One of the significant challenges facing researchers in artificial intelligence (AI) development is ensuring objectivity amid the rapid and ever-increasing flow of information online. With numerous tools now available to disseminate diverse data and information, it has become increasingly difficult for audiences to distinguish between truth and propaganda on one hand and between objective journalism and biased framing on the other.
This has underscored the growing importance of technologies designed to analyze, detect, and filter vast amounts of data. These tools aim to curb the spread of misinformation, combat rumors and fake news, and make the internet a safer space for sharing and accessing accurate information.
Fake news is defined as media content created and published with the intent to mislead or manipulate public opinion, often for political, economic, or social purposes. The methods for creating fake news range from simple manipulation of facts to sophisticated techniques like deepfakes, further complicating efforts to identify them.
In Abu Dhabi, Dr. Preslav Nakov, a professor and chair of Natural Language Processing at Mohamed bin Zayed University of Artificial Intelligence, is leading innovative efforts to develop AI technologies, particularly in analyzing the methods used in media to influence public opinion.
Among his most notable contributions is the development of FRAPPE, an interactive tool for global news analysis. FRAPPE provides comprehensive insights into the persuasive and rhetorical techniques employed in news articles, enabling users to gain a deeper understanding of diverse media contexts. According to Nakov, FRAPPE helps users identify how news is framed and presented in different countries, offering a clearer perspective on divergent media narratives.
FRAPPE’s capabilities also extend to analyzing media framing methods. Nakov explains that the tool allows users to compare how different media outlets address specific issues. For instance, one outlet in a particular country might emphasize the economic implications of climate change, while another focuses on its political or social dimensions.
AI is the cornerstone of FRAPPE’s functionality, enabling the tool to analyze complex linguistic patterns that influence readers’ opinions.
In a discussion with Asharq Al-Awsat, Nakov highlighted the tool’s capabilities, noting that AI in FRAPPE is fundamental to analyzing, classifying, and detecting intricate linguistic patterns that shape readers’ perceptions and emotions. He explained that the application uses AI to identify propaganda and persuasion techniques such as insults, fear-based language, bullying, exaggeration, and repetition. The system has been trained to recognize 23 subtle techniques often embedded in real-world media content.
Ensuring objectivity and reducing bias are among the main challenges in developing AI tools like FRAPPE. Nakov explains that FRAPPE focuses on analyzing the language used in articles rather than evaluating their accuracy or political stance.
To date, FRAPPE has analyzed over two million articles on topics such as the Russia-Ukraine war and climate change. The tool currently supports content analysis in 100 languages, with plans to expand its capabilities to additional languages and enhance the accuracy of its analyses, further strengthening its ability to comprehend global media patterns.
