The decade of the 1990s witnessed the rise of women’s empowerment as a policy objective which came into prominence with the Beijing conference. In this context, a focused attention was given to economic independence for women as one of the most important ways to promote gender equity, participation and decision making. The Beijing Platform for Action spoke of the need to promote women’s economic independence, including employment, and ‘ensuring equal access for all women to productive resources, opportunities and public services’. This dimension has become increasingly visible within the international policy discourse in recent years. The Millennium Development Goals on gender equality and women’s empowerment also adopted an increase in women’s share of non-agricultural employment as one of its indicators of women’s empowerment. ‘Full employment and decent work for all, including for women and young people’ forms one of the Sustainable Development Goals.
The basic logic behind the argument emphasizing economic independence for bringing empowerment among women is that participation in income generating activities will lead to direct access to resources, enhance women’s bargaining power and provide exposure in the public sphere to interact with a wider network of people which will increase their self-esteem and self-worth. UN women advocate that investing in women’s economic empowerment sets a direct path towards gender equality, poverty eradication and inclusive economic growth. At both international and national levels, economic empowerment of women has been advanced through various mechanisms designed for promoting financial inclusion, self-help groups, skill training, entrepreneurship, paid employment and self-employment. Increasing representation of women in formal employment has been identified as the one of the most effective channels here.
Among the economic sectors that have recorded a rising representation of women, the Information and Communication Technology industry holds a unique place. Acclaimed as the poster child of the New Economy, the ICT industry is considered to be non-discriminating and inclusive, offering equal opportunity for employment to privileged, marginalized and socially challenged sections of society all alike. It is characterised with several distinguishing features such as high degree of integration into the global economy; distinctive and apparently employee friendly human resource policies; women-friendly work environment thus making tech an attractive option for skilled female graduates. However, the story is not all fair as it seems to be.
Women are underrepresented in tech with only 20% of tech jobs held by women. As the future is increasingly being designed by innovations in tech, it is a disservice not only to women but humanity as a whole if women are not equally represented in its design. Absence of women from designing the world implies that their needs will not be represented and the outcome will be a world unsuitable for women. For example, initially car safety failed to take into account female anatomy and was only included in the US in 2011. Another striking example is Amazon’s experimental AI hiring tool that was being developed for mechanizing the search for top talent. In 2015, the company realized that its new system was discriminating on the basis of gender among candidates for developer jobs and other technical posts. The reason was that the tool was trained to sift applicants by identifying patterns in resumes submitted to the company for 10 year period. Since most applications came from men, the sifting reflected male dominance across the tech industry. Artificial Intelligence learns like babies do: it picks up data and knowledge from the world around it and so AI will further amplify and deepen the existing inequalities.
Inequality in the fields of data science and AI:
There is evidence of persistent structural inequality in the fields of data science and AI, with the career trajectories of data science and AI professionals differentiated by gender. Women in data and AI are under-represented in industries which traditionally entail more technical and overrepresented in industries which involve fewer technical. Previous research has shown that women in data science and AI have higher formal educational levels than men across all industries but there is still a high achievement gap which increases for those in more senior ranks.
The stark lack of diversity in the AI and data science fields has wide reaching consequences. Mounting evidence suggests that the under-representation of women in AI sets a feedback loop whereby gender bias gets built into machine learning systems. Although algorithms and automated decision-making systems are presented as neutral and objective, the fact is that bias penetrates and is amplified in AI systems at various stages. This can begin with the data used for training AI tools that may not be inclusive and carry the historical biases. The biases of the coders and designers may penetrate in modelling or analytical processes due to own (conscious or unconscious) values and priorities or resulting from a poor understanding of the underlying data and thus finally give birth to biased AI systems. It is significant to understand that technology does not operate in vacuum but within the environs of society we live in.
Call for inclusivity and representation in AI:
We urgently need deeper data and analysis on representation of women in AI in order to strengthen efforts to avoid hard-coded bias. We need to step up and use this data to ensure that biased technology is not developed in the first place. Including women in designing technology would lead to inclusivity and diversity. For instance, more women in tech would lead to better design of health apps such as period tracking apps wouldn’t focus almost exclusively on planning for pregnancy, getting hourglass figures or ‘ideal body shape’. Chatbots will not tolerate abuse and respond to sexist comments. Transcription softwares will identify different accents and scan machines will identify diverse skin tones and colors. And finally , this will make machines more intelligent and inclusive.
