April 2022 Summaries

The text discusses the importance of interaction models in conversational AI assistants, which are modeled after human conversation patterns to create a more natural experience for users. An interaction model should be informed by initial user research and define scenario ideations, information architecture, and conceptual models that capture the user's potential thought processes and questions. This model serves as a blueprint for other components of the conversational assistant, including NLU and dialog management. A well-designed interaction model influences every aspect of the conversation design experience and is crucial for creating a smart and effective conversational assistant. The text also highlights the importance of iteratively refining the NLU model based on live customer data and user feedback, and explains how dialog management plays a critical role in converting user inputs into machine outputs to create a natural conversational experience.

Every business’ conversation designs are going to be specific to their business, the team leveraging conversational experiences, and the maturity of the organization when it comes to conversation design. To make conversations more personalized, you can account for common scenarios like First Time vs Returning Users, Existing Customer vs Anonymous User, Different Product SKU Experiences, Online vs Offline Availability for Live Human handoff, and Business Logic Scenarios. By considering these personas and designing a conversational experience around them, businesses can create a more robust user experience that increases containment of conversations, answers more questions, and creates more satisfied users.

A study highlights the importance of effective conversation design in chatbots, particularly when dealing with unexpected user responses. The ideal approach is to provide users with options based on their previous utterances, rather than simply apologizing or admitting mistakes. This can help redirect the conversation and get users back on track. By incorporating features like paraphrasing, giving options, and using socially appropriate repair tactics, chatbots can create a more engaging and user-friendly experience. The goal is to build a network of "happy paths" that lead to successful interactions, rather than focusing solely on apologizing for mistakes.