Design Alexa Interactions to Change User Expectation towards Its Intelligence

Project background & Research problem

Many people perceive AI as this omnipotent black box that can solve all the problems. This unrealistic expectation undermines people’s appreciation of AI and often leads to frustrations when it inevitably makes mistakes. Voice-based agents, with human voices and embodiment, are particularly susceptible to such frustrations. This project investigated how alternative, non-human-like voice interactions could affect people’s expectations towards Alexa. We explored interactions that fundamentally shift the nature of the conversation between Alexa and its users, so as to change users and their expectations.

My work

I designed a voice agent with a pet's personality and its interaction with people on various scenarios. So users would lower their expectations but still had a pleasant experience.

Advisor

Prof. John Zimmerman
Qian Yang (PhD Candidate at CMU HCII)
Cole Gleason (PhD Candidate at CMU HCII)

Timeline

Jan - May, 2018