Title:
Study On The Impact Of AI Real-Time Emotion Suggestions On Users’ Social Experience In Online Conferences

Thumbnail Image
Author(s)
Li, Xingyu
Authors
Advisor(s)
Howell, Noura
Leigh, Sang-won
Harmon, Stephen W.
Advisor(s)
Editor(s)
Associated Organization(s)
Organizational Unit
Organizational Unit
Series
Supplementary to
Abstract
Advances in artificial intelligence (AI) have provided great opportunities for participating in social communication and carrying out empathic conversations through collaborating with people. However, Emotion AI with its currently low accuracy does not work well in social scenarios as the effect must be studied in the context it is being expressed, and observed emotion signals should not replace internally reported effects for affective computing applications. Affective Computing is a growing field making rapid strides toward improving Emotion AI. This project speculates, perhaps there will be a high-accuracy Emotion AI with an improved algorithm fitting complex social scenarios in the near future. To understand the effects of Emotion AI (artificial intelligence that learns to interpret and respond to human emotions) on people's interpersonal confidence, social experience, and emotion recognition ability, we designed a prototype, Adverb, providing real-time detailed emotion types in the online meeting software. This thesis study aims to investigate the following questions: How does emotional AI communicate with humans? What is the impact of AI information displays' advice on users' social intelligence and interpersonal confidence? What is the impact of AI information displays advice on users' emotion recognition ability? We conducted the user study employing both qualitative and quantitative methods, and thirty participants completed the survey and interview. This thesis is concluded with a discussion of the limitations and future work of human-AI collaboration. Our findings are as follows: (1) Displaying the categories of emotions from AI, overall affects users' social intelligence, however, there is no significant effect on social confidence. (2) Users wanted the AI to show its suggestions and explain its principles only when the users wanted it to do so. (3) In different social scenarios, different communication methods and information levels can be applied. Based on these findings, we discuss implications for user interfaces where users can collaborate with AI.
Sponsor
Date Issued
2022-12-19
Extent
Resource Type
Text
Resource Subtype
Thesis
Rights Statement
Rights URI