AI-Generated Content Labeling in Political Communication: A Comprehensive Review

In today’s digital age, misinformation is spreading rapidly through social media platforms, and it can be challenging to detect and correct false information. Recent studies have shown that automated AI tools are being developed to identify and remove false information from online sources, but these approaches face several pitfalls.
Firstly, the concept of misinformation is complex and multifaceted, encompassing a range of possible meanings, intentions, and interpretations. It’s not just about factual accuracy but also includes semantics, hidden meanings, and interpretations. Therefore, relying solely on automated content analysis approaches may not be sufficient in detecting false information.
Secondly, current AI-based methods focus either on the content or source of information, neglecting other crucial factors such as the intentions of content sponsors. This can lead to misidentifying false information as accurate or vice versa. Moreover, these approaches often rely on one-dimensional conceptualizations, which may not capture the nuances of more complex cases like conspiracy theories.
To overcome these limitations, scholars suggest improving information quality by countering false narratives through inoculation and pre-bunking (Lewandowsky & Cook, 2020). They have also highlighted consistent psychological factors that underpin susceptibility to false narratives, such as the lack of analytical thinking and reliance on intuition.
In conclusion, while automated AI tools show promise in detecting misinformation, they are not without limitations. A comprehensive approach is needed to address these challenges by considering the complexity of the phenomenon, the role of content sponsors’ intentions, and the need for improving information quality through countering false narratives and promoting digital literacy. By doing so, we can create a more informed society capable of critically evaluating online information and making informed decisions.