Gender inequality in AI. Myth or Fact?
For centuries, women have been subjected to inequality in treatment by the male dominated societies. Women have had to fight for their fundamental rights that theoretically, have been promised to all individuals by birth. Although the world has moved on to bigger and better things now, women continue to be subjected to partiality through indirect or subtle actions.
Artificial intelligence was initially thought of as the solution to such problems because it doesn’t require much human intervention, hence the possibility of any kind if bias was thought to be zero.
But what we didn’t estimate was that the algorithms developed by developers also represented a group of male dominated groups that had a patriarchal mindset embedded into their minds through the society. And that is how gender bias in artificial intelligence was born.
Although artificial intelligence is in its very initial stages right now and doesn’t affect our day to day lives very much but we must keep in mind that the future revolves around it and its applications.
A small bias right now will continue to expand as artificial intelligence finds its roots in almost all aspects of our lives. This minor bias will become a whole issue in the future, the same that the women population have been fighting against for a long time now.
Another possible way in which artificial intelligence is creating gender stereotypes is through voice assistants; majority of which have female voices. Such products instill a certain image in young minds that further develops into disrespect of a certain community, women in this case.
Several other biases have also been identified lurking in algorithms that create issues for women. Let us have a look at some of them here:
1) Stereotypes:
Algorithms work on datasets through which the machine learns. This is the very basic fundamental of machine learning. Many times, the quality of datasets is such that the algorithm starts associating certain tasks with certain communities. There have been instances in which algorithms started labeling data sets of basic work like cooking, cleaning, baby-sitting, shopping etc. with the female class and those of earning, working in offices etc. with the male class. Artificial intelligence is amplifying the pre-existing stereotypes in the society.
2) Discrimination on gender:
Several years back, Amazon was charged with the act of discriminating against women during its screening process before job interviews because its algorithm was awarding negative points to resumes with words like “female”, “women” etc. Amazon tried to make changes to the algorithm, but the bias was so much deep rooted that it had to scrap it altogether. This is just an instance of gender-based discrimination through AI. Such behaviour has been noted in a lot of instances when women have been denied credit cards whereas their husbands have been approved despite of the fact that both belong to the same income class and have similar credit scores.
3) Lack of inclusion:
Jobs in the IT industry have been male dominated since a long time. This lack of inclusion results in developers not taking into consideration the female point of view. This therefore results in all the biases aforementioned. This lack of inclusion also leads to women not getting the same comfort as men in many cases, one of them being the fact that many health apps do not have menstruation related issues listed. Also, voice recognition systems have been majorly tested on the male voices, which is why they have difficulty in identifying the female voice. All these issues would have been resolved internally only if female developers were also a part of software development teams in good numbers.
4) Cyber-safety:
One factor which might contribute to women remaining backwards as compared to men in technology in the future might be because of the fear that they have in exposing themselves to the world of internet. This fear is predominantly because they don’t feel as secure on internet as men do. This is because women are the major targets of cyber-bullying, online harassment, stalking etc. In remote areas, families tend to keep their women out of internet but impose no such major restriction on the males. Many pre-existing chat-bots exposed to the wrong kind of data have been accused of such issues.
There is no doubt that artificial intelligence poses a risk to equal gender inclusion in all spaces. But at the same time, it is a solution to all the pre-existing problems existing in the society including discrimination. This is due to the fact that artificial intelligence requires less human intervention than the present-day scenario and if human interaction is reduced, the society could refresh its pre-existing perceptions to start from a new phase altogether.
Let us discuss some ways in which artificial intelligence can become gender bias free in the future.
1) Inclusion:
All the above-mentioned issues can be solved if a female perspective is also considered seriously while developing algorithms. This is only possible if females are present in such development communities in more numbers. Increasing the representation of women in such male dominated areas will help in eradicating all biases. This further would motivate more women to rely on technology with more trust than they do now.
2) Education:
Inclusion is only possible through proper education. The inherent reason of the fact that the software industry is male dominated is that women aren’t encouraged to study more especially in south Asian countries. Only proper education will enable women to increase their representation in all areas.
3) Proper assessment:
There should be proper assessment of all companies and the algorithms they use to make sure that biases are not fed to the systems. Checks should be done to make sure that companies follow the mentioned ethics and do not indulge in activities that promote or exhibit discrimination.
4) Funding of women founded initiatives:
Despite all challenges, several women have been running various initiatives in technology. Such initiatives should be funded appropriately so that their development is not hindered. These initiatives will also promote other women to step into the world of tech and destroy pre-conceived notions.
Artificial intelligence is the future and all societies should make sure that the hindrances that are being accounted in the present world are not carried forward. Artificial intelligence will enable the world to start from a new perspective all together and will empower people to their basic rights without any discrimination. Hence, we should make sure that the biases in our ideologies are not carried forward as time progresses.
About the author:
Manasvita Sharma, 4th-year B.Tech student at Bennett University.
Connect with me on LinkedIn here.
