Is Sexism Impeding the Global Influence of Japanese AI and Robotics?

In her July 18, 2018 op-ed in Asahi Shimbun, Japan’s progressive paper, Noriko Arai, a famous Japanese mathematician and artificial intelligence specialist, addresses the issue of gender and sexism in the Japanese robotics and AI industries.

Her piece was prompted by questions asked by non-Japanese researchers about sexist implications of emerging Japanese technologies such as female-embodied receptionist robots. In particular, she addresses concerns regarding the ways in which perceived sexist ideals built into AI-driven technology will prevent those technologies from being taken up in markets outside of Japan. Such robots are often presented as cutting-edge developments in the Japanese robotics industry.

“Japanese corporations cannot expect global influence unless they start to seriously reflect on the gender assumptions that they unconsciously build into AI technologies and robots”

For instance, Kyōei Sangyō’s RecepROID, launched in 2017, is a highly-gendered receptionist robot. Named Kyōei Ai-chan and sold for 9 million yen ($81,600 USD), she can work as a multilingual receptionist. Equipped with latest AI and Natural Language Processing technologies, Ai-chan can respond to guests’ questions in Japanese, English, Korean, and Chinese, and move her face to simulate a real human. Ai-chan’s voice is high-pitched, using polite and recognizably gendered female speech. Her name itself — with added “chan,” used to refer to little girls — is rather demeaning; Kyōei Ai is reduced to a repetitive, non-skilled role — in a sense, a “proper” position assigned to women in Japan.

Photo by Andy Kelly

Professor Arai also discusses a new beauty evaluation AI technology in Japan. While she does not specify which one, she is referring to technologies like the one invented by Professor Toshihiko Yamazaki at the University of Tokyo — Japan’s premier university. He named his team’s latest AI technology “the engineering of physical attraction” (miryoku no kōgaku). His lab members showed various photographs of female faces to male interviewees and asked them to give these women ratings between 1 and 5. Yamazaki then had his AI technology analyze that user data using a built-in machine learning system to the “attractiveness” of all female faces. Yamazaki has ideas for applications of this technology to a wide range of purposes, for instance a system where women upload their pre-make up photos and then have AI suggest what kind of make-up they should put on to look their best.

Will this technology be useful for men? Yamazaki does not believe so because “while, according to psychological research, we know that there are universal evaluative standards for women’s faces, for the attractiveness of men’s faces there are multiple evaluative standards.” Yamazaki, of course, does not question the sexist implications of this technology; he does not challenge the idea that women should be always evaluated by the heterosexual male gaze.

Japanese corporations, which have traditionally focused on industries like steel, autos and semiconductors, have not had to explicitly address the issue of gender bias or sexism. This is no longer the case in the age of AI. Arai provocatively argues that no matter how much investment Japanese corporations make in robotics/AI, they cannot expect to have global influence or relevance unless they start to seriously reflect on the gender assumptions that they unconsciously build into AI technologies and robots.

-Post by Tomo Sugimoto