🤖 AI Summary
This study presents the first systematic evaluation of gender and cultural representation biases in mainstream text-to-image models (ImageFX, DALL-E 3, Grok) when generating occupational imagery for Saudi contexts. Using neutral occupation prompts, we generated 1,006 images; two Saudi native annotators performed double-blind labeling on gender, attire, setting, activity, and age, with discrepancies resolved by a senior researcher and supplemented by LLM-assisted analysis. Results reveal pervasive male bias—e.g., 96% male representation in DALL-E 3 outputs—as well as over-masculinization of technical and leadership roles and recurrent cultural misrepresentations in attire and environmental cues. We attribute these biases primarily to cultural homogeneity and societal stereotypes embedded in training data, rather than inherent model limitations. The work establishes the first Arabic-context-specific benchmark and methodological framework for assessing occupational representation fairness in generative AI, advancing equitable AI evaluation practices.
📝 Abstract
This study investigates the extent to which contemporary Text-to-Image artificial intelligence (AI) models perpetuate gender stereotypes and cultural inaccuracies when generating depictions of professionals in Saudi Arabia. We analyzed 1,006 images produced by ImageFX, DALL-E V3, and Grok for 56 diverse Saudi professions using neutral prompts. Two trained Saudi annotators evaluated each image on five dimensions: perceived gender, clothing and appearance, background and setting, activities and interactions, and age. A third senior researcher adjudicated whenever the two primary raters disagreed, yielding 10,100 individual judgements. The results reveal a strong gender imbalance, with ImageFX outputs being 85% male, Grok 86.6% male, and DALL-E V3 96% male, indicating that DALL-E V3 exhibited the strongest overall gender stereotyping. This imbalance was most evident in leadership and technical roles. Moreover, cultural inaccuracies in clothing, settings, and depicted activities were frequently observed across all three models. Counter-stereotypical images often arise from cultural misinterpretations rather than genuinely progressive portrayals. We conclude that current models mirror societal biases embedded in their training data, generated by humans, offering only a limited reflection of the Saudi labour market's gender dynamics and cultural nuances. These findings underscore the urgent need for more diverse training data, fairer algorithms, and culturally sensitive evaluation frameworks to ensure equitable and authentic visual outputs.