Trained with profile pictures collected from dating apps for gay men, the artist and the artificial intelligence (AI) collaborate to generate a series of images.
Based on generative adversarial network, the AI is supposed to learn the general style of the profile pictures, and generate “unreal” images of such style. However, due to a lack of number of dataset for training the AI, in other words, a failure to study, the images it generates look rather like Frankensteinish torsos.
An experiment questioning what is photography, this work uses Machine Learning (ML) to simulate how images are produced, stored, chosen and reproduced in the process of online dating. Nearly a case study of modern social networking, it shows how the notion of beauty has shaped our social behaviours thus influencing our self-cognition.
How similar the images look implies people’s perception of ideal lover and the representation of masculinity through social media, The distortions, because of the interference from ML metaphorize the influence of the capitalist mechanism and the society of the spectacle.
Even distorted and surrealistic, the generated images preserve the sensuality coming from the original footage. I print the images on different forms and materials, no matter they are commonly used in traditional art photography or not, questioning what photography is.
Liao, J., & Cassinelli, A. (2022). Too Good to Be True: An Art Project Using Machine Learning in the Context of Online Dating. Journal of Global Pop Cultures, (1).[link]
Hong Kong International Photo Festival: Too Good to Be True[Exhibition]. (2021). Jockey Club Creative Arts Centre, Hong Kong. [link]