Generative Genders: AI Image-Making and the Challenge of Representational Bias

The notion of gender is expanding in western culture at a time when AI is increasing in dominance. Consequently, the boundaries of creativity and cultural influence are shifting and becoming more permeable. AI image-generating technologies have emerged as powerful tools capable of producing captivating visual content. However, a critical examination reveals an unintended consequence: the reinforcement of typical western gender norms within these generated images. This project aims to unpack this phenomenon, highlighting the intricate interplay between technology, culture and the social construction of gender.

In the realm of art and visual representation, AI-generated images have gained prominence for their ability to combine elements, styles and concepts in innovative ways. Yet, as their algorithms draw from vast datasets, they inherit biases prevalent in the source material. In a global context, the dominant influence of Western culture on technology and media amplifies the propagation of Western gender norms – and reaffirms narrow stereotypes of ‘Other’ cultures. With the simple prompt ‘what is gender?’ plugged into Midjourney and then translated into multiple languages and fed back into the image generator again, the consequences are seen in the perpetuation of binary gender roles, heteronormativity, idealised physical appearances and cultural stereotypes. This work aims to unpack the aesthetic bias of emergent AI image generation, highlighting the intricate interplay between technology, culture, race and the social construction of gender.