Conversational UX Design CHI 2017 Workshop - overview

From Siri to Alexa to Cortana, conversational interfaces are hitting the mainstream and becoming ubiquitous in our daily lives. However, user experiences with such applications remain disappointing. Although it is easy to get a system to produce words, none of the current agents or bots display general conversational competence. Modeling natural conversation is still a hard problem. But in order to tackle it, conversational UX designers must possess a technical understanding of the structures of natural conversation. This workshop explores the intersection of user interface design and the design of natural conversation. It seeks to outline principles and guidelines for Conversational UX Design as a distinct discipline. Workshop participants will get their hands dirty building conversation flows.


While virtual agents and chat bots have been around for decades [1], there has been a recent resurgence of interest in them as major computer companies have released their own. Apple's Siri, Google's Assistant, Microsoft's Cortana, Amazon's Alexa, Facebook's M and IBM's Watson are just a few examples, not counting conversational agents by startups.

With persistent Internet connections and statistical algorithms, virtual agents are much smarter today than they were 20 years ago. While most of these systems accept voice input from users, a growing number accept text input, sometimes from standard applications like SMS and Instant Messaging. But although today’s virtual agents are often touted as “easy to build,” interactions with them are still awkward, confusing, limited and fraught with troubles in mutual understanding.

Conversational interfaces are very different from graphical user interfaces. In conversational interfaces, the graphical elements are generally minimal, for example, a chat history and text box or microphone button (Figure 1) or nothing at all. User interaction is conducted primarily through the words: typed or spoken. The interaction metaphor for these interfaces is the natural, human conversation.

Although natural language processing has given us powerful, automated tools for analyzing the spoken and the written word alike, it does not provide a model of how bits of language are sequenced by multiple parties into an interaction that is recognizable as a “conversation” [3]. Natural human conversation is a complex system [7,3], which Harvey Sacks called a “machinery” in its own right [6]. How to create a user interface that mimics features of such a machinery is nontrivial. But rather than avoiding this complexity by producing simplistic interactions, we should embrace the complexity of conversational systems because it “mirrors the complexity of the world,” while at the same time avoiding any complexity that is due instead to “poor design” [4].

The time is ripe for developing Conversational UX Design as a distinct discipline. Just as graphical user interfaces improved dramatically as visual artists became involved in development (1990s-2000s), so will conversational interfaces when conversation experts get involved. Rather than a background in the visual arts, conversation experts possess a background in the study natural conversation, for example, in the fields of sociology, communication, linguistics, psychology, etc. Conversation Analysis, in particular, offers over 50 years worth of rich, empirical studies of naturally occurring talk-in-interaction in a wide range of settings and languages, which offer formal, qualitative models of how natural conversation is structured. While the proposal to apply these findings to the design of dialogue interfaces may not be entirely new [1,5], it has become especially timely as new conversational technology platforms are becoming ubiquitous.

Conversation experts are keen observers of natural conversation and can articulate the mechanics of human conversation, which most others know only tacitly. For example, a conversation expert may describe the function of the word “oh” to mark speakers’ realizations [2] or how the phrase, “to the what?,” in response to “I’m going to the workshop,” elegantly elicits a repeat of a single word “workshop” [8]. Conversational UX designers use such observations of the machinery of human conversation in building conversational machines.

The goals of this workshop are to explore the intersection of UX design and the analysis of natural, human conversation in the context of text- or voice-based virtual agents and to begin to define a set of design principles and guidelines for today’s conversational platforms.