On August 10th, Nima Zargham, a PhD candidate from the University of Bremen, presented and hosted an informal presentation and discussion on his recent research about human-agent speech interaction.
Zargham focused his talk “Machine Talk: Speech and Human Interaction” on communication and AI-based programming issues when trying to have conversations with helpful programs like Siri, Cortana or Alexa. Using scenarios and storyboards, Zargham shared his recent study results in which participants chimed in on the usefulness and boundaries that should be set for agents in their homes. For instance, he found people generally did not want agents like a Google Home inserting themselves into a conversation between two people. Some participants found the idea of a proactively (without direct invitation) involved agent ruined the concept of conversationally intimacy in their homes. The balance between being proactively helpful and proactively invasive is exactly the balance that Zargham aims to find with Human-Speech interaction.
His overview of user problems, agent problems, and conceptual ideas for voice-based game integration sparked a fruitful discussion which went on to helping he preparation of his user-study that took place shortly after. Stay up to date with Zargham’s research by following him on Twitter.