9 August 2023 in Journal of Communication
The rapid deployment of artificial intelligence (AI) technology has enabled AI agents to take on various roles as communicators, such as virtual assistants, robot journalists, and AI doctors. This study meta-analyzed 121 randomized experimental studies (N = 53,977) that compared the effects of AI and human agency on persuasion outcomes, including perceptions, attitudes, intentions, and behaviors. The results showed that AI agents were as persuasive as humans in terms of overall persuasion outcomes. With regard to different types of outcomes, AI was less effective than humans at shaping behavioral intentions, but did not differ significantly from humans in eliciting perceptions, attitudes, or actual behaviors. Additionally, heterogeneous patterns were observed for different roles of AI communicators, directions of communication, experimental settings, and demographic segments. The implications of these findings for human–machine communication and persuasion in the era of AI are discussed.
3 July 2015 in Journal of Interactive Advertising
This study seeks to extend the emerging body and scope of research on consumers' attitudinal and behavioral responses to online consumer reviews by examining the role of message content characteristics. From this perceptive, this research broadens the understanding and importance of message characteristics to the persuasiveness of online consumer reviews. Specifically, this study investigates how consumers process online consumer reviews within the context of message characteristics, such as message type (attribute centric and benefit centric) and message valence (positive, negative, and neutral), and how this influences consumers' attitudes toward the review and product, perceived credibility of the reviewer, and purchase intention. Data indicate that message valence of online consumer reviews has a persuasive effect on consumers' attitudes toward the review and product, perceived source credibility, and purchase intention. Moreover, message valence interacts with message type, providing a more complex picture of product review effectiveness. Finally, data indicate that benefit-centric reviews result in higher recall than attribute-centric reviews.
14 February 2024 in Communication Research
The growing adoption of artificial intelligence in journalism has dramatically changed the way news is produced. Despite the recent proliferation of research on automated journalism, debate continues about how audiences perceive and evaluate news purportedly written by machines compared to the work of human authors. Based on a review of 30 experimental studies, this meta-analysis shows that machine authorship had a negative, albeit small, effect on credibility perceptions. Furthermore, machine authorship had a null effect on news evaluations, although this effect was significant and stronger (more negative) when (a) the news covered socio-political topics (vs. environmental topics) and (b) the actual source of the news articles was a machine (vs. a human). These findings are discussed in light of theoretical accounts of human–machine communication and practical implications for news media.
29 January 2020 in New Media & Society
Cyberbullying often happens in the presence of bystanders whose behaviors play a key role in changing dynamics of the situation. To examine the factors influencing cyberbystander likelihood of intervening in cyberbullying on social media, a 2 (degree of cyberbullying severity: high vs low) × 2 (level of interpersonal similarity: high vs low) between-subjects experiment was conducted (N = 132). Results indicated that cyberbystanders’ empathic responses depended on the severity of cyberbullying. The more empathy bystanders felt, the more likely they were to publicly and privately intervene. Moreover, the indirect effect of cyberbullying severity on willingness to intervene through empathy was more pronounced when the interpersonal similarity between victim and cyberbystander was low, rather than it was high.
8 July 2019 in Journal of Product & Brand Management
Purpose: In the context of celebrity endorsement, this study aims to demonstrate that the ways in which consumers adopt moral reasoning strategies (i.e. rationalization, decoupling and coupling) are largely dependent on the severity (i.e. high vs low) of celebrity transgressions and the degree to which they personally identify with the celebrity. Design/methodology/approach: A between-subjects online experiment (N = 144) with two conditions, representing high- and low-severity celebrity transgressions, was conducted. Participants’ attitudes toward the celebrity and endorsed brand, their purchase intention for the endorsed product and the degrees to which they identified with the celebrity and adopted the three types of moral reasoning strategies were assessed. Findings: The rationalization and decoupling strategies mediate the effects of highly negative information about a celebrity on consumer attitudes toward the celebrity and endorsed brand as well as on purchase intention for the endorsed product. In addition, consumers who identify strongly as fans of the celebrity in question are more likely to activate rationalization and decoupling strategies to process and evaluate transgressive behaviors than those with weaker fan identification. Originality/value: By exploring the ways in which moral reasoning and fan identification work in processing negative information, this study provides insights into the psychological process through which negative news coverage of a celebrity endorser influences consumer attitudes and purchase intention.
4 December 2020 in Digital Journalism
Studies have shown that uncivil comments under an online news article may result in biased perceptions of the news content, and explicit comment moderation has the potential to mitigate this adverse effect. Using an online experiment, the present study extends this line of research with the examination of how interface cues signalling different agents (human vs. machine) in moderating uncivil comments affect a reader’s judgment of the news and how prior belief in machine heuristic moderates such effects. The results indicated that perceptions of news bias were attenuated when uncivil comments were moderated by a machine (as opposed to a human) agent, which subsequently engendered greater perceived credibility of the news story. Additionally, such indirect effects were more prominent among readers who strongly believed that machine operations are generally accurate and reliable than those with a weaker prior belief in this rule of thumb.