Encouraging women in AI has never been more urgent. A study by the World Economic Forum noted a gender disparity of 78 percent male versus 22 percent female in AI and data science. This disparity isn’t just a challenge within the workforce. It reflects a highly nuanced issue that goes beyond any single workplace and if not addressed will have highly negative implications for society.

We have seen a lot of work to encourage girls and women to become interested in STEM and address gaps in digital skills at an earlier age than in the past. Yet now, there appears to be less effort to support women as they transition from higher education into a sustainable career in tech. This is a challenge for the industry. But the real problem is that as AI becomes ubiquitous in daily life, without a technology workforce that accurately reflects the structure of society, AI-based decisions are constrained by the limited societal and cultural biases of their designers. The impact of such homogeneity in AI decisions and bias has already been seen in examples such as the automation of credit card and mortgage applications, to resume screening and other areas. 

The industry challenge is not due to a lack of skills. Research from the Turing Institute suggests women are trailing behind men with industry-relevant skills such as computer science, data preparation and exploration, general-purpose computing, databases, big data, machine learning, statistics, and mathematics. Yet much of this is not due to formal skills, but rather confidence by women in stating these abilities during recruitment and in the workplace. In the tech world where technical skills are needed, soft skills are sometimes dismissed but in order to move forward, there needs to be a greater focus on leadership and mentorship to build confidence and encourage a more diverse workforce. We say that stereotypes must be combatted from a young age yet a gap remains. For example, within the tech sector, women generally have higher levels of formal education than their male counterparts yet academic citations are fewer suggesting there is a lack of confidence in sharing academic knowledge. The Turing Institute finds that only 20 percent of UK data and AI researchers on Google Scholar are women. Of the 45 researchers with more than 10,000 citations, only five were women. 

When I say that women need to have mentors and role models, I write from firsthand experience. It was only after winning a mathematics modeling competition in university that I considered a related career. This inspired me to write a blog on machine learning algorithms. The easy-to-understand method employed helped the blog garner over 5 million views, and eventually led to a career in programming. When I became a programmer and found myself working as the only woman in a room of men typically 10-15 years older, I struggled to relate and realized the need for a community of like-minded people.  

In April 2020 I started to manage operations for MindSpore, an AI framework developed by Huawei, just as it became open source. MindSpore is Huawei’s alternative AI framework to Google’s TensorFlow and Facebook’s PyTorch with comparable capabilities but 20 percetn fewer lines of code. Launched in September 2019, it is endorsed by major universities including Peking University, University of Edinburgh, and Imperial College. Today, MindSpore boasts over 1.3 million downloads and an interactive community indicated by over 19,000 issues, over 52,000 pull requests, and over 16,000 stars (the equivalent of a ‘like’ among developers). 

In 2021, open-source component downloads grew 73 percent YOY. With the rapid growth in the global adoption of open source technology, diversity in open source communities is also increasing. The MindSpore Women in Tech Community emphasizes seminar-like gatherings which provide women a safe space to discuss the challenges they face in the workplace. Mentoring is important. For example, in 2020, when the community was just in its infancy, a student at one of our events explained she was getting good grades but was worried about a career in programming. She sought advice from more senior programmers and tech leaders. By the time she graduated she had no need to worry and was able to choose from one of several offers. Not only did she feel more confident but was able to give back to the community by sharing her experience with new students, those who were now in the position she had been the previous year.  It is experiences like this that will keep women in tech. When they stay, tech also benefits. 

But encouraging women isn’t simply about creating diversity within the industry to enable greater gender balance. The benefits stretch beyond the sector and into the societal benefits. With the digitalization of many traditional sectors, the pervasive nature of AI demands that it not only provides efficiency but is also inclusive. It is only by broadening the pool of talent that we can avoid data-led decisions skewed by bias. Establishing communities that actively foster participation and diverse voices is an important step. 

Bias in AI starts with the initial formulation of problems. The questions are naturally constrained by the experiences of the designers and programmers. This in turn impacts the quality of the data and the way it is handled. So what will be the societal impact if there is not greater diversity?

• User experience (UX) for women will be not be as intuitive if it is there is not greater input at the design stage
• Economic discrimination whether assigning women’s resumes to lower paid jobs and access to financial resources will have a long-term impact
• Societal resources will be distributed unfairly whether affecting education, healthcare, or even safety
• Women will lose decision-making capabilities for fundamental day-to-day as decisions

So in conclusion, now that our lives are digitally-driven, we must ensure that women can enjoy the benefit of technology for generations to come rather than be negatively impacted.

© 2022 LeackStat.com