Abeba Birhane, Cognitive Scientist

Born in Ethiopia, cognitive scientist Abeba Birhane works in the nexus of critical race studies, machine learning, complex adaptive systems, and algorithmic bias. Through his collaboration with Vinay Prabhu, Birhane discovered that unpleasant photos and racist and sexist labels were included in large-scale image datasets like ImageNet and 80 Million Tiny photos, which are frequently used to construct AI systems.

Ethiopia is where Birhane was born. The Open University awarded her a Bachelor of Arts in Philosophy and a Bachelor of Science in Psychology. She finished her Ph.D. in 2021 at the Complex Software Lab in the University College Dublin School of Computer Science after earning her Master of Science in Cognitive Science in 2015.

2023

Birhane and Vinay Prabhu, chief machine learning scientist at UnifyID, examined the issues with massive picture datasets’ labeling, categorization, and data collecting in a study released in 2020. Thousands of artificial intelligence algorithms and systems have been developed using these datasets, which include ImageNet and MIT’s 80 Million Tiny Images. Birhane and Prabhu discovered that in addition to unpleasant photos, they included several labels and obscenities that were sexist and racial. As a result, the 80 million Tiny Images collection was voluntarily and formally removed by MIT.

Recently, Birhane has collaborated on studies on the obstacles to data sharing in Africa alongside Rediet Abebe, George Obaido, and Sekou Remy. Even when the data originates from Africa, they discovered that power disparities play a big role in the data sharing process. At the ACM Conference on Fairness, Accountability, and Transparency, their study was published.

Latest news about Abeba Birhane

• Mozilla Foundation has invested $30 million in Mozilla.ai, a startup aiming to create a trustworthy, independent open-source AI ecosystem outside big tech and academia. The company aims to prioritize transparency, accountability, and human agency in AI. Mozilla.ai will hire like-minded founders, developers, scientists, and product managers to create an independent, decentralized, and trustworthy AI ecosystem.
• Binance Labs has partnered with Delphinus Lab to develop an open-source zkWASM virtual machine, enabling trustless computation and application SDKs. This will enable developers to create zero-knowledge applications using their preferred programming languages, making them deployable across WebAssembly-supported environments. Binance Labs plans to invest in Delphinus Lab’s zkWASM-based application rollup platform, zkWASM Hub.

Latest Social posts of Abeba Birhane