Sexual harassment and violence in public places is a human rights issue that interests me especially because there is still a lot of mystery surrounding it and not enough people are aware of the consequences.

By sexual harassment I am talking about women that have to face catcalling, rude and lewd gestures, following, masturbation, verbal sexual and/or physical aggression or even rape by strangers when they go out on the street.

Some of you may think that there is nothing wrong with women receiving the occasional catcall on the street or that honking at someone you find attractive while they are crossing the road can be understood as a “compliment”. However, all these seemingly banal actions sum up to create a constant paranoia in women’s minds, especially young women, which seriously harms confidence and self-estem and affects basic human rights such as physical integrity, freedom of movement, safety and equality. In fact, all these actions, from “less” serious to more serious, belong to the same spectrum of violence against women which prevents them from experiencing the same freedom to roam alone as any man experiences.

Many of you might think that women in “developed” countries do not face these problems but according to a 2014 European Union Fundamental Rights Agency report on violence against women based on a survey carried out on 42,000 women from all 28 member states:

• 37 % of European women have avoided walking through specific streets or areas out of fear of being physically or sexually assaulted.
• 4 out of 10 European women avoid going or remaining in public areas where there are no people out of fear of being physically or sexually assaulted.
• 1 out of 7 European women avoid leaving their home alone out of fear of being physically or sexually assaulted.
• 18 % of violent physical and/or sexual aggressions perpetrated by an unrelated person to a victim aged 15 years or older takes place in the street or in public places.

And this is just the violence against women that takes place by strangers on streets and public places, without taking into account the violence that can exist within couples or from perpetrators that the victim may know (bosses, male friends or colleagues, etc.).

The last couple of weeks I have been hearing about a set of devices, notably wearables and phone applications, to help ensure women’s safety.

This is the case of the necklace/watches designed by Leaf Wearables called the Safer Pro. Leaf Wearables is an Indian start-up that has recently won a prize for women’s safety. Their wearables are an accessible, inexpensive way for women to feel safer when they are alone in the street. They are equipped with a discrete button in order to alert a community of responders and share GPS location, even in low-signal areas, when users find themselves in unsafe or violent situations and need help.

Personal safety wearables are a new trend intimately related to women’s safety. A lot of them are disguised as jewelry, but you can also find safety shorts. The Safe Shorts are a controversial wearable born in Germany and designed by a victim of sexual assault. She was attacked while jogging in the forest by herself (a situation too well-known by many women). These shorts are equipped with a cord that sets off an alarm when someone tries to remove the garment by force.

Now, while I am all for technology improving women’s lives and I applaud these initiatives (many of them founded by women), I have to be honest and say that it disgusts me that the problem still so-widely exists, enough so that companies are designing products in order to fight against these issues. It seems that, often, it is the victim’s job to figure out ways of protecting herself and the market supplying this demand, rather than society’s job to fix what is wrong with the education of so many men that decide to assault women to the extent of 40 % of European women being scared to be by themselves in streets and public places.

No, this is not the occasional psychologically-deranged person who decides to stalk or hurt someone, it is a systematic violence that leads to women deciding to change course while travelling home, invest in pepper spray and other protective devices, change clothes, take a taxi even though they do not want to, stay over at a friend’s house instead of going home at night, or asking to be accompanied by a friendly male – all of these actions that severely limit and threaten our basic freedoms.

It is the same culture that downplays catcalling and street harassment that makes products like these attractive to so many women – you never know when that guy who honked at you as he passed by will not be waiting for you at the end of the street or that the guy you called out for catcalling you won’t react violently. Catcalling and other low-spectrum behaviors are dangerous not in that they actually endanger women, but rather in that they add to the FEAR.

Technology is responding naturally to the demand on the market, and I am glad that it is providing women and other vulnerable collectives with shields from potentially dangerous and risky situations. But technology could also be a weapon against gender violence, by promoting education and training in gender equality.

Agree? Disagree? Leave your comments and participate in the discussion!

Cover photo taken from Stop Street Harassment. 

 

Prejudice is an inherently human feeling. We all have prejudice against certain people or type of people ingrained in our education. People judge each other based on appearance, skin color, belonging to a social class, behavior, disability, origin, education or affiliation. We learn this from our parents at home, from teachers and other children at school, TV shows, books, magazines, etc., Most of us know that feeling – ranging from positive to negative – when we first meet someone. It’s the snap judgment we make about someone’s piercings or tattoos, attire, accent or way of speaking or gender, for example.

A lot has been written about prejudice, and it even conforms philosophical and psychological thought. Is prejudice good? Is it necessary? Can it be viewed from an evolutionary perspective as a way for humans to economize thought and aid in the cognitive process by allowing categorization in situations where not enough time exists to make an informed decision? Prejudice helps us decide whom to trust and who not to trust when we don’t have enough information. On vacation with your friends, whom do you decide to ask for a picture on the busiest street in New York?

Edmund Burke, 18th-century Irish philosopher and statesman and founder of modern conservatism, wrote about the virtues of prejudice: “The individual is foolish, but the species is wise”.

Prejudice is human and its part of that learned social behavior that we cannot avoid, but rather rationalize so as to control it when it does appear. This is because as humans we also try our best to follow high moral beliefs such as that prejudice is wrong and that we should not give absolute value to snap judgments we make about people.

However, when we talk about Artificial Intelligence and Machine Learning, we are talking about creating technology that mimics human intelligence – including the biases that humans harbor. This means that robots and software that are run with this kind of technology will tend to reprodude the prejudices held by the designers as well as those people they interact with because machine learning techniques allow for the platform on which they are applied to keep learning from its users, like a child from its parents.

Which is why the issue of bias in AI and ML is so important. Intelligent machines can carry over subtle biases and cause real harm. Irene Sandler, the vice-president of Cognizant Accelerator, a California-based start-up incubator speaks about an experience she had with gender bias within artificial intelligence systems. She speaks about taking a sentence in English “He is a babysitter and she is a doctor” and translating it into Turkish, a language with no gender. When the sentence was translated back from Turkish into English, it came back as “She is a babysitter and he is a doctor”, meaning that the AI translator they were using had already been taught that the word “he” corresponds more to the word “doctor” and “she” to “babysitter”. You can see the short video here. 

These biases exist not because robots are sexist, but because the people designing them hold prejudices, even though they may not be outwardly sexist. If the team designing a certain software lacks in diversity and the data fed into the system is also biased, the result is things happening like in our example. In this example, no one is actually hurt, but there are other examples where these biases can actually seriously damage a population such as women, as a whole. Facial recognition, for example, is a service increasingly demanded from tech companies, especially by law enforcement. However, many allegations have surfaced about its accuracy. A study published in February 2018 by researchers from MIT Media Lab found that facial recognition algorithms designed by IBM, Microsoft, and Face++ had error rates of up to 35 percent higher when detecting the gender of darker-skinned women compared to lighter-skinned men. If law enforcement used this technology too, black women would face a much higher risk of being targeted unfairly for crimes not committed by them based on inaccurate and biased facial recognition.

Joy Buolamwini, a researcher involved with the MIT study and a black woman, talks about her personal experience with algorithmic bias in her Ted-talk which I highly recommend you watch. She talks about the “coded gaze” to refer to algorithmic bias and speaks about what needs to be done to reduce this scourge. The link to her video is here.

How can bias in technology be minimized?

1. Acknowledging that while AI can bring substantive benefits to individuals and society, it can also have a very negative impact which means that vigilance has to be maximized in areas of critical concern where the damage caused by bias can cause us to lose gains made my many decades of social progress.
2. Keep fighting for diversity in the tech world. Bias gets exacerbated when the people who are designing the technology all resemble each other. Bias can be reduced when a diverse group of people participates in the design process together. This also goes towards the data that is fed to the algorithms. If the people feeding data are diverse, the data will be diverse as well and prejudices such as those regarding women and certain professions will be reduced. Buolamwini refers to this as “checking each other’s blind spots”.
3. Ethical guidelines that acknowledge the reality of algorithmic bias and deal with it directly should be created and followed by all stakeholders involved in developing, exploiting and using AI technology. The European Union has recently published a Draft for Ethical Guidelines for Trustworthy AI which is a pretty good example of responsible “regulation” (these guidelines are not legally binding) in the form of soft law. The panel of experts that drafted these guidelines is made up of a diverse (information about the origins of each expert is not readily available but 22 out of the 52 experts are women in the European tech field – almost half!) group of professionals from a variety of backgrounds. This is a good start for Europe to leave its mark on the development of AI (still dominated by the US and China). These guidelines specifically mention the risk of discrimination that AI may entail and the need for diversity in design.

Interested in algorithmic bias? Please leave your comments!