Amani has had the privilege of engaging with speakers and thought leaders from various industries to share their viewpoints on different aspects of life. Grace Kwak Danciu, was the latest esteemed guest to join us and discuss Artificial Intelligence and Machine Learning.
Grace has been a Product Manager at Google for the past 18 years. She currently works on a team at Google called “Jigsaw,” a unit that explores technology-related threats to open societies. Originally from California, Grace now lives with her family in Zurich. Having a bachelor’s degree in Computer Science from Harvard University was not enough for Grace and she is currently pursuing a master’s degree in Social Innovation at Cambridge University. In our conversation, Grace went into detail on a practical example of a project where Machine Learning was used to assist in addressing human rights issues and the positive impacts thereof.
This project was implemented in collaboration with UPR Info (an NGO focused on promoting human rights), HURIDOCS (an NGO focused on Machine Learning) & Google. The project used machine learning technology to assist in streamlining and providing relevant recommendations made by UN member states on areas where other member states can improve human rights in their countries. This was done through a database called Uwazi where information is carefully sorted. It is also presented in a manner that provides easier access to meaningful insights.
Initially, this work was done manually by a Team that would sift through each recommendation and categorize it in the database. Understandably, this was a time-intensive exercise and with a small team working on it, this resulted in a backlog of uncategorized recommendations. Through this collaboration, they were able to harness the power of machine learning through automation to categorize the recommendations by taking human rights-related text and automatically assigning relevant topics to them. This resulted in a more effective execution of unaddressed recommendations in its backlog, cutting down the amount of time it took to update the database with recommendations. It also turned a three-step verification process into a one step verification process, freeing up time exponentially for the staff of UPR as a whole.