Let’s Chat! Human-Centric Methods for Designing Chatbots

How can we design better conversations? This project helps by applying and developing new team-oriented methodologies to make chatbots. Chatbots are programs that help people achieve goals through written conversation. However, chatbot makers and researchers appear to use few human-centric methods to design such naturalistic experiences. This work explores three proposed methods developed to help teams effectively design conversations, determining the goals and purpose of a conversation, crafting what is said and when it is said. These methods also illustrate experiential approaches for developing interesting artificially intelligent experiences, including participatory design activities, applying empathy and human-centered processes towards personality design, and tactile prototyping before working in code.

To understand central problems in making chatbots, this project uses analogous observation, expert interviews, and theoretical analysis, uncovering connections between conversation and trust. This research suggests designing for dialogue, personality, and conversational structure in order to create trustworthy chatbot experiences. The resulting collaborative methods are Improv, Bot Personas, and Conversation Mapping. These prototype design methods were iterated and tested over three workshops held with digital professionals. Bot Personas and Conversation Mapping proved to be very effective, while Improv remains imperfect.

In the course of this project, new insights are shared on a variety of chatbot design topics. These learnings include new approaches to personality design, empathizing with services and users simultaneously, techniques in visualizing conversation, novel techniques in structuring conversation, differences in desired versus practical conversations, exploring challenges unique to designing complex conversations, and ethics for chatbot behavior. Each indicate significant opportunities for future exploration.