Fairlearn is an open-source toolkit designed to assess and mitigate unfairness in machine learning and AI models. It emphasizes that fairness is a complex sociotechnical challenge, requiring both technical and contextual understanding to ensure AI systems do not cause harm.
Through a collaborative design thinking approach, we began by engaging with the global Fairlearn community, inviting participants to share images they felt represented their cultures or those they believed should be included. Each submission came with personal insights on the significance of the image and its alignment with Fairlearn’s mission.
We moved away from traditional, monochromatic character images to create a vibrant and inclusive visual language that breaks down gender binaries and represents a spectrum of identities. This transformation was pivotal in aligning the website’s visuals with Fairlearn’s mission.
Our final product is an updated Fairlearn website featuring a library of assets that reflect the diverse and inclusive nature of the community it serves. By engaging directly with participants and incorporating their perspectives, we ensured that the website not only meets but exceeds the expectations of both Microsoft and the global Fairlearn community.
