Propagated by the System: Why Faulty Models Could be Keeping Women Out of Tech Leadership Roles
MIT researcher Joy Buolamwini recently noticed something strange about facial recognition systems from Microsoft, IBM and China’s Face+++: While the systems recognized her white friends, they had a hard time with hers. When Buolamwini, who is black, put on a white face mask, the system’s accuracy improved.
Such work illustrates something that has been evident for a while – AI is a human invention that reflects the biases in available data sets and modeling techniques used by its creators. In other words, data rich with certain characteristics will be more effective at modeling those characteristics.
Something similar could be happening when it comes to funding female tech founders. No matter how great their ideas or how impressive their backgrounds, VCs may fail to recognize female founders and their companies because they’ve been trained to focus on a different archetype, creating their own version of the white-face problem. That could be a reason why only 17 percent of startups have a female founder and why female founders got just 2 percent of VC dollars in 2017. (The situation is even worse for minorities. In Silicon Valley, the lack of representation among African-American and Hispanic employees is known as the “2 percent problem.”) It’s time for the industry to acknowledge these potential biases.
Bias in action
Most men I know work hard to try to eliminate bias and some get offended when they feel attacked by this issue. I try to explain that biases and prejudices are universal and that it doesn’t necessarily mean that anyone has consciously done anything wrong. Our brains are exposed to huge amounts of information every day. The only way to make sense of it is to create shortcuts – generalized models that may work well on average but don’t work well in all cases. While we might be aware of some biases, many are unconscious and contained in those models.
To overcome our unconscious biases though, we need to work hard to better understand what we don’t understand – make the unconscious conscious so we can more easily do something about it.
For instance, at OpenView we spent a lot of time with the National Center for Women in Technology (NCWIT) examining our hiring practices. In doing so, we found many examples of biases in job descriptions that we didn’t notice prior to this work. For example, descriptors like “high-powered” and “action-oriented” may have masculine connotations. But that’s the thing about unconscious biases – you don’t know they exist, so you also don’t know which ones you’re still harboring. None of us are perfect, but awareness at least helps us see where we’re falling short.
Researchers have also found that male managers are more likely to criticize female employees for coming on too strong. Such bosses are also more likely to attribute a female’s accomplishments as part of a team effort and additional data has found that men are also less willing to put women in leadership positions.
Marica Morales-Jaffes, former Chief People Officer at PayPal adds that, “Team leaders play a crucial role in creating a team atmosphere that supports women. At PayPal we found that large teams led by women had almost twice as many female members verse teams led by men. Female team leaders have the power to be role models, mentors and champions in a very direct way. And the best female leaders become magnets for other talented women to join their teams.”
For decades, the archetypical startup founder has been a variation on Mark Zuckerberg – young, smart (usually from a top school), nerdy, focused, wearing a hoodie, and, of course, male. VCs have looked at other things too, of course, like the ideas and business model on which a startup is based, but they have clung to this Zuckerbergian image and it gets reinforced with every successful company created by a founder with that image.
VCs are pattern recognizers afterall. While on paper, such VCs would prefer to make a huge profit on whatever company they invest in no matter what the founder looks like, the reality is that they don’t necessarily have the pattern recognition of what a great female CEO looks like. That’s why would-be fixes like The Rooney Rule (which requires that NFL teams interview minority candidates for open head coaching and general manager positions) don’t work. What’s the use of submitting such candidates to interviews if the hiring manager (or VC in this case) already has a fixed idea of what their dream candidate looks like?
This bias is reflected in the questions that VCs ask female founders. A 2017 study found that VCs tend to ask men questions about potential gains and ask women about the potential for losses, for instance. This was the manifestation of what the researchers called a “promotion” orientation for male founders that focused on hopes, achievement and advancements. For women, there was a “prevention” orientation that looked at the potential for loss. As you might guess, a disparity in funding followed.
Steps toward equality
You can scream all you want about equality, but things aren’t going to change until there are more female tech founders as well as more females in other roles of leadership throughout companies. More women in leadership roles, more data points. More data points, less sampling bias. It’s that simple.
This is one of the reasons OpenView is investing our time, energy and resources into a partnership with Athena Alliance to contribute to getting more women on boards. Through our active partnership with this organization, we have created real change. Over 80 percent of the candidates we’ve suggested to our portfolio CEOs for board seats in the last year have been women. We’ve enhanced our network of top executive women who are actively engaged in discussion with our firm and our portfolio by 142 in the last year alone. And every one of our Investing Partners is personally connected to Athena, knowing that their connection to the organization and its network of powerful women is part of our strategic advantage to OpenView’s portfolio. By doing these things, we have actively created more data points of women in top leadership, and essentially re-trained our own eyes to truly recognize and engage all of the great female talent around us.
Our commitment to creating more data points with powerful women is also why OpenView launched Accelerate, which aims to match highly qualified female entrepreneurs with the resources and access needed to reach the expansion stage. The hope is that by helping more women leaders, we’ll contribute to changing the current investment archetype. If you’re a female founder with less than $1 million in annual revenues and at least five customers in the B2B SaaS space, we want to know you (learn more and apply here).
Bias is something innate in all of us. And it’s something we must all work to overcome. Small, incremental changes will help get us there, but we need to take the time to call out our own biases – conscious or not – to ask if we could and indeed should be doing more.