Artificial Speech is Culture:

Abstract

As conversational voice assistants (CVAs) like Alexa, Siri, and Google Home have become ubiquitous in some parts of the world, lab-based reception studies have captured perceived failures in their usability or accounted for ways that these devices reproduce problematic (often racialized and gendered) cultural conditions. But examining these devices solely as end products fails to account for the ways that UX design, like other expert labor, is a social practice born of daily workplace negotiations and inevitable compromises. This U.S. and Japan-based study argues that anthropological methods are critical to our understanding of conversation design and device training processes and explores why we must go beyond quantifying device usability to examine the work of the usability professionals themselves. For the conversational user experience (CUX) designers and researchers I interviewed and with whom I am conducting ongoing fieldwork, current technology frustrates their capacity to design for systems that go beyond being tools to also be enjoyable conversation partners. In some cases, the process of creating this category of AI turns the canonical literature on computers as social actors on its head, by identifying conscious intentions on the part of designers to replicate human personality and use it to drive the interactions we have with CVAs. It also suggests that the end users who most successfully adopt these tools are frequently those who best reproduce the culture of UX designers themselves.