Voice UIs, also called conversational UIs or conversational agents, have become mainstream technology: they are integrated into many modern smartphones and are also sold as standalone appliances for the home. Recent advances in hardware processing capabilities, coupled with improvements in machine learning techniques and the availability of significant datasets fuelling them, has led to increasing accessibility and commoditisation, meaning that broader design interest in ‘conversational’ interfaces has flourished.
Although there is a body of prior work related to conversational interfaces, HCI has yet to respond substantially to this recent proliferation of voice and—aspirationally—conversational interactions. Not only is there a paucity of studies of conversational UI in actual use (particularly from fieldwork), but also a growing need to understand how findings from practice-oriented empirical studies might feed into design via guidelines, frameworks, and so on. Thus, this workshop seeks to directly address this by bringing together HCI researchers investigating one or both aspects. We aim to use the workshop to share, and draw connections between the following:
- qualitative studies of different devices in use (existing consumer products, research prototypes, etc.) to help consolidate our understandings of user practices and experience;
- emerging design knowledge (e.g. principles, heuristics, frameworks, etc.) that can address and apply to future iterations of such conversational interfaces.
The design and study of voice-based and conversational interaction is diverse and interdisciplinary, drawing upon work from many backgrounds including sociolinguistics, design and user experience, sociology, and computational paralinguistics. Furthermore, the work lends itself to myriad research approaches, including ethnomethodological and conversation analytic work, interviews, and Wizard of Oz and technical probe approaches.
In this workshop, we particularly seek to encourage fostering an understanding of voice interaction as it happens, in order to generate design insights from observations of voice interaction practices. This practice-oriented approach, coupled with an interest in how such accounts map to design outcomes, is underexplored. Many challenges exist: how users develop new forms of talk that fit in with voice UI characteristics, how difficulties with transcription accuracy can be better handled through design, or how users deal with misunderstandings and unintended actions of voice-based interfaces. In addressing this and empirical studies of use, we are interested in a sense of design pragmatism: for instance, rather than relying on future technical improvements that seek to eliminate ‘error’ we are interested in errors and repairs as a fundamental feature of everyday talk, and how design can provide more resources for users of voice UI to repair and progress interactions.
This workshop brings together organisers from previous workshops at CHI and CSCW which explored the user experience of voice UIs and designing for collaborative action. In this workshop, we welcome both examinations of consumer devices (e.g. Amazon Echo, Siri, Google Home), as well as the study and design of research prototypes and experimental systems. Regarding the domains and situations of use, we are not only interested in established discretionary uses (such as interactions with household devices like the Amazon Echo), but also welcome consideration of design for settings where there are other constraints or challenges which present opportunities or requirements for voice interactions. For example, one might consider situations of use which can potentially benefit from hands-free interaction with technology, such as in-car, sterile environments, or during more mundane activities in the home, such as cooking. These situations of use also connect with voice UI applications for accessibility.
As organisers, our intention is for the workshop to draw upon the strong interdisciplinary foundation both within the CHI community and on the research of voice and conversational interface design in industry and academia. Through an open exchange of ideas and discussion, we will work to identify and position the role of HCI’s more explicitly practice-oriented approaches for contributing to the design of voice UI, and set an agenda for how this interactional modality (both in its design and study) might be addressed in conceptually novel ways.