Accepted Workshops
HUMANE-LLM: Workshop on Human-Centered and Responsible LLMs for Agency-Preserving Intelligent User Interfaces
Noemi Mauro, University of Torino, Italy
Julia Neidhardt, CDL-RecSys, TU Wien, Austria
Yashar Deldjoo, Polytechnic University of Bari, Italy
Ludovico Boratto, University of Cagliari, Italy
Dominik Kowald, University of Graz and Know Center Research GmbH, Austria
Large Language Models (LLMs) are reshaping Intelligent User Interfaces by enabling systems that can converse, personalize, recommend, explain, and support decision-making across complex user-facing tasks. Yet, their integration into adaptive interfaces also raises new questions about interaction design, agency, responsibility, transparency, and evaluation. The HUMANE-LLM workshop provides a forum for studying how LLM-based IUIs can be designed, evaluated, and governed in ways that remain human-centered, controllable, accountable, and aligned with users’ needs and values. The workshop brings together researchers and practitioners from IUI, HCI, recommender systems, NLP, responsible AI, and applied domains such as education, healthcare, and decision support.
https://humanellm.github.io/2027
SAMAS: Shaping the Future of AI-Mediated Communication in Asymmetric Scenarios
Ilhan Aslan, TH Köln
Matthias Kraus, University of Augsburg
Verena Fuchsberger, University of Salzburg
Matthias Böhmer, TH Köln
Elisabeth André, University of Augsburg
Timothy Merritt, Aalborg University
Human-to-human communication is increasingly technology-mediated. Contemporary societies face growing communication asymmetries driven by language barriers, cultural gaps, and digital divides. While rapid advances in wearables, LLMs, and affective computing offer real-time assistance, current research remains heavily engineering-centric. This interdisciplinary workshop shifts the focus toward human-centered and entanglement perspectives, examining these dynamics across two core scenarios: co-located one-to-one encounters using wearable displays, and one-to-many remote settings like interactive live streaming. Together, we will critically explore how users experience, negotiate, or resist AI augmentation to shape future, socially acceptable communication ecologies.
MINDXR: Psychological Risks in the Convergence of Extended Reality and AI
Andrea Bellucci (UC3M)
Niklas Ravaja (UH)
Giuseppe Riccardi (UniTN)
Giulio Jacucci (UH)
XR is becoming a primary interface for intelligent systems as AI migrates from screens into headsets, smart glasses, embodied conversational agents, and always-on multimodal interfaces. This convergence raises psychological risks to identity, autonomy, attachment, perception of reality, and agency. These risks are distinct from those posed by screen-based AI or XR in isolation. Existing work treats trustworthy AI, explainability, and immersive interaction separately, leaving the implications of their intersection underexplored. MINDXR convenes researchers from intelligent user interfaces, XR, psychology, affective computing, and responsible AI to characterize these emerging risks, share empirical evidence, discuss design and governance strategies, and establish a shared research agenda for immersive AI that supports rather than undermines psychological wellbeing.
https://www.mv.helsinki.fi/home/jacucci/mindxr/#cfp
HIMAS: Human-Centered Interactive Multimodal AI Systems
Gonçalo Marcelino, University of Amsterdam, The Netherlands
Melika Ayoughi, University of Amsterdam, The Netherlands
Sarah Binta Alam Shoilee, Vrije Universiteit Amsterdam, The Netherlands
Omar Shahbaz Khan, IT University of Copenhagen, Denmark
Andrea Mauri, Université Claude Bernard Lyon 1, France
Yen-Chia Hsu, University of Amsterdam, The Netherlands
Maximilian T. Fischer, University of Konstanz, Germany
Senthil Chandrasegaran, TU Delft, The Netherlands
Marcel Worring, University of Amsterdam, The Netherlands”
This workshop brings together researchers from multimedia analytics, information visualization, multimodal AI, and human-computer interaction to advance the design, development, and evaluation of Human-Centered Interactive Multimodal AI Systems (HIMAS). Organized as a full-day mini-conference, the workshop is structured around three themes: integrating multimodal AI with interactive interfaces, designing HIMAS for effective human-AI collaboration, and adapting these systems across domains. Through paper presentations, system demonstrations, and structured discussions, we aim to identify shared challenges, establish a research agenda, and foster a lasting cross-disciplinary research community.
LEXFLOW: Workshop on Agentic Workflows for Human-AI Legal Reasoning
Sirui Han(HKUST),
Yuyao Zhang(HKUST),
Yidan Huang(HKUST),
Chuxue Cao(HKUST),
Yujin Zhou(HKUST),
Guoying Lu(HKUST)
LEXFLOW is a full-day workshop on intelligent interfaces for agentic legal reasoning workflows. It brings together researchers from HCI, IUI, AI and Law, and human-centered AI to explore how legal AI systems can better support process-level supervision, including inspection, steering, verification, contestation, and accountability. Through invited talks, paper and demo presentations, workflow mapping, co-design activities, live critique, and collaborative synthesis, participants will develop reusable design patterns, workflow artifacts, evaluation questions, and a shared research agenda for more transparent and accountable human-AI legal reasoning systems.
https://trista-z.github.io/LEXFLOW/
AI CHAOS! 2nd Workshop on the Challenges for Human Oversight of AI Systems
Dr. Tim Schrills, University of Lübeck, Germany (primary contact)
Dr. Patricia Kahr, University of Zurich, Switzerland
Prof. Dr. Markus Langer, University of Freiburg, Germany
Prof. Harmanpreet Kaur, PhD, University of Minnesota, USA
Prof. Ujwal Gadiraju, PhD, Delft University of Technology, The Netherlands
Human oversight has become the central safeguard between increasingly capable AI systems and the people affected by them. With the EU AI Act’s high-risk obligations now in force, effective oversight by natural persons — including appropriate human-machine interface tools — has moved from regulatory vision to design reality. Yet oversight frequently fails in practice: overseers rubber-stamp outputs, miscalibrate reliance, or lack the visibility and controls to intervene, especially with agentic systems. Whether oversight is meaningful or merely nominal is largely decided at the user interface. AI CHAOS! convenes researchers and practitioners around one question: how do we design, build, and evaluate intelligent user interfaces that make human oversight of AI actually work?
https://sites.google.com/view/aichaos/iui-2027?authuser=0
Models in Human-AI Interaction: Methods and Challenges in the Generative and Agentic AI Era
Téo Sanchez, LMU Munich
Bhada Yun, ETH Zürich
Prerna Ravi, MIT
Laura Schütz, TU Munich
Anna Neumann, University Duisburg-Essen
Robin Chan, ETH Zürich
April Yi Wang, ETH Zürich
Qiaosi Wang, CMU HCII
Sumit Asthana, Microsoft
The mental model construct is widely used in HCI to refer to the knowledge structure people hold to reason about computing systems. Yet the construct is often used interchangeably with related concepts (e.g. folk theories, sensemaking) and methods of studying it are many and diverse. Generative and agentic AI systems further complicate mental model formation and elicitation as such systems are opaque by design and increasingly act on users’ behalf across files, applications, and on the web. Together, these challenges may hinder the commensurability of research on people’s mental models of AI systems. The MM/AI workshop calls for a critical reassessment of how we understand and study mental models in human-AI interaction research. It aims to foster theoretical and methodological exchange on mental models in human–AI interaction, identify open challenges, and develop directions for future research.