SK Telecom’s NUGU AI speaker. (image: SK Telecom)
SEOUL, Oct. 30 (Korea Bizwire) — The right timing for talking is more difficult than it appears.
This is because the context can differ depending on the mood of counterparts or the surrounding environment.
A South Korean research team has developed a new artificial intelligence (AI) technology that enables AI assistants to know when it is an appropriate time to talk.
The Korea Advanced Institute of Science & Technology (KAIST) announced on Wednesday that it found contextual factors in user circumstances that determine the optimal timing for smart speakers to talk.
After developing smart speakers for experimental use, the research team installed them in the rooms of 40 students who live in campus dormitories and collected as many as 3,500 data records over a period of a week.
As part of the experiment, the smart speakers asked if it was okay to talk when the user’s motion was detected or when a certain amount of time had passed. In response, the students answered and explained why.
With this experiment, the circumstance contextual factors that determine when the smart speakers should start or stop talking were classified into personal factors, social factors and motion factors.
Personal factors, for instance, can be divided into the degree of concentration on activities, the degree of urgency and busyness, the state of mental and physical condition, and the state of being able to talk or listen.
The research team expects the new technology to contribute to developing new AI assistants that can offer preemptive services such as reminding users to take medication or modifying schedules based on the actual progress of planned tasks.
Kevin Lee (kevinlee@koreabizwire.com)
