AI and Gender Equality: Addressing Biases In Data

The fight against gender inequality has made its way into AI technology, and we may not be able to control the outcome.

Artificial intelligence (AI) is not a new concept, but it’s only now starting to dramatically impact our daily life. AI holds boundless opportunities for our economies and societies, from contributing to sustainable environmental activity to increasing innovation and productivity. 

AI also has the potential to transform individual lives, particularly for women and girls with fewer resources. The emergence of high-paying tech jobs creates better opportunities for women in the workforce. Positive impacts would then flow onto their families, communities, and societies.  

But AI is not without risks, particularly when it comes to human rights and discrimination. Without careful consideration, our inbuilt gender biases can transfer from society into the technology that is meant to serve us.

Now is the time to ask ourselves – are we using AI to enhance the empowerment of women and girls around the world? Or are we reinforcing the gender divide, causing it to widen even further? And how does data play a role in reflecting gender inequality?

Women Are Less Likely to Work in AI Technologies

Despite the emergence of countless new jobs, women remain underrepresented across the design, use, and regulation of AI technology. 

Gender Inequality Makes Its Way Into AI

Dr. Kim, Secretary-General of The International Transport Forum at the OECD, warns that one of the most critical database biases we must be aware of is gender bias. There’s a historical practice of using data designed by men for a system that serves women, and the development of many technologies ignores the existence of inbuilt biases. If gender inequality is embedded within the fabric of society, and AI is developed from data that is input by humans, then if we’re not careful the same biases that exist within society will exist in AI too.

Discriminatory practices baked into AI have already emerged, as seen in Amazon’s failed AI experiment, a recruiting tool that was scrapped after finding it discriminated against women. This highlights how gender biases can make their way from the digital world into reality.

Meanwhile, NHS’s use of AI for early detection of liver disease was discovered to be twice as likely to miss detection in women than men. The result of this has health experts worried it will worsen inequality gaps, particularly for women and underrepresented ethnic groups. 

How can we ensure that AI benefits all of society?

AI needs to be gender inclusive. Systems must be developed in ethical and responsible ways that fully realize the benefits of AI and gender equality.

While data is a core part of developing AI, tackling issues around gender equality requires a focus on the entire end-to-end process. It’s a path that ensures AI develops in ways that improve the well-being of all humans. 

Aiming to regulate this technology is the OECD. Their framework for the classification of AI systems attempts to create trustworthy and responsible AI. Described by the OECD as a user-friendly tool assisting policymakers worldwide in classifying AI systems by their potential impact and providing a basis for risk assessment. Importantly, part of this process involves reviewing the input into the AI system and techniques that address biases in the data. 

Steps Forward: The Future of Women and AI in the Workforce

There is currently a lack of research on how AI will impact women in the future, and there are calls for more analysis in this area.

Shining a light on this topic is UNESCO’s 2018 report, The Effects of AI on the Working Lives of Women, and a 2022 Portuguese version examines the impact of advancing technologies on women and their working lives. The report intends to spark conversations about the future of gender equality, particularly in the workplace. This a timely discussion if we are to ensure women are not left behind as technologies continue to advance at an accelerated pace. 

The report offers six recommendations: 

  • Reskilling and upskilling women workers
  • Encouraging women in Science, Technology, Engineering, and Mathematics (STEM)
  • Accounting for contextual and cultural complexity
  • Leveraging multi-stakeholder approaches
  • Shaping gender stereotypes
  • Continuing applied research

Gender inequality is a societal issue, but it does not have to be part of AI. It is important to apply a critical lens to the data used to build AI and have the input of multiple voices throughout the process. 

It starts with acknowledging that biases exist and then working with policy and decision-makers on how to implement fixes to ensure AI benefits all of society, not just a few. Only by recognizing limitations and setting legislations and regulations can AI be a conduit for inclusion, not exclusion.


‘Want to stay up-to-date? Follow us on FacebookInstagram and Twitter
Interested in volunteering with CFG? Let us know
Not able to come to join us in Thailand yet? Consider donating
Not able to donate today? Look for opportunities in your community to work against gender-based violence and human trafficking, as these are universal issues.

Also keep updated on: