Keywords

1 Introduction

The male model prevails – the classic ideal of Man formulated by Protagoras, as “the measure of all things” [1] and the woman emerges as a deviant being, as different, as “other”, although the condition of contemporary woman is generally more favorable.

To speak of women as the main subject is, therefore, to question the male pattern that has been used to describe all of humanity. As Criado-Perez exposes in Invisible Women, masculinity and whiteness are assumed precisely because they do not need to be expressed. They are implied. They are unquestionable. They are the default. And that reality is inevitable for those whose identity needs to be mentioned, for those whose needs and perspectives are repeatedly forgotten [2].

According to the 2019 European Commission’s Report on equality between women and men in the EU [3], gender disparity is present in various social behaviours, which translates into the reinforcement of stereotypes, gender violence, lack of economic independence and disproportionate participation in various diverse areas. Women continue, in our cultural context, to play stereotyped and insulting roles, which devalue their status as specialists, whether in technical or scientific areas. In the political context, leadership positions are still mostly held by men, even with the implementation of quotas, Incidentally, according to data from the European Commission, men still outnumber women by at least 4 to 1 (i.e., <20% of women) [3] in technical or scientific fields.

Gender violence, moreover, is one of the greatest forms of oppression. Indeed, in the European Union, around half of all women say they have been victims of sexual harassment and one in three over the age of 15 have experienced physical and/or sexual violence. Further, in countries like Mexico, the number of femicides is not tracked in any tangible way [4]. In addition to femicide and sexual harassment, violence against women can also take the form of cyber harassment, honour-related violence, dating/domestic violence, and the practice of female genital mutilation [4].

The European Commission report [3] also points to the devaluation of mostly female occupations, such as positions related to assistance and education. When women occupy positions linked to the Scientific-Humanistic areas instead of Technological areas such as engineering, for instance, they often suffer a salary penalty for this. The European Commission’s report also points to a strong division, in the digital areas, related to “insensitive” working conditions, and proposes greater support and funding directed towards digital media, to achieve gender parity.

2 The Role of Design

If an intersectional perspective on the feminist movement allows us to explore the plurality of experiences—not only of gender, but also of other social and cultural aspects of individuals who do not see themselves in the dominant group—design practices must also recognize these ingrained power structures. In other words, it is necessary to promote a design practice that is cognizant of communities and individuals who are systematically socially and politically othered.

The author Luiza Prado [5] promotes an intersectional feminist perspective in design that offers a deep analysis about the impact of design objects in society. For Prado, it is pertinent to analyse how power structures and discrimination in the relationship between human beings and technology have been ignored. In this sense, her work proposes a design practice close to Critical and Speculative Design, developed by Anthony Dunne and Fiona Raby [6], but which, based on the observation of daily interactions with technology, allows for the role of these objects to be revealed within the complex systems of oppression of a capitalist, heteronormative, sexist, and racist society.

That is, design, by simultaneously addressing issues of race, class, or gender, and by challenging assumptions and preconceptions made about the role of various objects in everyday life, can produce a critical stance that encourages small changes toward a more equitable future.

3 The Absence of Female Representation in the Digital World

Statistically, the digital world is dominated by men, as women occupy only 21.5% of all jobs available in the digital context, according to the 2018 European Commission report entitled Women in the Digital Age [3]. However, the first digital computer, the ENIAC, was programmed by women in 1946, and for most of the 20th century, women were the dominant gender in programming [2].

According to a study published by the consulting company Accenture [7], only 68% of female university students chose to take coding and computing classes in 2017, in contrast to 83% of male students, which shows a strong discrepancy between the sexes. This disparity can be explained by the persistence of an unconscious bias related to the ability that each gender has, as well as the working conditions in the workplace.

These biases range from feeling a lack of freedom to present ideas in the same way as male colleagues, as well as the devaluation of programming and computer skills in relation to men’s skills—which discourages women from following a career in this sector. At the same time, the report [3] identified that the lack of diversity and representativeness, in a technological context increasingly marked by Big Data and algorithms, will imply that digital systems reflect only the values, information and ideas of their developers, who are, as a rule, male, middle class, privileged and predominantly white.

Although, currently, the recommendations of international agencies point toward the inclusion of more women in the digital sphere and the creation of an algorithm verification system, in addition to the devolpment of Artificial Intelligence systems to guarantee their neutrality, authors such as Schlesinger [8], Shaowen Bardzell [9] and Sasha Costanza-Chock [10] point to the difficulty of the field of Information and Communication Technologies in incorporating marginalized groups in its design process. In fact, it is possible to find situations of discrimination based on gender that could be avoided if more women (of different contexts) were part of the design team [2, 7].

Nowadays there is a plausible trend of trans people (specifically trans-feminine individuals) entering the fields of computer science. Research by Skye Kychenthal [34] may help address the social issues of gender difference in computer science with a broader and more detailed perspective, considering that gender difference issues exist across the gender spectrum.

4 Bias in Biometric Systems

Since the field of Artificial Intelligence is in such rapid development, it is essential to question what kind of information is provided to machine learning algorithms which help perpetuate and amplify stereotypes, based either on gender, race, age, and others. That is, how do Artificial Intelligence systems reproduce discrimination and cultural prejudices based on available data.

It is possible to conclude that “the machines are not just reflecting our prejudices. Sometimes they significantly amplify them” [2] because they learn from the information provided to them and thus directly reflect the values ​​of those who program them. For example, identifying faces is a simple task for humans [11], but for Artificial Intelligence it requires understanding the geometric structure (3D) and confronting it with external factors such as light and the position of the face.

Biometric systems work through two main processes: verification (or authentication) and recognition. The verification process determines whether a face exists, and the recognition system determines the identity of the image from a database. When facial recognition technology companies claim to have an identification rate above 95% [12], their verification is limited to their database sample.

Organizations and institutions have every interest in collecting information about the largest number of individuals, including an intention to supervise, control and punish the bodies of people and groups that hold less power in society, especially as these are seen as a “commodity” [13]. It is in this context that it is plausible to determine that the development of biometric technologies, which allow for the coding of certain physical aspects, can be an obstacle for marginalized communities, since they algorithmically define physical traits with which users might not identify, such as race or gender, or even the assumption of a person’s character, just by algorithmic observation of their characteristics.

4.1 The Case of Aspire Mirror

Joy Buolamwini, in 2016, realized that the facial recognition software she used in the university project “Aspire Mirror” could not recognize her face. The code, written in JavaScript, could not detect Buolamwini’s face, only that of her lighter-skinned colleagues. She noticed, when examining the database of the facial analysis technology at her disposal, that it was composed of 77.5% of male faces and 83.5% of white faces.

As such, the truth is that this supposed representative sample, from which the algorithm had to learned to verify and recognize different faces, was reduced to a small percentage of the world’s population: men and Caucasian people. This is further highlighted by the fact that the misrecognition rate of fair-skinned men is only 0.8%. [14].

In addition, through an intersectional analysis of Microsoft, IBM, and Face++ systems, the researcher, together with Timnit Gebru, observed that “darker-skinned women were up to forty-four times more likely to be misclassified than lighter-skinned males” [13]. It is possible, therefore, to confirm that social inequalities are also reflected in digital systems through the excessive representation of a very specific part of the population. Indeed, according to Buolamwini, these systems could only identify 10% of the world’s population.

Looking at biometric systems from an intersectional perspective and demanding transparency is exceedingly important to make these technologies less harmful to marginalized communities.

Thus, data scientists, programmers, and designers should be concerned both with developing bias-preventive mechanisms and with the effectiveness of the artifacts they develop, so that users are not directly or indirectly affected by an unfair status quo [13].

5 The Render Me Project

The project at the core of this paper, entitled Render Me deals with a satirical approach to the deterministic effects of biometric systems that reinforce gender stereotypes. It considers a speculative future where “skins” define the social construction of identity and only through these skins can the metaverse be explored.

The methodology applied to the project follows the research process developed by Anthony Dunne and Fiona Raby [6]—Critical and Speculative Design. This design methodology aims to reveal the values ​​and ideologies absorbed in digital systems and how they transform the relationship with human beings. This is especially relevant as Dunne and Raby specifically intend to take advantage of design as a tool for raising awareness of less recognized issues, allowing for the emancipation of human beings as consumers: “design can help to expand awareness about the consequences of our actions as citizens and consumers” [6]. As James Auger states, “speculative design proposals are essentially questioning tools” [15] and not proposals for new technological products.

5.1 Scenario

This project is intended to discuss gender from an intersectional perspective within a constructed society, postulated to have become completely conquered by technology and in which control and surveillance are no longer framed by a democratic system.

The digital economy has taken over all interactions and avatars are necessary for us to relate to the world (parasocial relationships). The idea of ​​human ceases to be that of “Man” espoused by Protagoras [1], to become that of the avatar. Offline identity has little impact and interactions are mostly conducted online.

Thus, a scenario is built where 3D masks (“augmented reality filters”) are needed for the avatars through which interaction in the metaverse is possible. These masks are digital skins that protect the physical identity of their users, both against data extraction and against surveillance and the capture of their real faces. The name of the brand, Render Me intends to exalt the connection between the concept of “render”—which, according to the Cambridge Dictionary [16], means “to translate words into another language,” or according to the Oxford dictionary, Lexicon [17], “to represent or portray artistically”—and the pronoun “Me”.

5.2 Digital Skins and the MBTI Model

For the development of these digital skins, a concept of persona was used, which comprises the notion of identity “for several individuals to inhabit” [18], based on a study of Carl Jung [19].

According to Jung’s psychology [20] the conception of persona is the version of the individual that is presented to the world, but in the context of the digital economy this idea has been characterized by the “sum of available data about an individual”—which results in their digital persona.

However, for the investigation of this project, the model developed by Isabel Myres and Katharine Briggs, “Myres-Briggs Type Indicator” (MBTI) [21], was chosen based on that of Carl Jung, as it is currently the most popular and widely used personality mapping method [22] in the context of appropriating behavioural surplus.

According to the MBTI, it is possible to analyse the behaviour and social identity of users based on four dichotomies: extroversion vs. introversion; sensing vs. intuition; thinking vs. feeling; judging vs. perceiving [23], which results in 16 different combinations/types. Each of these four dimensions concerns how the individual interacts, processes information, makes decisions and decides to live life, respectively.

We present a graphic with the MBTI and with the developed skins (marked with colours). It is our goal to continue this investigation and develop skins for the other personalities and genders (Fig. 1).

Fig. 1.
figure 1

Carl Jung Psychological Types, Myers-Briggs Indicator (MBTI), David Keirsey Temperament Sorter [30] as well as the masks.

Full size image

It is based on these studies that we intend to develop the “look and feel,” that is, an aesthetic of the digital skins for six of the personalities identified in the MBTI model. These skins would then function as a digital presence of the gender stereotypes associated with women and, at the same time, as “performance,” in the sense that they are controlled by the user.

The digital skins were developed through the Spark AR Studio platform, as it allows their circulation through the most frequently used social networks, Instagram in particular. For this reason, a moodboard was designed for each of the personas, which served as inspiration for the development of Augmented Reality filters.

The Helper Skin

The Helper focuses on highlighting the woman’s role as protector and caretaker of the physical and emotional well-being of men.

About the construction of this identity as an archetype, it is possible to trace, alongside the MBTI model, a more introverted and sensorial projection of the world and a more sentimental and perceptive way of making decisions (ISFJ)—that is, the type and personality of “Protector” [23]. As such, The Helper’s main qualities are sensitivity, cooperation, kindness, and empathy.

Fig. 2.
figure 2

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Helper skin.

Full size image

On a visual level, an effort was made to evoke both virtual assistants and the idea of ​​a “cyborg” by resorting to imagery projected in the Microsoft program Cortana. In particular, the filter shares the visual language of the videogame character Cortana, from Halo. Cortana is a hyperfeminine “cyborg”, represented by a “networked circuit full of currents of vibrating light” [24]. This is an idea also conveyed in The Helper’s skin through its representation in a hologram with “glitches” (errors). Additionally, pink was chosen because it is associated, in Western cultures, with feminine attributes, preserving the delicacy and sweetness linked to the figure of the woman (Fig. 2).

The Guardian Skin

The Guardian skin presents male stereotypes, such as qualities that enhance introversion and intuition, the predominance of rational thinking, a “Mastermind-type” personality, and preference for a style where the characteristic of judgment predominates (INTJ) [23]. Specifically, The Guardian presents a more assertive, cold, and competitive type of behaviour.

Fig. 3.
figure 3

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Guardian skin.

Full size image

As far as visual choices are concerned, The Guardian offers an association of geometric figures. There is a correlation between abstract figures and qualities associated with men, namely independence, roughness, and angularity. The qualities associated with women are, for example, sociability, softness, and roundness [25].

Given that The Guardian persona expresses a typically masculine archetype, the decision was made to represent the outline of the face in a bulk and superimpose it with angular geometric figures, typically associated with the male gender (Fig. 3).

The Madonna Skin

This skin is based on the imagery of the Virgin Mary and her adoration, as pure and virginal. It explores the binary and antagonistic understanding of female gender representation in digital systems, which reinforces the patriarchal system and restricts women’s autonomy. Freud’s Madonna-whore complex [24] consists of reductive visions produced by the male perspective, which end up being mirrored in digital systems.

A persona analogous to the artistic representation of Mary and the “Healer” personality-type of the MBTI model [23] was developed, with traits such as purity, affection, altruism, and warmth, and which assumes an introverted and intuitive functioning style—a more feeling and perceptive interpretation of the world (INFP).

Fig. 4.
figure 4

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Madonna skin.

Full size image

On a visual level, the skin appropriates the artistic representation of the figure of the Virgin Mary by the Renaissance painter Rafael Sanzio (1483–1520), Madonna with Child (1505). Likewise, it was decided to place a halo, in gold leaf, and to imprison its image in a frame in order to enhance the idea of ​​admiration and worship (Fig. 4).

The Hairesis Skin

Hairesis intends to represent the biblical figure of Eve and, at the same time, women in heretical communities of the Feudal period [26]. This personification of Eve presents herself as an antagonistic character to Maria. In this sense, it symbolizes the role of a seductive, confident, and vile woman, who is often represented in literature and cinema.

This persona has a “Performer” personality type (ESFP), that is, they are extroverted and confident in their sensory experiences [23], as they apprehend reality through their perception and emotional nature.

Fig. 5.
figure 5

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Hairesis skin.

Full size image

On a visual level, red was used to represent the idea of ​​seduction and conflict associated with the “femme fatale,” presented as an irresistible and dangerous figure with pronounced feminine traits (Fig. 5).

The Worker Skin

The Worker intends to emphasize the multiplicity of women’s tasks (professional, domestic, conciliatory, etc.). It is a persona with traits usually associated with women, such as a sense of responsibility, and a human, polite and cooperative side. Therefore, the construction of the archetype, based on the MBTI model of the “Counselor” personality (INFJ) highlights this introversion and the intuitive way one apprehends the world, as well as a cooperative and organized mode of operation [23].

Fig. 6.
figure 6

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Worker skin.

Full size image

Regarding the visual plan, The Worker features a small, coloured animation on a black and white background, covering the face, with the title of professional careers typically linked to the female gender, such as “caregiver”, “au pair”, “assistant”, “housework” and “nursing”, to recall its importance in the current cultural panorama (Fig. 6).

Since, according to the World Economic Forum, women take on extra work supporting the well-being, diversity, equity, and inclusion of their peers. But it's ‘invisible’ work—because companies aren't recognizing and rewarding [27].

The Manager Skin

The Manager intends to evoke a masculine archetype, as a way of appropriating the “sexist and misogynistic culture of the Silicon Valley tech community” [28]. That is, one that represents the desired qualities in men: competitiveness, leadership, and independence. This persona, in the context of this work, represents the annihilation of the intelligent and feminine woman.

The construction of this archetype, according to the MBTI Model, presents a “Field Marshall” personality type, that is, ENTJ, which translates into an extroverted way of interacting, an intuitive way of analysing the world, which values ​​logic and judgment in the way of living [23].

Fig. 7.
figure 7

Inspiration moodboard (images taken from Pinterest) and a screenshot from the Manager skin.

Full size image

The visual representation of this persona translates to the complete erasure of the female subject, through the construction of a figure represented by a contour line, hiding all the physical characteristics of the user and, thus, neutralizing their definition of social identity.

The decision to completely abdicate any visual symbol, in addition to the use of genderless tones: yellow and green, refers to the feminist perspective that “all human beings are equal” [29]. Choosing, therefore, to place this persona in a more neutral plane as possible (Fig. 7).

5.3 The Instagram Account and the Filters with the Skins

The Instagram account, which hosts the augmented reality filters, is part of the narrative and interactive logic and, therefore, a publication plan was developed to better clarify the concept of each of the six digital “skins” developed. These publications are in line with the graphic language developed for each of the filters, present the name of the “skin” and contain a small description (Fig. 8). The Instagram page can be visited through this link: https://www.instagram.com/rndr.me.

Fig. 8.
figure 8

Preview of the Instagram account feed of the “Render Me” platform.

Full size image

5.4 User Testing

To carry out the User Test, the “Single Ease Question” (SEQ) method was used, which allows determining the difficulty of performing one or more tasks using a 7-point scale [31]. This tool enables a metric analysis of the user experience and a greater correlation with a qualitative analysis, as it works for a variety of questions during the usability evaluation.

The SEQ makes it possible to measure the platform’s usability even from a distance—that is, without directly observing the user’s actions—and leaves room for an understanding of what exactly was the encountered difficulty.

During the User Test, it is important to understand whether the user can easily interact with the Instagram filters/skins, whether they encountered any problems, and their reaction/experience.

Participants

The evaluation and perception test of the users in relation to the filters/skins was carried out based on a questionnaire given to 12 people between the ages of 21 and 45. 41.7% of these were in the 21 to 24 age group (N = 5); 41.7% between 25 and 30 (N = 5); 8.3% between 36 and 40 years old (N = 1); and 8.3% between 41 and 45 years old (N = 1). 41.7% of the sampled respondents identified as being female (N = 5), another 41.7% as being male (N = 5), and 16.7% as another gender (other) (N = 2).

It is also acknowledged that 58.3% of the sample has a degree (N = 7); 33.3% completed a Master’s degree (N = 4); and 8.3% claim to have completed secondary education (N = 1). Most respondents (N = 5) work as a designer or similar (41.7%) but it was also possible to poll a consultant in ​​telecommunications, a student, a researcher, two project managers, a receptionist, and a shopkeeper in this survey sample.

The test was carried out through an online questionnaire where the following scenario was presented: “You are at home and decide to open the social network Instagram. You start watching Stories and one of your friends shares a video of himself wearing an augmented reality mask. You get curious and decide to click on the profile of the author of the mask. At that moment, you notice the link in the description and access the Rndr.me account.” The following tasks were requested: a) access https://www.instagram.com/rndr.me; b) read the descriptions of the skins in the Posts; c) select the AR Filters tab ; d) choose an augmented reality filter; e) select “try it on” and explore.

Results and Discussion

Regarding the analysis and interpretation of the Likert Scale responses, it is considered necessary to calculate the mean and standard deviation. To complete the test with this type of scale, each potential user marked their response in a spectrum of 7 degrees of agreement, from 1 (strongly disagree) to 7 (strongly agree), on the Likert Scale.

The following table presents the mean and standard deviation (STDEV) results of the responses obtained (Table 1).

Table 1. Mean and Standard Deviation (STDEV) results of the responses (Likert Scale).
Full size table

Given these results, according to the “Single Ease Question” method, most markers have a mean greater than 5.5, representing a score within the average on the SEQ scale [32]. This analysis shows that participants were able to easily interact with the skins and had a generally positive experience. The standard deviation (STDEV), in general, demonstrates that most of the values ​​that represent the responses in the Likert Scale of the test are not very homogeneous, since they are greater than 1.0.

6 Conclusion

From the analysis of the cultural meaning of gender, the analysis of biometric systems of facial recognition and the investigation of different perspectives on design practice, it was possible to develop a speculative artifact that communicates and questions the dominant culture, exposing established gender stereotypes.

It was based on this idea that a fictional brand was developed, under the name Render Me, which develops digital skins (augmented reality filters in Spark AR) that protect the physical identity of its users against data extraction. However, this is a satirical approach, as it explores how stereotypes associated with gender, race, ethnicity, and sexuality cross the physical world into digital systems. In this sense, an aesthetic was constructed for six different augmented reality filters, based on the gender stereotypes identified throughout this research article. This was also done in accordance with the MBTI model, which provides a behavioural dimension with cultural meaning imposed by the conception of gender roles.

The project proposed here intends to be an awareness tool regarding the perpetuation and amplification of gender stereotypes in digital systems, specifically in biometric systems.

In short, the Render Me project seeks to place the female subject as the main figure, abandoning the harmful notion of the man at the centre that served humanist desires, and inciting critical thinking about how technological objects reproduce long-formulated prejudices and ideals. Designers must remain agents of change, increasingly aware of these issues.