Facets: Issue 39

June 24, 2017

Algorithms aren’t racist. Your skin is just too dark.

Joy Buolamwini
"keep in mind that default settings are not neutral. They reflect the Coded Gaze, the preferences of those who have the opportunity to develop technology."

Having done extensive work to combat bias in machine learning, Buolamwini introduces the idea of “exclusion overhead”— the insistence that the underrepresented accommodate the biases of machine, rather than focusing on inclusive AI. Buolamwini’s Illuminate Series identifies the impacts of biased algorithms and underlines the need for inclusive, representative data sets which force the question: "Who are the default settings optimized for?"

The fight against racist algorithms

Tolulope Edionwe
"how do you help an algorithm unlearn racism?"

Our algorithms reflect our own implicit biases; how do we train them to behave inclusively when the people who build them do not? Edionwe outlines two current approaches, prevention (more inclusive data outputs) and reverse-engineering (more transparent algorithms), and looks at future challenges as well.

On being "good enough"

Katherine Daniels
"I don’t feel like I have the option of being just good at what I do; if I want to succeed, I have to be great."

In an industry oft-obsessed with “hustle”, the underrepresented can find themselves constantly pressured to prove themselves— to be even better. Daniels highlights that “good enough” is a set of shifting goalposts, and is often not enough for minorities. This implicit expectation of always-on is unsustainable; our focus should be on allowing each other to be successful, whatever shape that success takes.

Diversity in Tech: The Inspiration Gap

Jules Walter
"The problem is not that these role models don’t exist; they just aren’t visible. As a minority, you may not want to stand out because you don’t feel like you’ve accomplished enough yet."

While attending school in Haiti, Walter was told by a teacher that he and others in his class would never get into universities like MIT or Harvard, and the idea was easy to accept. The absence of visible role models can limit our aspirations— a concept Walter calls “the inspiration gap.” While recognising that it’s a big ask, he encourages the underrepresented to make themselves visible when possible, to help even more folks leap across the gap.

Also check out...

Greater than Code podcast with Laurie Voss.

Interesting insights from behind the scenes at npm, bookended by great discussions on non-broken technical hiring, what being a programmer means to us (and to others), and being gay in tech. 63 minutes.

#ChangeTogether: A Diversity Guidebook for Startups and Scaleups

The Change Together guidebook contains strategies designed to acknowledge exclusion in our environments, highlight disproportionate privileged, and center the marginalized.

Enjoyed this issue? Tweet about it!