IRL - In Real Life

2024

Machine Learning

Béatrice Lartigue - Lab212

In Real Life is a reference to the acronym IRL, which appeared in the early 2000s, as opposed to IVL - In Virtual Life, which refers to our interactions on the Internet and social networks.

In Real Life questions our identity as individuals within a social group, in the light of new technologies: What are the impacts of the development of artificial intelligence on our perception of our identity, and on our perception of the world around us?

The project based on Machine Learning draws on a corpus of Renaissance paintings, focusing on portraits of women. The images (generated using a text-to-image machine learning model) are based on the codes of classical painting, in terms of subject, composition, colour palette etc. However, on closer inspection, some disturbing details emerge, and physical anomalies are revealed.

The tools used to generate images using AI are constantly evolving, but to date they are still "imperfect" and "naturally" produce anomalies and incongruous results that raise questions about the nature of the images produced and their status.

Through this project I want to question both the representation of women in society in general and in the art world in particular, and the impact of new technologies on these representations. Some of their challenges are similar: the question of representativeness, with regard to certain cognitive biases linked to social class, ethnic origins, gender, age, physical characteristics...

The corpus of images that forms the databases on which generative AI algorithms are trained reproduces and amplifies the biases already at work, which is why it is essential to make it a subject of research, and to expose the issues and potentials at work to the general public. — Over and above the question of gender, whose boundaries are becoming less binary and more open to porosity over time, with In Real Life I want to question female representation and what it conveys. In particular, the word "woman" means "one who breastfeeds", but it also conveys the idea of generating and creating.

The notion of "reality" is also important: what makes a situation real or not? The perception of reality is a process that is both complex and endlessly reactivated. Reality is an ambiguous stimulus, and what makes us unique is the way in which we stabilise this ambiguity.

Through these portraits of women, I question the sometimes romanticised or idealised vision that emerges, just as through the sharing of our favourite moments on social networks. We offer a dreamlike vision of our daily lives, showing just one facet of a complex whole.

Finally, through this project and my work in general, as a woman working in the fields of art and new media, I hope to open up an area of discussion on these key issues.