Manifestations of Bias in AI-Generated Images Depicting the Lebanese-Palestinian Conflict Against Israel – A Case Study of Chat GPT-4

Section: Research Paper
Published
Sep 1, 2025
Pages
366-382

Abstract

This study addresses a pivotal issue in the realm of media bias, focusing on how thetechnologically advanced or dominant party can exploit such technology to defend its position or principles. It also examines how this party can interfere with media platforms to alter facts. Since media bias becomes more evident during conflicts, the researcher chose the Lebanese-Palestinian conflict against Israel to highlight bias in the production of designed images. Considering the widespread use of ChatGPT as a prominent media tool, it served as the subject of this study. The researcher began by asking the program for key descriptive phrases to characterize the conflict and then utilized these phrases to produce 20 images across ten themes. The analysis of each image, divided between the conflicting parties, was carried out by counting repetitions with the assistance of the program itself.The analysis reveals that ChatGPT relied on several strategies that appear deliberate, such as emphasizing positive aspects, using cultural symbols, and downplaying negatives to present Israel as an ideal and superior entity. In contrast, the program marginalized the positive aspects of Palestinians and Lebanese, reflecting clear media bias in the produced images. The key manifestations of bias in these images include: Overemphasis on Israels positives and exaggeration of Palestinian and Lebanese negatives.Exploitation of cultural and religious symbols to Israels advantage, including the appropriation of Palestinian cultural and religious landmarks like Al-Aqsa Mosque to align with Israeli narratives.Visual bias in presenting Israel as a strong, stable state capable of protecting civilians and providing humanitarian aid, while showcasing its modern weaponry and advanced military capabilities. Negative depictions of Palestinians and Lebanese, such as portraying them as weakly armed, struggling, and embroiled in continuous conflict, thereby perpetuating a biased narrative against them.The findings also highlight the programs tendency to frame Al-Aqsa Mosque as a symbol defended by Israel, reflecting efforts to reframe Palestinian religious and cultural symbols to serve Israeli narratives. Additionally, there was a noticeable imbalance in portraying cultural and religious identities. While Palestinian and Lebanese cultural markers (such as the hijab and keffiyeh) were heavily emphasized, references to Jewish identity were less prominent, indicating bias in how the parties were visually represented. Key manifestations of bias were observed in:1. Visual effects (colors, clarity, lighting).2. Visual composition of elements.3. Appearance of individuals or groups.

Download this PDF file

Statistics

How to Cite

Fadhil Abo Sakhar, H. (2025). Manifestations of Bias in AI-Generated Images Depicting the Lebanese-Palestinian Conflict Against Israel – A Case Study of Chat GPT-4. Adab Al-Rafidayn, 55(102), 366–382. https://doi.org/10.33899/radab.2025.155991.2286