Rutgers University researchers have discovered a significant flaw in the credibility evaluation process of algorithms that detect “fake news”. The algorithms rely on a credibility score for the “source” of the article rather than assessing the credibility of each individual article, according to the researchers.
Flaw in Current Algorithms
The researchers found that the use of source-level labels for credibility is an unreliable method, with article-level labels matching only 51% of the time. The labeling process has significant implications for tasks such as the creation of robust fake news detectors and audits on fairness across the political spectrum.
To address this problem, the study offers a new dataset of journalistic quality individually labeled articles and an approach for misinformation detection and fairness audits. The authors presented their paper at the 15th Association for Computing Machinery Web Science Conference 2023, held from April 30-May 1 in Austin, Texas.
Importance of the Study
The study highlights the need for more nuanced and reliable methods of detecting misinformation in online news and provides valuable resources for future research in this area. The authors expressed that validating online news and preventing the spread of misinformation is critical for ensuring trustworthy online environments and protecting democracy. Their work aims to increase public confidence in misinformation detection practices and subsequent corrections by ensuring the validity and fairness of results.
The study conducted by Rutgers University researchers indicates that the current algorithms used to detect misinformation in online articles are flawed. The researchers proposed a new dataset of individually labeled articles and an approach for detecting misinformation and fairness audits. The findings of this study highlighted the need for more nuanced and reliable methods of detecting misinformation in online news and provide valuable resources for future research in this area.