Today on International Women's Day, we celebrate the achievement of women everywhere—while bringing issues of disparity into the light. In my latest blog, I look at why women lag behind men in AI development and usage, how STEM, AI imaging tools, competing priorities, stereotypes, and bias all play a role in AI’s gender disparity problem—and what we can do to change that.
The AI gender gap
In an era where technology is meant to be a great equalizer, a striking gender gap persists in the realm of artificial intelligence (AI), both in terms of adoption and development. While men seem to be diving headfirst into the world of algorithms and machine learning, women, for various reasons, appear to be hesitating at the threshold.
Like most issues relating to women in technology, we need to first unpack the nuanced layers of the gender disparity to understand why women might not be as enthusiastic about AI as their male counterparts. Looking first at the usage side of the AI equation, survey data from FlexJobs shows noteworthy discrepancies between men and women and their relationship with AI. Based on data from working professionals, 54 percent of men are using AI in either or both their personal and professional life, while only 35 percent of women are adopting AI anywhere.
The breakout data shows the specifics of the gender gap in AI utilization, even showing where people are going rogue on usage without permission from the boss:
- Use in personal life: 15% women vs. 21% men
- Use in both personal/professional life: 11% women vs. 12% men
- Use at work, with manager’s permission: 5% women vs. 11% men
- Use at work, without manager’s permission: 4% women vs. 10% men
A divergent adoption curve
At first glance, the lack of gender diversity in AI usage may seem perplexing, especially when technology is ostensibly gender neutral. However, a closer examination reveals a complex interplay of societal expectations, educational opportunities, and nagging image stereotypes.
Historically, STEM fields, the breeding ground for AI enthusiasts, have been predominantly male dominated. Despite women having a banner year and making significant strides in closing the gender gap, women still find themselves underrepresented in tech-related leadership and software development roles.
This underrepresentation manifests itself in the AI landscape, where women might feel like they're once again navigating uncharted territory. A BBC report confirms that theory, noting that women leading AI development report that the hesitancy to embrace AI stems from, well, STEM—or more specifically the ongoing shortage of women studying and working in STEM fields.
AI expert Jodie Cook, the founder of Coachvox.ai, explains in the BBC report, “The current trend in the adoption of AI tools appears to mirror this [STEM] disparity, as the skills required for AI are rooted in STEM disciplines."
The trust deficit: Women and AI skepticism
Beyond the systemic barriers, women also seem to have a palpable trust deficit with AI technologies. This skepticism is rooted in the historical biases embedded in algorithms. From facial recognition systems that struggle with diverse faces to predictive policing algorithms amplifying existing prejudices, AI has not been exempt from inheriting human biases.
Women, rightfully cautious given the biases we face in many aspects of life, are naturally hesitant to fully embrace AI technologies that may perpetuate female biases. Piling onto this challenge is the lack of diversity in AI development teams, which contributes to a narrow perspective during the design phase and can inadvertently overlook the needs and concerns of women.
Competing priorities
Women further face challenges when it comes to exploring and engaging with AI due to the constraints of our busy schedules. Balancing family commitments, household responsibilities, and career aspirations can be demanding, especially for single mothers. While we recognize the potential benefits that AI offers, our attention is often divided among various priorities, making it difficult to fully embrace emerging technologies.
Exposing AI’s Image Problem
In the BBC report, one female business strategist noted that using AI for content generation is just “heavy photoshopping” and moves away from authenticity. She is concerned about the growing trend of AI imaging tools being used by people to create the “slimmest, youngest, and hippest versions of themselves,” which could lead to a place where we have only AI-generated appearances, and her clients won’t even recognize her in person.
Her concerns point directly to one of the distrust AI hot spots for females: images. Women are inherently skeptical of computer image enhancement, as countless photos that feature women’s bodies have generated more harm than benefit. Unfortunately, AI algorithms are off to a rocky start in this area when put to the test.
For example, when classifying this image from the US National Cancer Institute depicting a clinical breast examination, several AI platforms had an epic fail.
Image credit: National Cancer Institute
Google classified the image as having the highest score for racism, while Microsoft AI was 82 percent confident that the image was “of a sexually explicit nature,” even though it was simply one of the 16 photographs taken to demonstrate the recommended steps of a patient’s regular cancer screening. And this is only one example of the many times where AI and image recognition have not been trustworthy companions.
Closing the gap: Paving the way for inclusivity
So how do we mind the gender gap in AI development, before women fall behind? Here’s a few ideas to get us started:
- Advocate for increased representation of women in AI-related fields. This is a crucial step, as ensuring diverse perspectives will inform the future development of AI technologies. A diverse range of voices are needed to actively contribute to the creation—and refinement—of emerging AI technologies.
- Remove educational and access disparities. Providing more STEM programs for women will open doors for them at AI startups and established tech stalwarts.
- Strive for transparency in AI algorithms. We need rigorous testing for biases. A lack of consideration for gender-specific experiences can contribute to a sense of alienation and will fuel female distrust.
- Share and promote the benefits of AI. Women are sponges for information. Forums and roundtables on the topic will help build trust among women to use AI, especially in areas where it has proven to be useful, such as job searches. A piece of this is to amplify the voices of women frontrunners leading the way in AI development.
As we strive for a future where technology is a force for equality, dismantling the barriers preventing women from fully participating in the AI revolution is not just a technological goal, but a societal one. It's time to forge a path where women are not just minority users but are active contributors and shapers of the AI landscape.
