What is AI4media about?
The AI4media initiative with a total of 30 partners aims to become a European "centre of excellence" that delivers the next generation of AI advances for the media sector. Apart from the development of new technology, the project wants to make sure European values of ethical and trustworthy AI are embedded in future AI deployments.
The project will be setting the research and innovation agenda of the field, while also making concrete contributions to relevant research and innovation areas. During the project, more external researchers and collaborators will join via funding and exchange programmes. A collaborative PhD program will develop a new generation of talent for Europe, also bringing together academia and industry.
Key AI4Media research areas include:
New learning paradigms and distributed AI (e.g. manifold learning, decentralised graph mining, distributed heterogeneous learning, and on-line knowledge transfer)
Explainability, robustness and privacy in AI (e.g. training of robust models, assuring explainability in all instances, and fairness in the light of changing social standards)
Content-centered AI (e.g. audio-visual content understanding and analysis)
Human- and society-centered AI (e.g. online communities, social network analysis, polarisation/hate speech/disinformation, and AI in computational politics).
AI4Media is coordinated by the Multimedia Knowledge and Social Media Analytics Lab of CERTH.
What is DW Innovation's role in AI4media?
In cooperation with ATC, DW will run one of the AI4media use cases: We'll focus on the needs of journalists, who require more powerful features for the tools currently used to verify content and detect disinformation. This will involve tests concerning journalists' and decision-makers' trust in AI-driven support tools and information.
The use case will extend the Twitter analytics tool TruthNest (ATC) and the collaborative verification platform Truly Media (ATC/DW) with advanced AI-based functionalities, ensuring that they are coupled with features related to "Trusted AI". User classification, detection of user similarities, synthetic image detection, and trend detection/monitoring are among the planned new functionalities.
Our use case work includes collecting requirements, co-designing prototype AI functions, and user testing in the context of news journalism, content verification and legal/ethical frameworks for AI (e.g. related to explainability, transparency, fairness, privacy, and trustworthiness).
With regard to the wider context of using AI in news media, we'll also make contributions in the area of policy recommendations and the social-economic impact of AI.
Finally, we'll help define other use cases evensioned for the project (seven in total), all of them media-related.
AI4media links–and a workshop!
AI4media official website (currently under construction): ai4media.eu
AI4media on Twitter: @AI4media_EU
AI4media is hosting a free online workshop (on Generative Adversarial Networks for Media Content Generation) on October 1st. Join us: AI4media GAN Workshop