Human-Computer Interaction (HCI) is undergoing a profound epistemological shift. For decades, the field has developed a rich repertoire of research methods informed by its interdisciplinary origins. These methods have enabled HCI researchers to explore phenomena across a wide range of domains—from desktop work environments to everyday life—with approaches that span controlled experiments, ethnographic fieldwork, and design interventions. However, the increasing integration of artificial intelligence (AI), and particularly generative AI (genAI), into both interactive systems and research workflows is disrupting the foundational assumptions behind these methods.

This Dagstuhl Seminar invites participants to examine how HCI research methods must adapt in light of these developments—and what new opportunities and risks this transformation entails. The seminar addresses a timely and pressing set of challenges. As AI systems become more capable of generating content, modelling behaviour, and even interpreting findings, they shift our understanding of what it means to design, evaluate, and study interactive systems. This raises new questions about the role of researchers and participants, the definition of rigour, and the boundaries between empirical, conceptual, and constructive work. The seminar will explore how AI can be positioned not only as a topic of study within HCI but also as a tool that shapes the research process itself. This includes the use of AI to produce knowledge, assist in design and evaluation, and even act as a collaborator in framing research questions.

Key topics of discussion include:

The use of AI as an autonomous or assistive research tool in HCI, and the implications for researcher agency, ethics, and validity;
The generation and application of synthetic data, including simulated interviews and user behaviours, exploring if and when such data can meaningfully replace or complement human studies;
The evolving perceptions of AI autonomy and consciousness, particularly in how they influence user trust and blur distinctions between human and machine contributions;
The redefinition of evaluation metrics to better capture trust, transparency, and other user experiences unique to AI-powered systems;
The tensions between methodological traditions in HCI and the demands of rapidly changing technological capabilities.

We will also devote attention to broader epistemological and ethical issues, including questions of research ownership, reproducibility, and the potential erosion of human-centred values in a climate increasingly driven by speed, automation, and cost-efficiency. At the same time, we see opportunities to reimagine how HCI methods can evolve—through interdisciplinary collaboration, engagement with industry, and the development of new frameworks for method design and validation.

This seminar is intended for researchers and practitioners from across the HCI spectrum. A particular emphasis will be placed on bridging academic and industry perspectives to ensure that proposed approaches are both conceptually grounded and practically viable. Through structured discussions and creative exploration, the seminar aims to create in a shared research agenda outlining priorities for advancing HCI research methods in the context of AI. It also aims to produce guidelines on the use of synthetic data and a set of community statements reflecting diverse viewpoints on the evolving role of AI in HCI research.

Creative Commons BY 4.0

Nitesh Goyal, Jasmin Niess, Yvonne Rogers, and Pawel Wozniak