The Effects of AI-based Credibility Indicators on the Detection and Spread of Misinformation under Social Influence

Lu, Zhuoran; Li, Patrick; Wang, Weilong; Yin, Ming

doi:10.1145/3555562

Cited by 17 publications

(3 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This line of inquiry is important because it might provide a possible explanation for the mixed findings concerning the effects of fact-checking labels in that people's evaluation of labels could affect the accuracy evaluation of or engagement with posts containing misinformation. As fact-checking labels on social media are provided by various sources, ranging from institutions such as independent fact checkers (e.g., PolitiFact, Snopes) and news organizations 3 to general social media users to algorithms (Lu et al, 2022;Seo et al, 2019;Yaqub et al, 2020), we examined people's perception of the effectiveness of fact-checking labels based on four different sources: (a) third-party fact checkers, (b) news organizations, (c) algorithms, and (d) social media users (i.e., crowdsourcing or community labels). We asked participants to rate their perceived efficacy of each fact-checking label after showing them a visual example of how social media platforms label posts containing misleading or inaccurate information so that they could understand what we meant by fact-checking labels.…”

Section: Discussionmentioning

confidence: 99%

Journalistic interventions matter: Understanding how Americans perceive fact-checking labels

Jia,

Lee

2024

HKS Misinfo Review

View full text Add to dashboard Cite

While algorithms and crowdsourcing have been increasingly used to debunk or label misinformation on social media, such tasks might be most effective when performed by professional fact checkers or journalists. Drawing on a national survey (N = 1,003), we found that U.S. adults evaluated fact-checking labels created by professional fact-checkers as more effective than labels by algorithms and other users. News media labels were perceived as more effective than user labels but not statistically different from labels by fact checkers and algorithms. There was no significant difference between labels created by users and algorithms. These findings have implications for platforms and fact-checking practitioners, underscoring the importance of journalistic professionalism in fact-checking.

show abstract

Section: Discussionmentioning

confidence: 99%

Journalistic interventions matter: Understanding how Americans perceive fact-checking labels

Jia,

Lee

2024

HKS Misinfo Review

View full text Add to dashboard Cite

show abstract

“…A widely adopted debunking approach is to apply warning tags, labels, or indicators, during the misinformation presentation after fact-checking by professional organizations or artificial intelligence (AI). Empirical user studies reveal that those warnings are generally effective in reducing participants' belief in misinformation (Clayton et al 2020;Yaqub et al 2020;Jia et al 2022;Kreps and Kriner 2022;Lu et al 2022). Yet, the efficacy of the warnings can be impacted by factors such as warning specificity (e.g., general warnings introduce bias, reducing belief in real news), warning design (e.g., simple and precise warning language), the source of the warnings (e.g., fact checker and community), and extra fact-checking details.…”

Section: Misinformation and Its Correctionmentioning

confidence: 99%

Reliability Matters: Exploring the Effect of AI Explanations on Misinformation Detection with a Warning

Seo,

Lee,

Lee

et al. 2024

ICWSM

View full text Add to dashboard Cite

To mitigate misinformation on social media, platforms such as Facebook have offered warnings to users based on the detection results of AI systems. With the evolution of AI detection systems, efforts have been devoted to applying explainable AI (XAI) to further increase the transparency of AI decision-making. Nevertheless, few factors have been considered to understand the effectiveness of a warning with AI explanations in helping humans detect misinformation. In this study, we report the results of three online human-subject experiments (N = 2,692) investigating the framing effect and the impact of an AI system’s reliability on the effectiveness of AI warning with explanations. Our findings show that the framing effect is effective for participants’ misinformation detection, whereas the AI system’s reliability is critical for humans’ misinformation detection and participants’ trust in the AI system. However, adding the explanations can potentially increase participants’ suspicions on miss errors (i.e., false negatives) in the AI system. Furthermore, more trust is shown in the AI warning without explanations condition. We conclude by discussing the implications of our findings.

show abstract

“…However, they are costly, ethically challenging, require hiring fluent speakers for each language, and difficult to sufficiently scale to typical volumes of user-generated content. Purely automated solutions have remained on the horizon, as they struggle with evaluating facticity and have reduced impact in sharing behavior in the presence of social influence Lu et al (2022).…”

Section: Introductionmentioning

confidence: 99%

Limiting Factors in the Effectiveness of Crowd-Sourced Labeling for Combating Misinformation

Bak-Coleman¹

2023

Preprint

View full text Add to dashboard Cite

Crowd-sourced labeling has been proposed as a solution for reducing the spread of misinformation on social media. Although there is evidence that labeling misinformation can reduce engagement at an individual level, we lack insight into whether crowd-sourced labels can have similar impacts at scale. To estimate the efficacy of crowd-sourced labeling, we adapted a computational model of viral misinformation spread parameterized by a corpus of 12.9 million tweets from the 2020 US presidential election. Our simulations reveal that individual-level reductions in engagement from crowd-sourced labels did not directly translate into commensurate reductions in the spread of misinformation. This is due to a number of factors, including the time required to surface a helpful note, the proportion of tweets labeled, and the extent to which labels target highly-engaging content. Yet even if crowd-sourced labels are prompt and well-targeted, they are unlikely to be as effective as combining more traditional forms of content moderation. Overall, crowd-sourced labeling-based may be better used as a supplement to traditional content moderation rather than a standalone replacement.

show abstract

The Effects of AI-based Credibility Indicators on the Detection and Spread of Misinformation under Social Influence

Cited by 17 publications

References 37 publications

Journalistic interventions matter: Understanding how Americans perceive fact-checking labels

Journalistic interventions matter: Understanding how Americans perceive fact-checking labels

Reliability Matters: Exploring the Effect of AI Explanations on Misinformation Detection with a Warning

Limiting Factors in the Effectiveness of Crowd-Sourced Labeling for Combating Misinformation

Contact Info

Product

Resources

About