ABSTRACT
Artificial intelligence (AI) is everywhere. It has proved very useful in areas such as domestic robots, extended virtual reality, assistive technology, medical advancements, and video games, to name but a few. In the field of machine translation, AI reproduces human behavior by processing extensive data which, inadvertently, encapsulates gender bias. As a result, the construction of gender identities is asymmetrical and favors the masculine gender. This poses risks, including translation errors with potential legal consequences that hinder the intended legal and social progress envisaged in the laws. In order to illustrate how gender bias in machine translation affects the legal domain, this chapter first explains how training data and algorithms contribute to current conceptualizations of gender bias in AI and in machine translation. After an examination of gender representation imbalances in machine translation models, the chapter discusses some key examples that address masculine as the default translation in the language pair English–Spanish in four machine translation engines. The chapter closes with a call for interdisciplinary research as a means to addressing the challenge that MT poses in the legal domain and, based on the impact of the law in our daily lives, in society at large.
