Seulement 20% des biographies sur Wikipédia en anglais concernent des femmes. Pour aider à apporter plus de représentation sur Internet, nous sommes …
– When you think about human history, right? A lot of times it’s been written by a certain type of people, often men. The victor tells the story. The world around me that I see is a lot more diverse than what I see on Wikipedia. I want everyone to be able to go
To Wikipedia and see something that they value. My name is Angela Fan, I’m a researcher at Meta AI, working on solving the problem of representation on Wikipedia using artificial intelligence. For me, it really started like. quite early on. So when I was in school, I think I wanted to write a biography
About Eleanor Roosevelt. And I remember thinking like, okay, there’s a lot of books but there’s mainly books about men. And so that kind of idea, like, it stayed with me throughout life. It still manifests today, right? But I don’t think people are like, going to the library to read the like,
Written encyclopedias anymore. If you wanna be inspired by someone, if you wanna know about something, Wikipedia is really like, one of the default knowledge sources on the web. In a month, over 1 billion unique devices access Wikipedia. There’s very few articles about women, maybe only one in five, very little representation
Of non-binary people or people in the LGBTQ Plus community, very little representation of people outside of the Western world. – Representation matters. We, as editors, are mostly white, mostly middle class, mostly in Western countries. So senior editors definitely have a role in this and in trying to challenge their own biases.
My name is Alexandre Sirac, I’m a French Wikipedia editor and the localization editor for French at Meta. – Near the end of my PhD, I was kind of thinking, you know, I really want to work on things that speak to the problems that I’ve personally experienced. At Meta AI research,
We have the freedom to follow our curiosities and unearth research breakthroughs without being tethered to like, a specific product launch or a corporate goal. Donna Strickland, she won a Nobel Prize but actually she did not have a Wikipedia article. And the reason was because there wasn’t a lot
Of content on the web about her. – All of Wikipedia is based on sources, external sources. So we completely depend on the biases of the current reporting landscape. – You shouldn’t need to win a Nobel Prize to have a Wikipedia article. This case of biographies, like it’s so meaningful to me
But how do we tackle the problem of representation on English Wikipedia itself? Text generation has the potential to write like, entire paragraphs. Could this be a way for people to write Wikipedia articles basically with a draft first? The technology that we develop tries to mimic the human process of writing a Wikipedia article.
First, we create a dataset. We use it to teach our machine learning model through pattern recognition like, what kind of biographies we want to produce. We have technological tooling to make writing Wikipedia articles much faster and allow people who want to kind of focus on challenges of representation to feel empowered
To make changes to Wikipedia. If you want it to be different in the future, like, we need to start thinking about that now. I want people in my field to kind of engage in this task, to understand that when we take on the challenge of long text generation, there’s a lot of opportunity
For people to represent their own diversity and their own unique viewpoints and bring it online. – Everyone can edit Wikipedia, everyone can bring their contribution. And that’s the whole strength of Wikipedia, as much as it is a way to have testimonials of how the world was, is, and will be.
– One of the things that people talk about when they talk about career advancement is like, having a role model. Like, see other people who might have their similar stories and be inspired by those people, understand more about their work and kind of ultimately, see yourself in other people.
And so I think that’s why representation, particularly on Wikipedia biographies is so important.