REPORTS

Report on the 6th BAIRAL Research Meeting for 2022 “Who is Ema?: Examining Representations of Race and Gender through Virtual Bodies Generated by AI”

Nozomi Ohtsuki (Research Assistant of the B’AI Global Forum)
Kyoko Takeuchi (Project Assistant Professor of the B’AI Global Forum)

・Date: Tuesday, October 18, 2022, 6:00-7:30 pm (JST)
・Venue: Zoom meeting
・Language: English
・Guest Speaker: Hanako Lowry (MA student, Graduate School of Interdisciplinary Information Studies, the University of Tokyo; Graduate Student Member, B’AI Global Forum)
・Moderator: Nozomi Ohtsuki (Research Assistant of the B’AI Global Forum)
(Click here for details on the event)

On October 18, 2022, the 6th “BAIRAL” meeting in 2022, a research group organized by the Research Assistants of the B’AI Global Forum, was held online. The guest speaker for this session was Hanako Lowry, a graduate student in the Master’s Program of the Graduate School of Interdisciplinary Information Studies at the University of Tokyo and a B’AI Global Forum graduate student member. Lowry spoke about the ongoing project, “ema.” “ema” is a design and art-based project that uses AI as a creative critique tool to examine how the intersection of contemporary structural inequalities and new technologies will create the future. In this workshop, Lowry first explained how the project was born and the process of designing “ema,” followed by a general discussion.

According to Lowry’s presentation, the birth of this project was enabled by her interest in the origins of the image of young Caucasian women used so frequently in the Japanese media as a symbol of mixed-race identity. In addition, an encounter with a student studying AI at a class exhibition at the University of Tokyo and a subsequent critical examination of the biases in AI led Lowry to begin the “ema” project.

In this project, Lowry generated images using AI trained on a dataset of existing representations of mixed-race bodies obtained from Google image searches. Lowry then showed the images, particularly the face of “ema,” to people at the exhibition to see how they would react. Lowry emphasized in the design that the face was very consciously created, as neither the dataset nor the process by which the AI-generated the image could ever be neutral. In the new “ema,” designed after feedback on this prototype, Lowry focused on the fashion and advertising industry, which promotes and commercializes aesthetics and mixed-race appearances to the public. Lowry made “ema’s” body appear as a model and examined the interaction among industry professionals, “ema,” and people accessing the website.

In the general discussion that followed, the participants discussed the possibilities for people to develop a sense of familiarity and attachment to “ema,” and how the second project, which is related to the industrial world of fashion and advertising, maintains the critical elements that Lowry had shown in the first project. In the second project, Lowry stated that she is updating her understanding of bias and media by first looking at how AI models are received in the industry, as well as presenting a picture of what will happen if certain practices continue to escalate at their current rate. In this regard, it was further pointed out that it would be interesting to create mythical creatures such as unicorns and mermaids to test the limits of the fashion industry. On a related topic, in the debate over virtual bodies, there is a position that attempts to create a body without gender and without race. However, Lowry pointed out that such bodies were actually expressed through the lens of the real world and did not escape the representation of gender and race. Lowry’s discussion will allow us to critically examine the virtual bodies that will be increasingly used by the fashion industry in the future, and we look forward to the ongoing new projects and exhibitions.