Sara Shuman

Sara Shuman AI Conversational Designer Portfolio

AI Conversational Design

by

Designing AI-Powered Conversational Video ChatBots for USC

Striking the right balance between usefulness and approachability.

When it comes to designing chatbots, the goal is always to produce an efficient yet human-like interaction. The University of Southern California (USC) Digital Repository, which hosts an array of digital collections, recently approached me to build an innovative and immersive AI-Powered video-based interactive interview experience that uses conversational video – video that talks back.

The Challenge

As a conversational UI/UX chatbot designer, the challenge was to create an engaging and interactive experience for visitors to explore the treasure trove of information within the USC Digital Repository utilizing a conversational human touch. One of the primary goals was to make the experience memorable using speech-to-text inputs that gently guide users through the conversation so that they can easily discover the information about that person.

The Solution

After carefully considering the needs of the repository and user expectations, we developed a highly engaging chatbot called “Interactive Interviews“. With this feature, users can now dive into the repository’s vast collection of interviews and interact with them as if they were conducting the interview themselves.

Using cutting-edge technology, the chatbot demonstrates an excellent understanding of its subject matter, creating smooth, dialogue-like interactions between users and the vast content store available on the USC Digital Repository. To understand how this chatbot is designed and harnessed, try and ask Lora King’s interactive interview a question: Chat with Lora King Now

Key Design Elements

We focused on the following key design requirements to create a successful AI-Powered video-based chatbot:

  1. Context-awareness: The chatbot utilizes speech-to-text natural language inputs as it adapts its conversational flow based on user input and informational needs.
  2. Simplicity: We’ve designed the chatbot with an easy “click and hold” to ask a question UX, enabling users to approach it with very little hassle or learning curve.
  3. Engaging conversational flow: We crafted a natural and engaging conversational flow, using open questions and options for further inquiry, inviting users to explore the interview video responses.
  4. Error recovery: When the chatbot stumbles, misinterprets or fails to understand the user’s input, it provides options to redirect users without affecting the overall experience. Using a back-end SaaS tool, we were able to conduct supervised AI training on missed answers through chat logs.
  5. Efficient: Our conversational AI-powered video chatbot enables users to get their questions answered effortlessly without having to watch long video interviews or read multiple pages of interview transcripts.