INSIDE MOMENTIVE

Sr. Director of Engineering Jing Huang shares thoughts on representation, the less obvious forms of bias, and how to build a more equitable workplace.

Molly Smith

March 14, 2022 | 5 min read

SHARE THIS ARTICLE

When I went to college, I majored in computer science. In the years since then, I’ve been on engineering teams at leading tech companies, been a technical founder at my own startup, and been to innumerable technical conferences. Even though I have spent a lot of time as the only woman in the room, I have been fortunate enough to have female co-founders and mentors working together along the way 

My experience overall has been highly positive. I’ve never felt like my opinion was less valued as a result of my gender and I’ve never doubted my capability because of my gender. But that doesn’t mean that it wasn’t a problem. Even among the best-intentioned, open-minded teams, if you don’t have diversity, you’re going to have bias. 

Some people hear “bias” and think about prejudice or discrimination.Those things can be related, but they aren’t the same. Your team might not be racist or sexist, but they can still be biased—without any fault of their own. Bias comes from having limited experiences of the world, which shape your perceptions and priorities. Everyone is biased. The only way to have a team, or a product, that isn’t is to hire for diversity. 

When you don’t have diversity at the team level, it can manifest at the product level. People tend to solve the problems that they experience personally and design solutions based on the people they have access to, which siloes innovation into certain groups. Silicon Valley has a well-documented problem with male-centered innovation, driven by companies that are largely founded and controlled by men.

For example, the default setting on VR headsets is too broad for 90% of women’s pupils, resulting in them getting sick after usage much more often than men do. Many phones and game controllers are also sized for men’s hands. Not to mention the different head and facial shapes from different ethnicity groups. The technology is designed to be optimal for the kinds of people that are building it.

Now, extrapolate that problem to machine learning. Our teams are building algorithms that learn and replicate—sophisticated programs that can unintentionally amplify bias or build it into the DNA of that technology, like the infamous MIT and Stanford study that found that commercial facial recognition programs had an error rate of 0.8% for white men and over 34% for women of color.

The people who are left behind in these areas aren’t blind to what is happening. Recent Momentive research unsurprisingly found that people of color are especially concerned about bias in AI. They know that they will be the ones left unserved if this technology is built by people that don’t look like them. 

According to the researcher, Inga Vailionis, ”In the business world, when algorithms are trained using biased data, there are major ramifications for companies’ products, user experience, and reputation. From recruiting tools and credit card application algorithms to UX features like automated photo-cropping, when AI is fueled by biased data, it can lead to bad predictions, discrimination, and the reinforcement of racist and sexist systems.”

Engineering programs consistently skew male. According to Statista, 91% of software engineers in the U.S. 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’s a lot of bias to overcome. 

Hire for diversity. Momentive has set public goals related to diversity, including goals specifically for technical positions. DEI is about much more than meeting a number, but at the same time, the only way to address bias is to ensure that there is some level of representation in these influential roles. Sometimes, this means slowing our hiring down later than we’d like. That’s okay—we know that the investment is worth it to us in the long run.

Create standards for development and testing. Even if your team isn’t balanced from the get-go, establishing a set of standards for inclusion and/or decision-making can reduce the risk of bias in your engineering work. You can also create processes for reviews or stakeholder sign offs that emphasize equity as well. 

Keep leadership diverse. I was lucky enough to have a personal mentor in Robin Ducot, the CTO here at Momentive. Many women in STEM are not so lucky. Having diverse leadership signals to potential employees that you are an equitable organization where they will have a fair chance at being promoted. It also ensures that someone in a position of power has their considerations in mind when important decisions are made about products, internal processes, or strategy.

Create an ERG for underrepresented people in STEM. Community can be a powerful way to overcome adversity in the workplace, brainstorm new ideas, and establish more of an influence on the way that a team develops. Momentive has an employee resource group (ERG) that is specifically for women on our tech teams. I think this helps us create more of a culture of balance.

I’m not the only person on my team advocating for diversity. My colleague Peng Jiang recently pointed out that “It is vital that the AI-powered insights the team has built could reflect different perspectives and therefore drive positive social changes.” We’re all aligned around this value, and that makes it easier to act on it.

I love my field, and I’m very aware of the impact that teams like mine have on day-to-day life and on what the future looks like. My hope is that we can find a way to spread that influence more equitably so that the technology of the future serves everyone, regardless of gender or gender identity.