Product engineers need to take action to address their own bias and prioritize diversity to create inclusive innovations that can benefit everyone.

Jing Huang, Senior Director of Engineering, Machine Learning

July 8, 2022

4 Min Read
Colorful wooden stick with DIVERSITY word. diversity concept
Hafiez Razali via Alamy Stock

When I went to college, I majored in computer science. Since then, I’ve spent a lot of time as the only woman in the room. According to Statista, 91% of software engineers in the US identify as men. That’s wildly out of proportion for a group that should be about half of the population -- especially for a role that’s so influential on modern life.

That said, I was fortunate enough to work with talented female co-founders and mentors along the way. I learned that even among the best-intentioned, open-minded teams, if you don’t have diversity you’re going to have bias. Here is why lack of gender diversity in engineering is a problem and what you can do about it.

Strive for Representation in Your Data Sets

As we develop new technology, it's crucial that we bake in diverse perspectives from the start. An infamous MIT and Stanford study found that commercial facial recognition programs had an error rate of 0.8% for light-skinned men and over 34% for dark-skinned women. This is proof that people tend to solve the problems that they experience, which siloes innovation to certain groups. When we turn a blind eye to perspectives other than our own, we risk creating innovations for only some of us, not all.

Now, extrapolate that problem to machine learning. Without a representative data set, you run the risk of introducing unintentional bias. Unfortunately, there is no easy cure-all to defining a diverse data set, and it is very difficult to create universally applicable standards in the machine learning context.

What we do know is that artificial intelligence programs amplify the perspectives of the data they’re fed. From a team diversity perspective, if you want to tap into data diversity, you need more examples and data points to back it up and be credible.

At my company, we build algorithms that learn from and replicate the two million questions answered on our platform every day. We are also building machine learning algorithms to help users on our platform create better questions by avoiding double-barreled and insincere questions. High-quality responses start with high-quality questions. Even if your data input isn’t balanced from the get-go, casting a wide net helps reduce the risk of bias in your engineering work.

Keep Leadership Diverse

Building an equitable workplace is more than meeting a number. Addressing bias requires ensuring some level of representation in influential roles. I am lucky enough to have a personal mentor in Robin Ducot. As our CTO, she advocates frequently on getting more women into leadership roles. Diverse leadership is a signal to employees that you are an equitable organization where they have a fair chance at career advancement. It also ensures that someone in a position of power has their considerations in mind when making important decisions.

This starts at recruiting where even language in job descriptions can be a barrier for some women interested in applying. This is where AI comes in handy. We use a tool to help screen job descriptions for bias. You might also consider hiring talent sources focused full time on identifying underrepresented candidates. For director and above positions, try interviewing women and underrepresented minorities before making any offer. Sometimes, introducing steps like these requires slowing down your hiring process more than you’d like. That’s okay; you should know that the investment is worth it in the long run.

Invest in an Employee Resource Group

Communities like employee resources groups (ERGs) are a powerful way to ensure fair representation. They can also spark positive effects on your products. This is important, considering Silicon Valley has a well-documented problem with male-centered innovation. Research found that VR headshots are too broad for 90% of women’s pupils. This resulted in them getting sick after usage much more often than men do.

Because ERGs are often composed of people on many teams, they can infuse diverse perspectives across the business. Our research also found that 62% of workers consider DEI (diversity, equity, and inclusion) to be “an important factor in our company’s ability to drive success.” Meanwhile, nearly half of C-level executives consider DEI “a distraction from their company’s real work.” ERGs can address this “DEI disconnect,” while brainstorming new ideas and influencing how the company develops.

Representation might be the less obvious form of bias. Yet, as we develop new innovations, it's vital to ensure technology serves everyone, regardless of gender or gender identity.

About the Author(s)

Jing Huang

Senior Director of Engineering, Machine Learning, SurveyMonkey

Jing Huang is Senior Director of Engineering, Machine Learning at SurveyMonkey. She leads the machine learning engineering team, with the vision to empower every product and business function with machine learning. Previously she was an entrepreneur who devoted her time to build mobile-first solutions and data products for non-tech industries. She also worked at Cisco Systems for six years, where her contributions ranged from security to cloud management to big data infrastructure.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like


More Insights