Reflections from a Woman Founder: Why Women Must Be Better Represented in Both AI Technologies and Data Sets When companies have gender diversity in their leadership teams, they outperform their peers.
By Meghan Gaffney Edited by Russell Sicklick
Opinions expressed by Entrepreneur contributors are their own.
I'm a co-founder and CEO of a health tech company that specializes in automation solutions. As such, I am reminded almost daily of the fact that I represent the exception, rather than the rule, in both the healthcare and AI industries. Let's set healthcare aside for the moment. According to a recent article in Forbes about the future of AI, women currently only represent about 18 percent of the C-suite across AI companies. Zooming out to the broader workforce, the numbers unfortunately aren't much better. Despite the fact that the gender gap is closing in many industries, drilling down into data and AI roles specifically, only 26 percent are held by women.
The degree to which women are missing from the workforce is a critical issue for AI
Women are grossly underrepresented in the emerging technology workforce. Yet, research shows that when companies have gender diversity in their leadership teams, they outperform their peers. Excluding women from roles developing solutions with AI means that powerful products will never make it to market.
The workforce effect is compounded by the underrepresentation of women in the data sets AI is built to work on
But women aren't just missing from the AI workforce — we're also often underrepresented in the very data sets AI technology is built to process. This is particularly alarming in healthcare because the currency we're dealing in is human lives. The implications of the lack of data inclusivity in the industry are expressed particularly well in a recent piece in HealthcareITNews: "Artificial intelligence tools in healthcare, as with any other software, are not immune to bias — especially if they have been trained on data sets that do not accurately reflect the global population." Bias in training data sets means that companies are creating products that fail to deliver on their promise. Eliminating bias in the data, as well as having diverse company leadership, leads to improved product development and tools that achieve meaningful results.
Three ways AI entrepreneurs can address inclusivity issues
There is no shortage of articles that offer solutions to the inclusivity issues in the AI industry. But what can't be researched, of course, are learnings from other entrepreneurs. With that in mind, I am sharing three strategies that I have found to be successful.
#1: Build diverse teams and create a culture where employees have permission to talk openly about inclusivity
My co-founder and I are both healthcare outsiders, and we see that as an advantage. Thus, we've doubled down on hiring people that are also interested in applying non-traditional frameworks to healthcare to solve its most intractable problems. In doing so, we've built a culture where no idea feels like it falls outside the boundaries of appropriate company conversation. This includes the topic of inclusivity or lack thereof in AI.
The core of our company's mission is to improve data accuracy, and our devotion to data and science drives everything we do. We've had conversations about how there aren't enough women in AI and tech. We've also tackled the tough question of the lack of diversity in AI data sets. When we do, we've armed employees with foundational resources on the topic.
Related: Fellow Women in Tech: Where Do We Go From Here?
#2: Ensure that women are well represented in your organization, especially in the C-suite and roles that drive the development of technology
Hire the best people you can is still the best business advice I've ever received. But in order to do that, you have to make sure that you are interviewing all of the best possible candidates for every job. Challenge your recruiters to present diverse slates of candidates for every role — especially executives, product leaders and software developers. I found some of our most effective leaders only after pushing recruiters to present broader slates of applicants.
Given the statistics around the AI C-suite referenced earlier, I'm especially proud that today we are one of the few companies in the sector that has a woman chief technology officer. And when women who aspire to enter the industry look at our website, they don't just see that woman CTO — they also see a woman CEO and woman chief people officer. We hope that enables them to envision similar professional futures for themselves.
Related: There's a Massive Gender Gap in AI, but Tech Education Programs ...
#3: Take a critical eye to the data sets your AI technology is purpose-built to parse and speak up if they aren't inclusive
A recent study published in PLOS Digital Health drives home the huge impact of biased data sets (and there are many other studies that have reached this same conclusion) in healthcare: "Given that AI poorly generali[z]es to cohorts outside those whose data was used to train and validate the algorithms, populations in data-rich regions stand to benefit substantially more vs. data-poor regions, entrenching existing healthcare disparities."
Having worked as a political consultant in D.C. when the Affordable Care Act was being debated, I'm very attuned to the health inequities in our current healthcare system. Now, as the CEO of a health tech company, it's my responsibility to ensure that our technology doesn't ever inadvertently contribute to the problem.
With these tips in mind, every company leader should feel empowered to talk about bias. And, educate their employees about building AI systems that serve the diverse communities in which we live.
