Although research on discrimination and inequality in artificial intelligence does exist, most of it was conducted by Anglo-Saxon researchers and used algorithms implemented in the United States as concrete study cases. In line with its objective to support multidisciplinary research projects that mobilize data sciences, and to catalyze - in France and Europe - the reflection about equity in artificial intelligence, Fondation Abeona has carried out, in partnership with researchers in computer science and economics from Télécom Paris, a research project on algorithmic biases.
Following this first report, Fondation Abeona joined forces with the Caisse des Dépôts et Consignations (CDC) as part of “Operational AI Ethics research chair” at Telecom Paristech to detect bias risks and optimize equity in Mon Compte de Formation, a search engine and database dedicated to professional training.
Building on the study carried out with Télécom Paris, the Foundation has published a report in partnership with Institut Montaigne which aims to provide concrete solutions to limit potential abuses and restore confidence in algorithms.
Among the main recommendations, the report advocates third-party auditing of giving public high-impact algorithms: Fondation Abeona and Institut Montaigne continue this work by setting up a task force dedicated to examining this recommendation in the context of the upcoming European “AI Act” regulation.