Home Artificial Intelligence International Women’s Day: The Gender Gap in AI

International Women’s Day: The Gender Gap in AI

by Dimitrios Koutrotsios
International Women’s Day: The Gender Gap in AI

With more than a century having passed since the pioneering women of the early 20th century began making strides in the workplace, gender inequality in the labour market should have been consigned to history long ago.  But the sad reality is that there is still a very long way to go: even on the most obvious metric – the gender pay gap – the world is only 68.4% of the way to equality, and at current rates is not going to achieve full parity for another 131 years.

If it were not bad enough to learn that gender equality is unlikely in our lifetimes, there are certain sectors and industries where the picture is even more bleak: AI in particular is trailing the pack on equality of opportunity, with women accounting for only 22% of its professionals. Why is the number so low, and how can we change it? We, at DailyAI, look at the stats and sit down with Agnieszka Suchwałko PhD, COO of QuantUp, and Alysia Silberg, the founder and CEO of UnemployableAI to get an insider’s look at gender discrimination within the industry:


AI jobs are booming, but not taken by women

There can be no doubt that AI as an industry is booming. Even going back to 2020, a LinkedIn report identified “Artificial Intelligence Specialist” as a top emerging job in the US market, and in the four years that have followed the role has seen hiring growth of a staggering 74% per annum. The demand for employees in the sector is undoubtedly the strongest it has ever been, but the supply is proving to be decidedly male-focussed, and little has changed on that front for more than a decade.

In light of this, Alysia feels it is essential to highlight the historical context that the original coders were women, particularly during the mid-1900s wartime efforts. This underscores the foundational role women have played in the development of computing and technologyDespite the fact that women account for 47.7% of the global workforce, and are more likely to hold both a bachelor and master’s degree than their male counterparts, nowadays they account for only a quarter (26%) of all AI/data positions in the workplace.

Against this statistical backdrop, there can be no denying that there is a huge glass ceiling in the AI industry, and in many ways it is a systemic issue rather than a product of direct discrimination. It is most obvious and measurable from the profile of the current workforce, but it has its roots in a much earlier stage of life.


ICT has 500% more male graduates than female

Even before most future professionals begin to think about entering the workplace, the seeds of disparity are very often already sewn by what are – seemingly at least – free educational choices. Recent research by the World Economic Forum shows that the percentage of males taking degrees focussed on information and communication technologies is 8.2%, almost 500% higher than females who choose to focus on this area (1.7%).

None of this is to say that the resulting inequality is self-imposed. Far from it: it is hardly surprising that so many smart and aspirational young women entering university feel that their education would be better focussed on another sector, not least because there are so few female faculty members in the tech world. The Stanford Institute for Human-Centred AI, for example, found that women make up just 16% of AI-focussed tenure-track faculty.

Despite the rocketing growth of AI in the past decade, nothing much is changing with this early and all-important feeder into the workforce. In 2019, for example, women accounted for 22% of AI and computer science PhD programs in North America, with growth of just 4% from the same statistical category in 2010. This snail’s-pace progression in the upper echelons of academia is a global problem, replicated worldwide, with the number of women taking artificial intelligence and computer science PhDs stalling at around 20% for the past decade and currently shows no sign of shifting.


It was 20 years ago that Agnieszka chose to study for a degree in Computer Science and she isn’t surprised to find that little has changed over the past two decades:


“In our [class] there were three girls among more than 20 guys. If someone had done a summary comparing the results by gender, the difference would have been very clear. More was required of us, and we managed”.


Women expected to work in ‘purpose-related’ fields

Even for those courageous enough to take on a male dominated field, getting to the point of qualification is just half of the struggle. By doctoral level, for example, one of Agnieszka’s female classmates had already moved into a different area, while Agnieszka and the only remaining woman switched to a program with a more practical focus. During her PhD in Biocybernetics and Biomedical Engineering, Agnieszka found that she was often directed towards projects which had tangible applications, rather than more theoretical ones.

Agnieszka’s experience is by no means unique, and brings to life the hypothesis laid out by Emma Fernandez at the Esade 4YFN in March 2023, that girls are, from a young age and throughout their lives, pressured by society and stereotype to focus their energies on work that is “purpose-related”. Technology is rarely perceived as something with a tangible purpose; it is seen as a tool, rather than a way of achieving a measurable benefit. This is of course a misconception, not least in view of the recent scientific and health breakthroughs attributable to artificial intelligence, but that does not stop it standing in the way of equality.

The point the Esade Panel were striving to make is that gender inequality in the technology world can be traced back even earlier than university – even as far back as infancy – and is ingrained in our social structures. It begins with something as small and innocent as the language we use to communicate to children about the purpose of technology, while gendered toys and games help to instil society’s expectations about boys and girls. Robots and computer games, for example, are still often seen as boys’ pursuit, leaving many girls feeling out of touch with technology from a young age. This fuels a lack of confidence that manifests itself even in the basic stages of education, with recent research by Teach First revealing that 43% of girls lack confidence in science, compared with only 26% of boys.


Fernandez puts the point in a nutshell:


“children never choose things they don’t know”.


The path to equal representation in the tech field must, therefore, begin at school and there are a number of relatively simple ways of making real progress on this front, whether by up-skilling teachers or investing in STEM initiatives for young girls.

In Agnieszka’s view, we could benefit from going back further still to the preschool level.


“We need to focus on developing partnership relationships between the genders from an early stage… The future is up to us”.


Alysia agrees that education is important, but calls for a more diversified approach.


“Advancing gender equity in AI requires a multifaceted approach, including education, non-gendered tools, and promoting emotional intelligence. My mission aligns with the UNAI’s goals, emphasising the need for systemic changes to support women’s involvement in AI. By focussing on these areas, we can empower women to become leading forces in AI and technology, driving positive change and innovation for the betterment of society.”


Female representation in AI is important

In November 2023, a year after the launch of Chat-GPT, OpenAI’s CEO Sam Altman was temporarily replaced by the company’s long-term CTO, Mira Murati, who has been named as the ‘most interesting women in technology’. While Murati has now relinquished the role to Emmett Shear, her influence lives on and she is credited with helping launch AI into the mainstream.

Unfortunately, however, Murati is the exception rather than the rule in the tech world. Young girls and teenagers have traditionally had very few female role models in the AI sector, which in turn makes it that much harder to spark enthusiasm, let alone passion, for the field. While Elon Musk and Sam Altman are almost household names, vanishingly few will have heard of Fei-Fei Lin, who created ImageNet, or Elaine Rich, whose work established the foundations of AI research and paved the way for further developments in the field.

Artificial intelligence, like science, also has a propensity to suffer from what has been dubbed ‘The Matilda Effect’: the tendency for women’s contributions to be overlooked, downplayed or attributed to male colleagues. Agnieszka now works alongside her husband and male partners who always look at the know-how and not the gender of the team, but she hasn’t always been able to avoid prejudice:


“Unfortunately, even my husband did not believe in me initially; although my mother-in-law is still an active architect today. So I repeated it like a mantra: “you did a doctorate, which means you are no dumber than the men you work with”.


Alysia describes how in order to get ahead she too had to be accepted by her ‘male peers’ but was determined not to lose her identity in the process:


“My path has involved leveraging AI to level the playing field and normalise my voice in a domain where I’ve often been one of the few women in the room. Working alongside some of the most innovative founders in Silicon Valley has been both challenging and exhilarating. It has required me to navigate the nuances of being accepted as one of the “boys,” all while maintaining my identity and integrity. My success in this field has not just been about fitting in; it’s been about breaking barriers and reshaping the landscape to be more inclusive and equitable for women.”


Like many women Alysia and Agnieszka both had to work harder than their male colleagues in order to prove themselves. Agnieszka did not allow this to knock her self-esteem:


“Confidence, in particular, is something that no one can give you, or even help you build. People may try to make you feel bad about yourself, but you can fight back. You are different from the people around you, and you know it. Use that difference, for it is your extraordinary power with which you will build your good future”.


Mentorship is the best way for women to learn from each other

There’s no doubt that Agnieszka and Alysia have both worked hard to get where they are and fought off the naysayers along the way. While it is important we recognise and celebrate these achievements, and those of other female tech pioneers like Mira Murati, true progress in the field will only come when women’s achievements are no longer considered unusual or unexpected.

There is hope for change though, with organisations like WLDA (Women Leaders in Digital and AI), created by Asha Saxena, springing up not only to encourage more women to enter the field, but also borne of a belief that mentorship and peer-to-peer feedback is the best way for women to learn from and uplift each other.

This is something that Agnieszka can get behind:


“To get more girls and women interested in AI, we need real examples, real women’s stories, to show that it’s possible. People who are mentors have authority: respect and influence. So they can help girls and women who dream of working in AI to see how their assets can accelerate their careers and open doors in the AI industry. We need others who are stronger than us to show us that we are good enough to do it.”


Alysia herself is both the Founder and General Partner of the investment firm Street Global, where she mentors tech startups and helps them go public:


“My commitment to mentoring and supporting the next generation of women in AI is rooted in the conviction that women possess all the necessary qualities to harness the power of AI effectively. They bring unique perspectives, empathy, and a nuanced understanding of social implications that are crucial for the ethical development and deployment of AI technologies”.


WLDA’s focus is naturally on empowering women to expand their leadership capabilities, but one of their many strategies is to recruit male allies within the industry, who can create impact in parity and equity. This echoes Agnieszka’s own professional mantra, to support the strengths of every person regardless of gender or age. Outside of mentoring programs, Agnieszka recognises that teamwork within businesses is vital:


“I also believe very much in the power of a team. Any initiative with teamwork and sharing of responsibilities to show different perspectives of a challenge is very valuable. In most cases, the difference between men and women is fictional when it comes to the work we do, and it is up to us to notice it”.


Equality improves the quality of your product

71% of people believe that adding more women to the AI and machine-learning workforce will bring much needed perspectives to the industry. There is a real issue at present with natural language processing, a key component of common AI systems like Apple’s Siri and Amazon’s Alexa, developed primarily by men, demonstrating distinct and negative gender biases. Similarly, there have been issues with computer vision systems for gender recognition reporting higher error rates in recognising women, particularly those with darker skin tones. This is often attributed to an incomplete or skewed training data set, generated without adequate female input.

In Alysia’s experience it’s how we approach these discussions that is of paramount importance:


“My experience in Silicon Valley has shown me the importance of shifting discussions from tokenism to proven results in promoting gender equity”.


Agnieszka feels that this problem lies with the world at large and that technology should not itself be blamed for failing to be equitable and inclusive:


“The world still isn’t designed to meet the needs of men and women equally. I would like the phone to fit my hand and pocket just as it fits my husband’s hand and pocket. I would like the mannequin representing a woman used in crash tests to be not just a scaled-down mannequin of a man but to have a woman’s physiognomy taken into account. We as a society need profound change. Fortunately, it’s happening. We’re approaching the turning point”.


AI is becoming such a key component of everyday life that there is a risk of underrepresentation in this field having a wider impact on society, and setting back equality efforts across the board. Take the gendered nature of robotic systems, for example: with robot waiters, receptionists and telemarketing bots invariably programmed by men to use female voices, there is an obvious potential for gender stereotypes to be reinforced.

One crucial element that we often fail to discuss in our countless debates about gender bias and stereotyping is, in Agnieszka’s eyes, our responsibility to use AI:


“Even though we know about both sides of AI (the bad and the good), as humans we’re still not eager to unanimously choose the right side. So I have to address and emphasise that there is no global and massive pressure to tackle all forms of bias in AI projects”.


UNESCO insists governments take action

While Agnieszka feels that there is no pressure to tackle these biases, UNESCO disagrees. In their Recommendation on the Ethics of Artificial Intelligence they address the fact that AI may be trained on personnel datasets that represent pre-existing human-hiring biases, which often feature a strong male skew and could result in AI-systems favouring male candidates over female ones. As part of a targeted package of actions, they recommend dedicated funds for policies which support women and girls, to ensure that they are adequately represented in AI systems.

UNESCO insists that governments should be implementing gender action plans, for integration into national digital policies. Many would argue that steps such as these are crucial for promoting and advancing women’s participation in the digital sector, but Agnieszka feels that imposing further regulation on the private sector isn’t the answer:


“We don’t need more restrictions. We need more enticements. Instead of introducing new rules or obligations, we need to focus on supporting those organisations that base their development on hiring wise managers. This will lead to more organisations wanting to be like them. You can’t change people’s minds with more bureaucracy”.


Alysia, however, feels that dedicated funding for gender-related schemes and integrating gender action plans into national digital policies, are both essential measures for creating environments where women can thrive in the digital sector. This in turn will contribute significantly to technological advancements and innovation.

When it comes to gender inequality, Agnieszka is adamant that companies should be tackling this at the recruitment stage:


“Start with the people you hire. Rely on managers with strong reputations. They can create a supportive and inclusive environment for everyone. They will do internal audits to check whether the hiring process is based on expertise rather than gender. They will listen to everyone regardless of gender, formal education, or seniority, showing that everyone is valuable, each in their own way, and that we will gain the most by working together. As a Chief Operating Officer, I oversee whether the hiring process in my company is fair and whether people feel valued. It’s extremely important, because in AI if you want to build A-teams, you have to select different personas and then rely on them”.


Alysia agrees that it is up to AI organisations to ensure that they’re creating a supportive environment for female colleagues:


“Organisations must create inclusive environments that encourage women to excel in AI. Recognizing the pivotal role of AI Engineers and the unique contributions women can make is crucial. By valuing the human factor and intelligence that women bring to the table, companies can foster positive disruptions and advancements in AI technologies. Supportive policies, mentoring, and career development opportunities are key to achieving this”.

Source Link

Related Posts

Leave a Comment