FANTASY FOOTBALL - HOW JACQUI MCASSEY EXPLORED FEMALE FOOTBALL FANS THROUGH ARTIFICIAL INTELLIGENCE

On the 6th of February, 2023, I posted an image to my Instagram page of an older female Inter Milan fan standing beside the San Siro stadium. The caption gave no further context to the image but did prompt someone to message me asking what project I was working on in Italy. There was nothing unusual about me posting about female fans; I had been documenting women fans through the GIRLFANS project since 2013, and through a research study in 2018, I preserved the untold football stories and photographs of older life-long fans—work that first featured in MUNDIAL. 

This, however, was different. The supporter did not exist. 

She was a textual manifestation of a football fan created using Artificial Intelligence with the words: 90-year-old woman wearing an Inter Milan football shirt, standing by the brutalist San Siro stadium in Milan—v4**-.

A few months earlier, in my day job as a lecturer in fashion communication at Liverpool School of Art & Design, our final-year students had explored AI platforms, their image generation process, and the complex ethical challenges they create. I wanted to use it myself to form my own opinions and understand machine learning, so I chose Midjourney, which creates ‘beautified’ art through its server, Discord. The output generated is derived from a dataset, which is a collection of publicly available images that have been scraped from the web or art collections and archives, and an initial scroll through Midjourney’s stream of images showed that most of the illustrations and photographs were of young, Caucasian, wide-eyed women—clearly driven by the dataset and through its user prompts. 

"To test this, I saw an opportunity to continue making images of older football fans based on the real-life stories collected from a decade of research and imagine the fans I had yet to meet but who hopefully existed somewhere in the world."

Each image I made was generated by a prompt, a text-based description of the person. From an ethical viewpoint, I gave myself some parameters and rules for my image generation. I avoided adding existing images to the prompt. I avoided using the name of any photographer, designer, or artist, meaning the image was generated from Midjourney’s dataset only. I did name the football club and their kit sponsor and added some details about their environment. 

Each person was described as an older woman, and I included prompts to describe ethnicity, knowing that without this, the woman would probably be prettified, young, and pale-skinned. I generated multiple images in minutes, many revealing significant challenges and oddities, the signifiers of machine-made art. Anatomical details were skewed; women had seven-digit hands or faces with three eyes, though this has improved over the last twelve months with Midjourney v6.