints into first-person sensorimotor experience, the glove supported students in reframing user problems and generating more context-sensitive design responses. These findings position embodied pain simulation as a pedagogical approach for empathy-oriented design education and provide a transferable framework for related learning contexts.
Hand function impairment significantly impacts the daily lives of a large population. However, conventional technologies for rehabilitation typically rely on pre-defined rules, failing to adapt to the user's dynamic training state. This article presents the design and evaluation of AdaptGrip, an ultra-personalized adaptive training system for hand function rehabilitation. The system predicts real-time training load by capturing users' cross-modal data, establishing a human-computer interaction loop that synergistically adapts training difficulty and content. We evaluated the role of AdaptGrip and its adaptive components within a single-session short-term training context through a field user study (N = 42) and post-study expert interviews (N = 3). Results indicate that, compared with non-adaptive, AdaptGrip better regulated training load and improved usability and training experience. The qualitative findings further support the advantages of AdaptGrip. We discuss several design insights to inspire future research on ultra-personalized adaptive systems.
