current Research

As large language models (LLMs) become increasingly central to human–AI interaction, traditional chatbot interfaces—typically designed as linear, single-window chat lists—are no longer adequate. Such layouts constrain how users can review, reuse, or reorganize information from long and complex conversations. Users often lose context, struggle to manage prompts effectively, and experience a growing cognitive burden as interactions extend beyond simple question–answer exchanges.

Our research explores how interface design can redefine the way people collaborate with LLMs. Our goal is to help users regain control, reduce mental load, and make prompting more transparent and effective. Through our work, we aim to establish new design principles for next-generation LLM interaction environments that are intuitive, flexible, and user-centered.

Associated Studies

Graphical user interfaces (GUIs) are the primary medium through which people interact with digital systems, yet enabling AI to truly understand them remains a major challenge. Traditional approaches rely heavily on manual inspection or rule-based analysis, which are costly, time-consuming, and limited in scalability. To make AI systems more capable of reasoning about usability and visual design, it is essential to teach them how to interpret and leverage GUI structures directly.

Our research investigates how AI agents can comprehend and utilize GUI information to detect usability issues more efficiently. We develop models that can identify display problems, predict users’ perception, and diagnose potential design flaws with minimal human intervention. Beyond building specialized models, we also explore how large language models (LLMs) can be guided to interpret and reason over GUIs, enabling a new generation of AI systems that support low-cost, intelligent usability evaluation.

Associated Studies

Virtual environments are evolving beyond immersive visualization into dynamic spaces where people work, create, and collaborate. However, supporting seamless interaction in such environments poses unique challenges—from designing accessible graphical interfaces in VR, to reducing rendering costs in virtual production, to enabling AI systems that assist users’ tasks in real time. These challenges demand technologies that bridge perception, cognition, and intelligent system design.

Our research develops intelligent techniques that enhance interaction within virtual worlds. We explore how AI can understand spatial context, support human actions, and optimize system performance to make virtual environments more efficient and adaptive. Through this line of work, we aim to create human-centered, AI-supported virtual experiences that expand both usability and creative potential.

Associated Studies

In addition to our core research topics, we continue to explore a wide range of themes in Human–Computer Interaction (HCI). Our interests extend to GUI, extended reality (XR), and human factors—areas where software, interaction, and human experience intersect.

Through these ongoing explorations, we aim to identify new challenges and opportunities that complement our main research directions. By integrating emerging technologies and human-centered design perspectives, we seek to broaden the scope of our contributions to the evolving HCI landscape.