Gender bias in tech

“I am he or she on my Google account?” – Gender bias in technology

I have been employed in a consulting company for more than two years, and my work communication on a daily basis is in Croatian and English. I sent countless emails in English and Croatian, and what started to frustrate me was the auto-correction of words into the masculine gender. In addition to that, I started to ask how Google Translate or my Gmail Account automatically offers male translation. By translating between cultures in this way, technology creates a metaspace that perpetuates gender insensitivity and male domination. Translation is not just a linguistic but a sociological endeavor as well, thus making it automatic or AI presenta a huge challenge in making not just unwanted mistakes but severe impacts in reproducing gender bias. In English and Croatian, there are mostly gender-biased words that predominantly include masculine terms. In general, the English language alone does not distinguish between a woman and a man through certain nouns, thus reproducing sexism in language. While on the other hand, the Croatian language primarily uses masculine nouns, as well as verbs that refer to persons in the masculine gender. Sexism in the Croatian language is most evident in the names of professions and occupations where the occupation is determined by the social construction of gender roles. For this reason, we have a “director” and “expert” in the masculine gender, and “the nurse” in the feminine gender.

But am I, according to the Croatian language, a “stručnjak” (expert) or a “stručnjakinja” (expert/ess). In English it is “simple”, I am an “expert” (he). So, my Gmail system and Google Account remind me every day that I’m a man (or should I be)?

Right at the beginning, I want to emphasize that this is not another one of those feminist blogs – this is a blog about social responsibility and the need to raise awareness of social inequality that is reproduced on a daily basis in one way or another. It is necessary to critically approach technology and the impact it has precisely because of its immanent use today on the global level and impact on people, especially young ones. I will address the issue of the impact of technology in the reproduction of gender inequality and sexism by trying to give several arguments that point out the reproduction of gender bias through the technology we use every day and the “how-to tackle that” steps.

Technological world – is it more suitable for men or for women?

Technology is seen as the “engine of progress” of world development (Crewe and Harrison 1998, 30). According to that, for the past 300 years, societies have been judged as progressive or backward depending on their knowledge of (West) science and technology (Crewe and Harrison 1998). But what technology is? Technology can be defined as an object which does something that works or helps (Everts 1998, 5). Thus, technology has occupied our daily lives and thus forms the dominant social discourse. There are the key factors that determine the social impact of technology on the women and men who design, build, purchase, and use it. I am giving attention to all these elements here. So, I’m assuming you’re familiar with biases in the field of technology, especially in the workplace. Algorithms within technological systems are designed to automatically recognize words and give them a gender. So for the noun doctor, surgeon, and director, it is “normal” to belong to a man, and the adjective “beautiful” or the noun teacher or nurse, refers to a woman.

In general, gender bias has been unconsciously internalized to us since childhood through primary and secondary socialization. Social science explains to us how gender roles and gender inequality are reproduced in our daily lives. So that kind of gender stereotyping also applies to management positions in the business world where a larger number of men are present, so it is “normal” that a director is a man, that an expert is a man, that a scientist is a man. So it’s not “weird” why Google corrects all my nouns about leading business positions in the male gender when I am using croatian. Thus, in the context of technological platforms, gender bias in machine translation is one of the many research topics being eagerly explored by academics and industry stakeholders alike. Gender bias has been shown to reduce translation quality, particularly when the target language has grammatical gender. Algorithm bias can also directly exacerbate the lack of gender representation in tech. The algorithm that is used in technology and the virtual world (such as Google, social media, Amazon) had been trained on historical narratives, which were of course mostly male. The algorithm, therefore, assumed that male candidates were preferable.

What can help us – “how to” against gender bias in technology

So, tackling the gender bias problem is an important part of technological reproduction.  Even though in 2018 Google promised to take the first step to solve gender bias in Google Translate, the change is not visible in every language and cultural sphere (as Croatian for instance). Although we live in a world of technological revolution where technology enters every sphere of our lives, we still have a certain impact and responsibility in our daily lives without technology involved. We ourselves can resist gender bias and the reproduction of gender inequality through our daily actions and impact on our everyday environment. We start with our personal and social responsibility. By becoming aware of this problem (gender bias) and talking about it with people around us, we are already making a change and raising awareness about a particular problem.

  1. Identifying Bias: The first step towards fixing the biases is identifying them. As we know, the biases are so deeply embedded that it’s often unconscious and overlooked. Consciously seeking those biases and identifying them will help us figure out ways to address them.
  2. Raising awareness: When we become aware of a problem and identify it, it is important that we talk about it with others and make the problem visible. Raising awareness is one step in the process of positive change (positive impact).
  3. Building Structural Equity in the business core: Having Diversity, Equity & Inclusion efforts in organizations that encourage equity and promote inclusivity would be a great way to begin. Breaking the glass ceiling for women in leadership positions, not only women but other marginalized groups as well. Being transparent about opportunities, and encouraging diverse groups to pursue them will help build structural equity. 
  4. Impact (Business) Culture: Problem identification and awareness of our potential for impact is an important step in this process, and the involvement of different actors in change is another factor. Impact Business culture is important to set within the work team in order to recognize and try to reduce the reproduction of social inequalities (whether it is gender, age, or ethnicity…).

To conclude, when developing a new tech product, it is necessary to involve experts in social impact who will identify the needs of users, and create a heterogeneous user persona map in accordance with their user-pathway. The first stage for measuring effectiveness of the technology is user and usability testing, which can highlight those biases that may surface when using a certain technological solution. The status of identity is a sensitive issue in today’s virtual world. We talk about male and female, but what about transgender people? Automatically changing one’s identity characteristics such as gender through technology perpetuates certain social inequalities and discrimination, and thus violates the rights of users. An accidental mistake or an oversight through generic technology that is used every day can be a million times worse than intentional sexism, because they are much wider and affect people on a global level.

This blog is written by Danijela Paska, Impact Specialist and Analyst in Impact House, and edited by Tamara Puhovski, Impact House CEO.