RightsCon Toronto 2018

Social media platforms can be a toxic place for women. The widespread inequality and discrimination against women that is embedded in our society does not simply disappear when you click open your internet browser or tap on your Twitter mobile app. Amnesty International’s research shows that women from all walks of life face violence and abuse online. But exactly just how much of our twitter feeds and social media timelines are filled with such content remains unclear. Organizations, academics, as well as companies themselves, have turned to machine learning to analyse and detect patterns and trends of abuse on social media platforms to help tackle this ever-growing problem. Following the June 2017 General Election in the UK, Amnesty International developed an algorithm to detect online abuse against women parliamentarians. The findings demonstrated the extent of abuse that women in politics are subjected to as well as the intersectional nature of online abuse. At Wikimedia, thinking about how to create systems that are transparent for users and also facilitate user participation to tackle online harassment continues to shape their approach to combating online harassment. In 2017, Google Jigsaw launched Perspective API which uses machine learning to help identify and moderate toxic comments. The proposed panel discussion will include participants from Amnesty International, Wikimedia, Google Jigsaw and Element AI about using machine learning to combat online abuse. In addition to speaking about how the respective individuals have used machine learning in their past and present work, the panel will also be an opportunity to discuss the successes, limitations and future developments in using machine learning to combat violence and abuse online.