Artificially Intelligent Nipples

Gender Diversity in Machine Learning

If you’ve been reading the news lately, it can feel like every part of artificial intelligence is a decisive subject. Should research be public or private funded? Will it help our economy or will we be enslaved by robot overlords? Should we regulate it or let it exist freely? However, there is one issue not enough people are paying attention to: how women are treated in the field.

I genuinely believe that artificial intelligence will be humanity’s greatest invention and that it will shape our race for centuries to come. So, there is clearly a lot at stake. Diversity in machine learning research is something we have to think hard about. If you raised a child in the absence of women, it probably will not learn to treat them properly. Why should we think about artificial intelligence any differently?

All the hype in machine learning attracts the brightest and most promising individuals in computer science. However, meritocracies like this can create problems in diversity. Machine learning (the recent practical surge in the belief of AI) is often a hostile environment. Just getting started with the subject can be stressful, requiring pre-requisite knowledge in otherwise rare forms of differential linear algebra and extensive schooling in computer science. In addition, publishing is often a crapshoot with thousands of submissions to top-tier publication venues (kinda like university admissions). As a male, I feel the stress of academia’s publish or perish mentality, but I can only imagine it when sexism is thrown into the mix.

The amount of women in computer science, although improving, is still pretty low. Even worse, the lack of diversity becomes amplified when zooming into the sub-field of machine learning. I don’t want to discredit all the progress we’ve made, though. The tech community has made enormous progress with a variety of programs and events to celebrate women in the tech industry. Recently, some new progressive inclusions were made through the workshops Women in Machine Learning and Black in AI.

In this blog post, I want to talk about one specific conference in machine learning - arguably one of the most competitive and influential events in all of tech: the Neural Information Processing Systems conference. More specifically I want to discuss its official short hand name.

Originally, it used to be called NIPS.

I’m writing this blog post because the NIPS-fiasco is a relatively unheard of issue both within computer science and the greater academic community. However, the implications are large. Even if it’s only the name of a conference that is offensive, it’s also the precedent that is set regarding the power of minority voices in a largely male dominated field.

I really like the lab I do research in. Almost half of our team are women and a majority are people of colour and we definitely have fun. But most importantly, we do interesting work to help computers understand human language. And if women aren’t able to contribute equally, we may never reach this goal.

Read more posts or learn more about me.