The European Broadcasting Union (EBU) has launched the AI Hub, a new platform designed to harness artificial intelligence (AI) technology for public service media organizations. This initiative aims to enhance how EBU Members use AI to better serve their audiences through customized solutions based on open-source models. The AI Hub, developed by EBU’s Artificial Intelligence and Metadata (AIM) group, fosters a collaborative ecosystem where AI and media experts can co-create and refine AI models tailored specifically for various media applications.
The AI Hub is designed to accommodate the fast-paced evolution of technology, allowing EBU Members to adapt by utilizing robust open-source models as foundational tools. These models are instrumental in a variety of tasks, from content analysis to content creation, thus addressing diverse media needs in an efficient manner. One of the significant features of the AI Hub is its provision of private spaces for testing AI models, maintaining the confidentiality required when working with proprietary content. Additionally, it includes a user feedback system for evaluating AI tools and a discussion forum to encourage collaboration between AI users and developers.
At launch, the AI Hub features three key projects aimed at enhancing media experiences: the MetaRadio project, a facial recognition system for television programs, and a fake news detection tool. The MetaRadio project enriches radio content by incorporating advanced metadata, thus transforming user experience through improved navigation and discoverability. Originating from a collaboration with Radio France, the project utilizes annotated datasets to detect key questions in radio programs while benefiting from translation support from the EBU’s EuroVOX initiative.
The facial recognition project is tailored specifically for television content and focuses on identifying significant participants in broadcasts while excluding incidental characters. This system is developed with privacy considerations in mind, ensuring compliance with the General Data Protection Regulation (GDPR). Collaboration on this initiative involved prominent broadcasters such as Rai from Italy and RTS from Switzerland, leveraging a unique dataset of annotated videos to enhance content relevance during broadcasts like political debates.
In addition to these projects, the AI Hub includes a sophisticated fake news analyser that employs natural language processing and machine learning. This tool analyzes the linguistic attributes of news articles to identify misleading information and assigns reliability scores based on various linguistic features. Together, these tools exemplify how AI can enhance the richness and accessibility of media content, illustrating a proactive approach by the EBU to tackle the challenges of misinformation in journalism.
Through the introduction of the AI Hub and its initial projects, the EBU aims to create an environment that promotes knowledge sharing and collaboration among its members. By embracing open-source AI technologies, members are encouraged to participate in ongoing developments and future expansions of the AI Hub. The goal is for the media industry to not only maintain its competitiveness in a rapidly evolving landscape but also drive growth in a responsible and innovative manner. EBU Members interested in participating in these initiatives are encouraged to join the AIM group for further engagement and information.