Comparative Study of Gender Representation in AI-Generated News Images: Cross-Cultural Evidence from Chinese and Western Media Platforms
The rapid adoption of text-to-image (T2I) tools in newsrooms raises pressing questions about the gendered visual conventions embedded in AI-generated imagery. This study investigates cross-cultural differences in gender representation through a controlled four-model comparative experiment. Two Chinese-origin models (ERNIE-ViLG, Tongyi Wanxiang) and two Western-origin models (DALL-E 3, Midjourney) generated 400 images across ten news-related occupations, coded along five dimensions of gender and visual presentation. Results reveal a cross-cultural gap in female representation, with Chinese-origin models producing female figures in 30.5% of cases versus 42.5% for Western-origin models, a divergence most pronounced in high-status occupations. Both ecosystems nevertheless shared stereotyped patterns in care-oriented roles and in the gendered encoding of visual authority through posture, attire, and age. These findings indicate that cross-cultural divergence operates primarily through the intensity of stereotyping rather than its presence, positioning model selection and prompt design as consequential cultural decisions for news organizations.
References
- Thomson, T. J., Thomas, R. J., & Matich, P. (2025). Generative visual AI in news organizations: Challenges, opportunities, perceptions, and policies. Digital Journalism, 13(10), 1693–1714.
- Simon, F. M. (2024). Artificial intelligence in the news: How AI retools, rationalizes, and reshapes journalism and the public arena. Tow Center for Digital Journalism, Columbia Journalism Review.
- Bianchi, F., Kalluri, P., Durmus, E., et al. (2023). Easily accessible text-to-image generation amplifies demographic stereotypes at large scale. In Proceedings of the 2023 ACM Conference on Fairness, Accountability, and Transparency (pp. 1493–1504).
- Eagly, A. H., Woo, W., & Diekman, A. B. (2012). Social role theory of sex differences and similarities: A current appraisal. In The developmental social psychology of gender (pp. 123–174). Psychology Press.
- Liu, B., Wang, L., Lyu, C., et al. (2023). On the cultural gap in text-to-image generation. arXiv preprint arXiv:2307.02971.
- Liu, Y., Lin, A., Peng, S., et al. (2026). Occupational gender bias in Chinese generative AI models: Cross-model evidence of stereotypical amplification and systematic underrepresentation. Systems, 14(3), 286.
- Paik, S., Bonna, S., Novozhilova, E., et al. (2023). The affective nature of AI-generated news images: Impact on visual journalism. In 2023 11th International Conference on Affective Computing and Intelligent Interaction (ACII) (pp. 1–8). IEEE.
- Naik, R., & Nushi, B. (2023). Social biases through the text-to-image generation lens. In Proceedings of the 2023 AAAI/ACM Conference on AI, Ethics, and Society (pp. 786–808).
- Chen, Y., Zhai, Y., & Sun, S. (2024). The gendered lens of AI: Examining news imagery across digital spaces. Journal of Computer-Mediated Communication, 29(1), zmad047. https://doi.org/10.1093/jcmc/zmad047
- Girrbach, L., Alaniz, S., Smith, G., et al. (2025). A large scale analysis of gender biases in text-to-image generative models. arXiv preprint arXiv:2503.23398.
- Sun, L., Wei, M., Sun, Y., et al. (2024). Smiling women pitching down: Auditing representational and presentational gender biases in image-generative AI. Journal of Computer-Mediated Communication, 29(1), zmad045. https://doi.org/10.1093/jcmc/zmad045
- Ananthram, A., Stengel-Eskin, E., Bansal, M., et al. (2025). See it from my perspective: How language affects cultural bias in image understanding. In International Conference on Learning Representations (pp. 55615–55636).
- Tao, Y., Viberg, O., Baker, R. S., et al. (2024). Cultural bias and cultural alignment of large language models. PNAS Nexus, 3(9), pgae346. https://doi.org/10.1093/pnasnexus/pgae346