Exploring the potential and challenges of voice interfaces
Conversational artificial intelligence (AI) refers to digital systems that use natural language to communicate with users. Thanks to the advances in deep learning and powerful language models, such as ChatGPT, a new generation of conversational AI systems has emerged that enables conversations with human-like qualities.
But how capable are these technologies? How well can they hear our voices and understand us? And how much information can companies decode from hours and hours of voice data without sitting down and listening to it all? In this 90-minute workshop Michael Shorter and Melanie Rickenmann used custom made provotypes that were informed by the Speculative Design Phase of the project. As participants interacted with these provotypes, they found themselves in the shoes of data collectors and gained an understanding of how much can be understood from our conversations. They also explored ChatGPT’s ability to analyze our conversations and experienced visualizations of this data. By the end of the workshop, participants gained valuable insights into the potential, as well as the limitations, of large language models.