Project Overview
The news infrastructure is the gateway to a well-informed public. Access to a balanced news offering from professional journalistic organisations should not be hampered by digital barriers created by search engines and social media. In addition, journalism does not succeed well in connecting its products to the user needs of a section of society. There is a lack of genuine attention for and direct involvement of the public in the creation of the journalistic offering. Could AI play a role in better meeting the needs of users from different target groups?
Approach
A central question of this research is to uncover whether a community annotated audience model could support the journalistic sector in creating more inclusive news reporting and, if so, under what ethical and legal conditions this would have to take place. This project aims to establish an AAVA (AI Audience Validation Assistant), that helps journalists identify harmful content in reporting. This project further investigates under what conditions training data can be inclusive and if a model trained on this data may be able to contribute to a better representation of a pluralistic society, including vulnerable or unreachable users.
Methodologically, this project will work with focus groups, community annotation, and data analysis.
Expected outcomes include reports, publications, and workshops with media professionals to support the sector in responsible AI applications, helping create content that resonates with diverse audiences.
This project is still ongoing. More information will follow.
This research is made possible in part by a financial contribution from the Dutch Journalism Fund.
