Al Jazeera, RTÉ, and Associated Press, with project participants V-Nova, Metaliquid, QCRI & Tech Mahindra are currently exploring ways to intelligently automate the identification of on-air content using AI, in order to measure and report against metrics set by external regulatory bodies, as well as for accurate measurement of internal editorial, creative and technical standards.
AI indexing for regulatory practice
This project has a first objective of overperforming generic AI indexing solutions in terms of efficiency and/or accuracy, by combining a media-specific face/object detection and recognition algorithm with an innovative multi-resolution compression format. This project has a dual scope:
1) Foundation stage – a key aim is to make indexing faster and more accurate by combining custom AI algorithms with innovative data formats, unlocking savings and real-time/live applications.
2) A Compliance Use Case (Fact-Checking/News Coverage/Political Bias Detection)