MedGPT is a curriculum-aligned AI platform developed to integrate artificial intelligence into undergraduate medical education. Grounded on Medicine teaching materials — including lecture slides, PDFs, and time-linked lecture videos — MedGPT allows students to ask questions and receive responses aligned to their current phase of study, with precise source citations that trace concepts back to specific lectures and resources. The platform supports guided questioning and targeted quiz generation to help students clarify concepts, test understanding, and engage more deeply with course material at an appropriate cognitive level. In parallel, MedGPT provides instructors with anonymized learning analytics, supporting reflective teaching practices and curriculum refinement by making areas of student difficulty more visible. This initiative explores a structured, curriculum-embedded approach to AI-supported learning that remains closely aligned with institutional teaching goals and undergraduate learner needs.

A curriculum-aligned AI learning platform supporting medical students with guided questioning, quiz generation, and faculty learning analytics.

MedGPT: A Curriculum-Aligned AI Platform for Medical Education

MedGPT

Responsiveness in large language model (LLM) applications is widely assumed to be critical, yet the impact of latency on user behavior and perception of output quality has not been systematically explored. We report a controlled experiment varying time-to-first-token latency (2, 9, 20 seconds) across two taxonomy-driven knowledge task types (Creation and Advice). Log analyses reveal that user interaction behaviors were robust to latency, yet varied by task type: Creation tasks elicited more frequent prompting than Advice tasks. In contrast, participants who experienced 2-second latencies rated the LLM's outputs less thoughtful and useful than those who experienced 9- or 20-second latencies. Participants attributed delays to AI deliberation, though long waits occasionally shifted this interpretation toward frustration or concerns about reliability. Overall, this work demonstrates that latency is not simply a cost to reduce but a tunable design variable with ethical implications. We offer design strategies for enhancing human-LLM interaction. Accepted and to be published at CHI 2026.

Investigating how response latency affects user behavior and perception of LLM output quality across different task types.

The Impact of Response Latency and Task Type on Human-LLM Interaction and Perception

LLM Response Latency

Drones have gained traction as a versatile form of assistive robots for Blind and Low Vision (BLV) people. Nonetheless, novel interaction techniques are required to enable BLV people to communicate with drones naturally. In this work, we built an LLM-powered assistive drone for BLV users. We leverage an LLM to translate high-level user goals to step-by-step instructions for the drone and to extract visual information from images. Through a formative study with BLV users (N=9), we identified envisioned use cases and desired interaction modalities. Then, we took a participatory and iterative approach to build a prototype, incorporating feedback received from 3 BLV users, as well as 5 domain experts. Finally, we conducted a user study with an additional 6 BLV participants to evaluate the iterated prototype, and received positive feedback. This work is contributing to a growing body of research on harnessing the power of LLMs to build a more inclusive world. Accepted and to be published at CHI 2026.

An LLM-powered assistive drone enabling blind and low vision users to interact with their environment through natural language.

Towards LLM-powered Assistive Drone for Blind and Low Vision Users

LLM Drone

Gastric interoception influences eating behavior and emotions, making its modulation valuable for healthcare and human-computer-interaction applications. However, whether gastric interoception can be modulated noninvasively in humans remains unclear. While previous research indicates that abdominal-sound-driven haptic feedback resembles gut sensations, its impact on gastric feelings and gastric interoceptive behavior is unknown. We conducted three experiments totalling 55 participants to investigate how gut-sound-driven audio-haptic feedback applied to the stomach (1) affects user's feelings (2) influences perception of hunger and satiety levels and (3) influences gastric interoceptive behavior, quantified with Water Load Test-II. Results revealed that audio-haptic feedback patterns induced the feelings of hunger, fullness, thirst, and stomach upset, increased hunger level, and significantly increased volumes of ingested water. This work provides the first evidence that audio-haptic stimulation can alter gastric interoceptive behavior, motivating the use of noninvasive methods to influence users' feelings and behaviors in future applications. Accepted at CHI 2026 — Honorable Mention.

Investigating how gut-sound-driven audio-haptic feedback applied to the stomach can modulate gastric feelings and eating behavior.

VisceroHaptics: Investigating the Effects of Gut-based Audio-Haptic Feedback on Gastric Feelings and Gastric Interoceptive Behavior

VisceroHaptics

ANISMA is a software and hardware toolkit to prototype on-skin haptic devices that generate skin deformation stimuli like pressure, stretch, and motion using shape-memory alloys (SMAs). The toolkit embeds expert knowledge that makes SMA spring actuators more accessible to human-computer interaction (HCI) researchers. Using the software tool, users can design different actuator layouts, program their spatio-temporal actuation and preview the resulting deformation behaviour to verify a design at an early stage. The toolkit allows exporting the actuator layout and 3D printing it on skin adhesive, which allows pasting the device directly on the skin. To test different actuation sequences on the skin, a user can connect the SMA actuators to the customized driver board and reprogram them using the visual programming interface.

A toolkit to design more natural feeling touch feedback for extended reality applications.

ANISMA: A Prototyping Toolkit to Explore Haptic Skin Deformation Applications Using Shape-Memory Alloys

ANISMA

We introduce Prompt-to-Touch, a proof-of-concept of a multi-step pipeline that generates haptic effects based on the textual description of a haptic experience. Our pipeline first translates the haptic effect description to a sound effect description using our Foley-Interpreter component. It then uses a text-to-audio model to generate a sound effect from the sound description. Afterwards, the sound effect is converted into a perceivable haptic effect using our Dynamic-Audio-Processor. Finally, the haptic effect is post-processed to compensate for actuator-specific frequency response characteristics. We validate our concept in two preliminary human evaluation studies (n=20, n=10) and a technical analysis. Our results indicate that our pipeline can generate effects to enhance immersive multimedia experiences, abstract desktop/XR interactions, and social communication applications. They also still reveal significant potential for further improvement. Our pipeline could be used to develop future text-driven haptic design and automation tools. We provide open-source code to support future extensions.

Automatic generation of touch experiences based on their textual description.

Prompt-To-Touch: Enabling Automatic Haptic Effect Generation from Text Prompts

Prompt-To-Touch

LitPanda is an on-going project in development. LitPanda can analyze and synthesize insights from hundreds of papers at the same time and translate report/paper drafts into fully fledged research papers. LitPanda has the potential to accelerate common tedious research tasks by a factor of 1000x, allowing researchers to focus on innovative solutions and saving research organizations valuable time and money. If you are interested in this project, please reach out!

Accelerating research through a human-AI collaborative research platform.

LitPanda: A human-AI collaborative research platform

LitPanda

Prior work has shown that providing spatio-tactile feedback through vibration motors mounted at the finger pad can enrich various virtual reality (VR) applications. However, in mixed reality (MR), the user frequently interacts with the physical world. To enable seamless MR interactions, it is therefore essential to keep the finger pad free to preserve the user's natural sense of touch and support in-air as well as on-surface interaction spaces. In this paper, we explore PhantomFolds, an unobtrusive nail-mounted device consisting of two linear resonant actuators (LRAs) located at the lateral nail folds to provide spatio-tactile feedback for in-air as well as on-surface interactions in MR. In a series of three studies, we investigate the perception of spatio-tactile feedback produced by PhantomFolds under various conditions. Our results show that (1) PhantomFolds can leverage the funneling illusion to produce spatio-tactile feedback at the finger for in-air as well as on-surface interactions, (2) tactile feedback produced with PhantomFolds can successfully increase the perceived realism in MR applications without impeding user interaction, (3) phantom sensations (created by the funneling illusion) are perceived more localized when users touch a surface, and (4) while participants could not consistently feel touch illusions below the finger when touching a surface as described by prior work, our results indicate that a per-user calibration could increase the success rate of this effect in the future.

Exploring touch illusions below the finger using finger nail mounted vibration motors.

PhantomFolds: Exploring Unobtrusive Spatial Tactile Feedback Produced by Two Fingernail Mounted LRAs for In-Air and On-Surface Mixed Reality Applications

PhantomFolds

UbiPen uses a single finger-nail mounted inertial measurement unit (IMU) to transform ordinary pens into digital input devices. We demonstrate how our technique enables the detection of pen-to-surface touch events, allows tracing 2D pen-tip movements and supports discriminating between different input surface and pen types. We showcase several applications that blend analogue and digital functionalities while preserving the traditional pen writing experience that most digital stylus devices struggle to replicate.

Transforming ordinary analog pens into digital input devices.

UbiPen: Digitizing Analogue Pens With a Finger-Mounted IMU

UbiPen

We present TORO (https://www.toro.ac.nz/), a web-based application that aims to simplify the time-consuming literature research process by providing an easy way of searching, exploring, screening, and comparing literature inside a visual paper graph. Unlike many automated tools, the user actively explores papers and constructs the graph. Users can add papers by searching across different publishers or pasting DOI links and bib-files. Users can analyse relations between papers in one view, look up common keywords, access paper details and categorize them easily. Expanding references allows exploring interesting streams of literature and performing search strategies like snowballing easily across platforms. To manage large numbers of papers and identify interesting work, users can define filters and highlighting criteria. We present our initial design and implementation and discuss the results of a preliminary user study with 18 researchers. The results indicate that TORO already helps researchers' exploration process and highlights opportunities for future work.

Quickly discover scientific work through a visual literature exploration tool.

Toro: A Web-based Tool to Search, Explore, Screen, Compare and Visualize Literature Authors

Toro

Emerging research has demonstrated the viability of on-textile actuation mechanisms, however, an easily customizable and versatile on-cloth actuation mechanism is yet to be explored. In this paper, we present ClothTiles along with its rapid fabrication technique that enables actuation of clothes. ClothTiles leverage flexible 3D-printing and Shape-Memory Alloys (SMAs) alongside new parametric actuation designs. We validate the concept of fabric actuation using a base element, and then systematically explore methods of aggregating, scaling, and orienting prospects for extended actuation in garments. A user study demonstrated that our technique enables multiple actuation types applied across a variety of clothes. Users identified both aesthetic and functional applications of ClothTiles. We conclude with a number of insights for the Do-It-Yourself community on how to employ 3D-printing with SMAs to enable actuation on clothes.

A prototyping platform to create functional clothing that can deform itself.

ClothTiles: A Prototyping Platform to Fabricate Customized Actuators on Clothing using 3D Printing and Shape-Memory Alloys

ClothTiles

Short-form video platforms like YouTube Shorts captivate users with engaging content, but their potential for promoting incidental learning remains underexplored. We present Curious Shorts, a conceptual framework that extends the Hook Model, designed to enhance curiosity-driven exploration and incidental learning on these platforms. In Study 1, we empirically tested two designs that incorporate 'curiosity nudges' — interactive prompts that spark curiosity and encourage further exploration — with follow-up videos to satisfy that curiosity. Results show that specific, question-driven prompts proved most effective, significantly boosting curiosity and encouraging more focused and intentional viewing compared to the baseline. Study 2 examined whether this design enhances incidental learning without compromising engagement. Findings confirmed improved learning outcomes. However, when applied to a realistic viewing environment interspersed with entertainment videos, engagement remained high while learning benefits diminished. We conclude with implications for balancing learning and engagement on short-form video platforms and propose directions for future research.

Investigating how exploratory questions can drive curiosity and learning on short-form video platforms.

Curious Shorts: Curiosity-Driven Exploration and Learning on Short-Form Video Platforms

Curious Shorts

Designing effective haptic feedback is challenging due to the subjective nature of touch and the fact that multiple people cannot easily share and evaluate touch experiences. In this work, we propose CoTacs, a collaborative haptic toolkit to address the challenges of haptic feedback design by enabling designers to explore haptic experiences together using group collaboration and allowing them to refine ideas quickly through early feedback during the design process. Two design sessions with five collaborators each show that CoTacs enables users to design, evaluate, refine and explore haptic feedback together. Participants leveraged group feedback to improve their designs and inspire creative ideas for the second collaboration round. Our work demonstrates how collaborative haptic toolkits can enable synergistic effects through interactions between the designers, benefiting haptic feedback design while still exposing several limitations. We discuss different opportunities for the design of future collaborative haptic toolkits and haptic group collaboration techniques.

Enabling haptic designers to collaboratively explore novel touch experiences.

CoTacs: A Haptic Toolkit to Explore Effective On-Body Haptic Feedback by Ideating, Designing, Evaluating and Refining Haptic Designs Using Group Collaboration

CoTacs

SwiftPrompt started as one of my private hobby projects for MacOS. Since using it, I have found it to be so convenient that in many cases I do not use the traditional ChatGPT interface and Google Search anymore. This and the positive feedback from my friends convinced me to make SwiftPrompt accessible to everyone. I hope the tool will be useful for you too!

Integrating AI into desktop workflows using a lightweight prompt terminal.

SwiftPrompt: A lightweight prompt terminal to integrate AI in your desktop workflows

SwiftPrompt

The Haptic Vest 2.0 is a wearable vest that uses a front-facing depth camera and an array of 128 vibration motors to let its wearer feel an image of the obstacles in front of him and their individual distance. It is part of our BMBF-funded project Personal Photonics that aims at making photonic, or optical, technologies like lasers and depth cameras accessible to the DIY maker community. Imagine putting on the Haptic Vest, which is actually more like a tight-fitting, broad belt across your lower chest, then closing your eyes, and starting to walk. As you approach a lamp post that you are in danger of running into with your right shoulder, you start feeling a vertical column of motors vibrating on the right half of the vest. As you move closer, the vibration gets stronger and slowly shifts towards the right. You make a small left turn to correct your course, and the vibrating column moves further to your right, indicating that you will now walk past that lamp post and no longer hit it as you continue walking.

A wearable vest that enables vision impaired people to feel their environment.

Haptic Vest: A wearable vest that enables vision impaired people to feel their environment

Haptic Vest

The Badge Maker does not only make badges, but also bachelors! I developed this open source CNC milling machine as part of my bachelor thesis. The Badge Maker did not only receive great interest by children, but was sought after by many visitors at the many exhibitions that we visited.

Devices like small sized desktop 3D printers and milling machines are already finding their way into our private households. These personal fabricators may be beneficial for many different applications like prototyping, spare part fabrication or individual product design. But, so far, many of these machines still show to be cumbersome to use. Novice users often have to face several challenges before they can operate such devices and achieve their desired results. These challenges may also form an entry barrier for people who are not familiar with such devices and prevent them from getting into touch with this technology.

My approach to solve this problem was the 'Badge Maker'. This stand-alone device should provide novice users, like children, the opportunity to gather early experiences with the fabrication technology by offering them an intuitive and easy-to-use milling machine. It enables them to create their own illuminated acrylic name tags in a safe and self-directed way.

Enabling novice users to gather first experiences with personal fabrication machines.

Badge Maker: Developing an Easy-to-use System to Design and Build Illuminated Acrylic Name Tags for Children

Badge Maker

This is an on-going project where I explore how different feedback modalities influence user performance in on-the-go mixed-reality gaze interactions.

Exploring how different feedback modalities influence user performance in on-the-go gaze interactions.

Moritz Alexander Messerschmidt