The effect of artificial newscasters on cognitive and emotional responses to televised news

Xu, Xiaoyu

Permalink

https://hdl.handle.net/2142/117761 Copy

Description

Title

The effect of artificial newscasters on cognitive and emotional responses to televised news

Author(s)

Xu, Xiaoyu

Issue Date

2022-11-21

Director of Research (if dissertation) or Advisor (if thesis)

Wise, Kevin

Doctoral Committee Chair(s)

Wise, Kevin

Committee Member(s)

• Ball, Christopher
• Vargas, Patrick
• Yao, Mike

Department of Study

Inst of Communications Rsch

Discipline

Communications and Media

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

AI newscaster, human-likeness, human-computer interaction, automated journalism

Abstract

Machines are getting increasingly intelligent. Not only do they facilitate our daily life, but some of them have also engaged humans in communication processes. Of particular interest is the use of artificial newscasters, which engage audiences in a mass communication context. As a new phenomenon, little is known about their effects on the psychological processing of broadcast news. Two experiments were conducted to examine effects of artificial newscaster on viewers’ experience of broadcast news watching, specifically the cognitive resource allocation to and the encoding of news, as well as the subjective emotional evaluation of news. In both studies, news valence was examined as the moderator. The dynamic change of these effects over time with repeated exposure was investigated as well. Study 1 compared artificial newscasters with human newscasters with a 2 (Newscaster Type: Artificial vs. Human) x 2 (News Valence: Neutral vs. Unpleasant) x 5 (Message Repetitions) mixed design (N = 85). Subjects were randomly assigned to either an artificial newscaster or a human newscaster. For each subject, half of the trials were neutral news, the other half are unpleasant news. Secondary task response time indicated cognitive resource allocation; memory sensitivity (d’) reflected the encoding of news. Self-reported valence and arousal dimensions of Self-Assessment Manikin (SAM) were collected for each news trial to indicate the subjective emotional evaluation of the news content being presented. Both cognitive resource allocation and emotional response were explored over the course of both STRT probes and news exposure trials to assess how these responses change dynamically over time. Study 2 further explored the effect of the human-likeness of artificial newscasters with a 3 (Human-likeness of Artificial Newscaster: low, medium, high) x 2 (News Valence: Neutral vs. Unpleasant) x 5 (Message Repetitions) mixed design (N = 111). The measures for Study 2 were the same as for Study 1. Results of Study 1 showed that relative to the human newscaster, the presence of an ultra-realistic artificial newscaster elicited no difference in cognitive resource allocation to and encoding of news content. Nor was the subjective evaluation of valence and arousal of news different. Dynamic change of cognitive resource allocation as indexed by STRT suggested that viewers exposed to the artificial newscaster allocate cognitive resources differently from those exposed to the human newscaster: viewers exposed to the artificial newscaster allocated less resources to news watching at first, but after the first probe, they allocated more resources to news watching. Results of Study 2 showed that human-likeness of artificial newscasters had no effect on the cognitive processing of news. However, exposure to a least humanlike artificial newscaster induced viewers to perceive unpleasant news even more unpleasant than those exposed to a medium humanlike artificial newscaster. No significant difference was found between valence ratings between the medium humanlike artificial newscaster and the most humanlike artificial newscaster. The results are discussed in relation to the mind perception theory, categorical perception, and the Media Equation.

Graduation Semester

2022-12

Type of Resource

Thesis

Handle URL

https://hdl.handle.net/2142/117761

Copyright and License Information

Copyright 2022 Xiaoyu Xu

Owning Collections