인공지능의 목소리, 규제할 수 있을까 ―젠더 편향 해소를 위한 제안―

This article investigates the effectiveness of proposed solutions addressing gender bias in AI and evaluates their appropriateness. Globa tech companies have developed and marketed a digital assistant with a female voice, targeting male consumers. However, the issue lies in feminized digital assistants not merely perpetuating existing gender inequalities but exacerbating them. These digital assistant voices reinforce gender stereotypes, portraying women as kind, docile, and submissive, while fostering false attitudes about sexual harassment. This inclination towards female digital assistants stems from the overwhelmingly male workforce in tech companies, highlighting a severe gender divide within AI technology compared to other professional fields. Numerous international and national initiatives have aimed to address this issue. Proposals have emerged to eliminate defaulting to female voices, explore gender-neutral digital assistants, and collect gender-sensitive data. Efforts have also focused on designing algorithms to mitigate gender inequalities reinforced by biases in digital assistants. In the education sector, policies and legislation bolster girls’ entry into science and technology, addressing gender imbalances within the tech field. However, these measures struggle to rectify deeply entrenched gender inequalities. Specifically, existing measures tackling gender bias in AI have limitations. Some introduce a male option to avoid defaulting to a female voice but fail to confront gender binary stereotypes or consider diverse gender identities. The creation of gender-neutral digital assistants within male-dominated tech companies raises questions. Additionally, discussions about the individuals responsible for designing or auditing gender-sensitive data and algorithms are absent, hampering their collection and programming. Despite policies and legislation aimed at enhancing women’s participation in technology, a significant under-representation of women persists within the field of AI technology. To effectively address gender bias in AI, it is required to devise solutions that consider the social and political contexts of gendered power. In a predominantly male-dominated society where male perspectives and experiences prevail, awareness of gender bias in AI might be lacking. Raising social consciousness about gender bias in AI necessitates educational initiatives and dedicated research. Specific legal support should also be provided to help address gender bias in AI. Developing gender-sensitive data collection methods and algorithmic designs requires a legal support institute focused on identifying and proposing solutions to gender-related issues. Such endeavors require training experts at the intersection of technology and gender through affirmative action.