Framing of Diversity in Tech

Jason Smith Jason Smith

September 7, 2019

pexels-diva-plavalaguna-6147031.jpg

“Diversity in Tech”. This seems to have been a rallying cry in Silicon Valley as of the past few years. All of the major tech firms talk about their desire to hire more and more underrepresented groups.

Diversity in tech is very important for a variety of reasons. I mean over 75% of each group represented uses a smart phone. We can see similar numbers in broadband users. From a pure business perspective, one could see that it would make sense to have a workforce developing products that represent their user base.

Education and skills are important but only paint a small picture of all the skills required to build the technology of a global brand. In a world filled with women, men, black people, white people, Latinx, Asian, Christian, Muslim, Jewish, Buddhist and everything else in the world, we can’t make products that only appeal to a small subset so diversity is needed to create diverse thought and diverse products.

Then of course there is the moral argument. Do we believe that only a select group of individuals are qualified to work a job? Is it possible that just due to genetics, some groups of people are inherently unqualified in these fields? We are seeing more and more Black and Hispanic developers and women make up the majority of STEM graduates. It is pretty clear that the talent is there so if we are a fair and moral society, workforce numbers should reflect that.
But with the moral and business argument, why are we seeing articles such as this one from Tech Crunch telling us that the gap is worsening. But why can this be? I mean, it’s 2019 right? We are all super woke and had a black president and all saw “Get Out” right? Why are we still having a problem with underrepresentation?

I was going through twitter one day and a contemporary of mine, Bryan Liles ( @brianl ), tweeted this:

“I’m not underrepresented anymore. Y’all are over-represented. Not going to use a negative to describe my existence. I’m Bryan, and I’m supposed to be here.”

Those words stuck with me and really reframed my thinking of not just the situation but myself. The problem isn’t that we are under-representing ethnic minorities, women, LGBTQ. The problem is that we are over-representing a specific group. This has resulted in an oversampling of years of data causing an inherrent bias in processes.

Often times people say it’s a pipeline problem. “Well we just don’t have enough people applying so what do you expect?” The problem isn’t pipeline as much as it’s a the culture. Tech companies (and quite frankly most companies) have created an archetype of what an ideal employee looks like. The problem is that the samples they used to create that archetype come from a small pool. Now while they very well may not intend for the archetype to be a white male, the qualities that they assign this “ideal employee” fits.

In fact, Amazon last year learned this the hard way when their AI to help reduce recruiting bias actually increased it. Why did this happen? Well the data used to train the model came from 10 years of resumes and a significant portion were from male applicants. Another way to look at this is that after using years of hiring data, Amazon’s AI determined that a man was the best candidate. Well case closed right? Men are just better at tech and we should move on, right?

Actually no, this is not case. The problem is that the computer is being fed inaccurate data that isn’t representative of the truth. If I wanted to train a model to identify what a doctor looks like and 90% of the photos I fed it were of men, the machine would “learn” that doctors tend to be male. The truth however is very different. In fact, we have more women graduating from med school than men.

This is far from the only AI faux pa that we have seen. There are stories everywhere on this issue. This is very interesting as this is inadvertently revealing the deeper truths of reality. Even when we are well meaning, our experiences subconsciously create a pattern of “normal” and we continue to draw from that same pool of experiences (data) to create more jobs.

Part of the reason we seeing fewer black and Hispanic recruits is because many tech firms don’t look for them where they are. They have been used to looking for people a certain way which won’t work in the name of diversity.

Then there is the inclusion aspect. Diversity is simply the act of putting a bunch of different people together in the room. Inclusion is making them all feel welcome and heard. I can pour a bag of multi-colored Skittles in a bowl and call it diversity. But if I only pick out the red ones from the bowl to eat, I am not being inclusive.

In order to be accepted, code switching is a common method different groups use so that they can blend into the “sample group”. This further creates an environment where one may not feel welcome coming as they are. They have to try their best to become what they think people want them to be.

One can only be someone else for so long before it takes a toll on them. Being black in tech can often feel isolating and sexual harassment is still an issue in tech. These are just some of the few issues that cause the diversity issue.

The problem is not under-representation but rather over-representation that is skewing our views and expectations. In order to fix this problem, we need to frame it in that way. When tech companies and society as whole realize that due to an excess of one type of person, we have created an archetype that permeates our culture, we can then expand our expectations of people to truly be inclusive, meeting them where they are instead of making them come to us.

Cover Image courtesy of Diva Plavalaguna on Pexels