TLDR: A study analyzed 120,000 YouTube comments during the June 2025 Iran-Israel conflict using AI (BERT, XLM-RoBERTa) and critical analysis. It found a strong overrepresentation of pro-Palestinian and anti-US/Israel discourses, with Iran emerging as a central, though often criminalized, actor in digital conversations, suggesting a shift from traditional media narratives and highlighting algorithmic biases.
A recent groundbreaking study delves into the complex digital landscape of the Iran-Israel conflict, specifically analyzing 120,000 comments posted on YouTube during a critical 12-day period in June 2025. Titled “AI Applied to the Analysis of the Irán-Israel Conflict: Mapping Discourses on YouTube,” this research, led by Alvaro Vallejo RamÃrez, offers a unique perspective on how online conversations are shaped by both user sentiment and underlying algorithmic biases.
Understanding the Digital Battlefield
The study’s primary goal was to identify the prevailing discursive positions regarding the key actors involved – Iran, Israel, the United States, and Palestine – and to understand the influence of media and algorithmic biases on these digital exchanges. The conflict itself, which erupted on June 13, 2025, with an Israeli strike on Iranian infrastructure and a swift Iranian retaliation, marked an unprecedented escalation in the Middle East, quickly becoming a focal point for global digital discourse.
A Mixed-Methods Approach
To achieve its objectives, the research adopted a sophisticated mixed-methods design. The quantitative phase leveraged advanced natural language processing (NLP) techniques and machine learning models, specifically BERT and XLM-RoBERTa, to classify the vast number of comments into ten distinct categories. These categories ranged from “Pro-Israel” and “Pro-Palestina” to “Anti-US” and “Comments without a clear stance.”
The qualitative phase complemented this by conducting a critical analysis of the broader media context and ideological narratives. This involved manual annotation and supervised training, ensuring that the statistical robustness of the AI models was integrated with a deep contextual understanding of the geopolitical and cultural nuances at play. The researchers collected comments from 3,000 YouTube videos in Spanish, English, and French, focusing on content published two weeks before and two weeks after the 12-day conflict.
Key Findings: A Shift in Digital Narratives
The study’s results are particularly revealing. It found a significant overrepresentation of pro-Palestinian and anti-United States/Israel discourses within the YouTube comments. Conversely, pro-United States and anti-Palestinian positions were found to be marginal. This suggests a strong alignment of the analyzed public with anti-hegemonic causes and narratives of resistance.
One of the most striking findings was the emergence of Iran as a central actor in the digital conversation. Traditionally less visible in global media, Iran’s presence surged dramatically during the conflict, both in terms of support and opposition. This indicates a potential narrative shift away from previous hegemonic frameworks, where Iran was often sidelined or portrayed negatively. However, the study also notes that this increased visibility often came with a “criminalizing” narrative.
The research also confirmed the pervasive influence of algorithmic biases. These biases, inherent in platforms like YouTube, were found to amplify certain discourses while limiting others, often reinforcing dominant Western perspectives. Despite this, the study observed a decline in support for Israel and the United States as the conflict progressed, challenging traditional pro-Western narratives.
The Power of Likes: Reflecting Public Sentiment
Analyzing the number of “likes” on comments provided further insights. Comments supporting Palestine and Iran received significantly more interactions compared to those favoring Israel or the United States. The researchers interpret this not just as a shift in digital public opinion, but as a reflection of growing sensitivity among the digital audience to the structural violence against the Palestinian people, particularly in light of events often described as a “genocide.”
Dichotomous View: Iran vs. Israel
When comments were grouped into two broad blocks – “Iran” (Pro-Iran, Pro-Palestina, Anti-Israel, Anti-US) and “Israel” (Pro-Israel, Pro-US, Anti-Palestina, Anti-Iran) – the “Iran” block accumulated significantly more comments (15,218) than the “Israel” block (9,291) during the active conflict period. This quantitative difference suggests that the digital agenda was not solely dominated by official Western discourse, but became a space where narratives aligned with Iran and its allies gained greater resonance.
Also Read:
- Unpacking AI’s Political Compass: How Large Language Models Reflect Democratic and Autocratic Biases
- Unveiling Self-Preference: How Large Language Models Develop Human-Like Bias
Conclusion: AI Illuminates Digital Asymmetries
The study concludes that the digital discourse on YouTube during the Iran-Israel conflict reflects a clear transformation in the visibility of geopolitical actors. While algorithmic and media agendas initially favor Western powers, the eruption of alternative narratives in digital spaces can challenge this hegemony, especially during intense conflicts. The research underscores the critical role of AI and computational analysis in understanding these complex dynamics and the often-overlooked narrative disputes in international conflicts. For more detailed insights, you can access the full research paper here.
