Periodic Reporting for period 2 - AI4Media (A European Excellence Centre for Media, Society and Democracy)
Reporting period: 2022-03-01 to 2023-08-31
Following a series of breakthroughs in Artificial Intelligence, new technologies are emerging, ushering innovations in all aspects of business and society, impacting also the media, journalism and politics.
The media market is already benefiting from AI support across the value chain, including tools & services for information analysis, media editing, content generation, content optimisation, content recommendation, audience preferences analysis, etc. Society and politics are also affected by AI while concerns are rising that the combination of the power of major social media platforms and other AI technology companies with the large-scale automation capabilities of AI could prove detrimental to society and democracy.
AI technologies are disrupting the media industry through advances in content creation, analysis, and distribution, by enhancing user experience but also by offering deeper insights into complex social processes that unfold online or offline. AI technology could help shape the democratic role of media by enabling new ways of being informed, of entertainment, artistic creation, political participation, and decision making. In addition, AI could support the relationship between media providers and their audiences, helping to align with the needs of media users, and also cut down operating costs and free up resources for more in-depth and quality work. In addition, AI can create opportunities for the better realization of public values like media diversity, freedom of expression, and inclusiveness.
Motivated by these challenges, risks and opportunities, AI4Media aims to i) deliver the next generation of core AI technologies for the media sector, ii) ensure the development of ethical and trustworthy AI, and iii) reimagine AI as a human-centred, trusted and beneficial enabling technology for media and society.
To reach its goal, AI4Media has set out the following objectives:
• Define Europe's research agenda for media AI, monitor AI impact, provide policy analysis & recommendations, and establish an AI Media Observatory.
• Extend Europe's expertise in AI for media, focusing on four areas: (1) emerging ML paradigms, decentralised learning and AI at the Edge; (2) robustness, fairness, explainability, and privacy in AI; (3) AI for multimedia content analysis & creation; and (4) human- & society-centred AI technologies.
• Contribute to the development of a European ethical brand of AI for media applications.
• Establish a world-recognised PhD programme on AI.
• Support and enrich the European AI-on-demand platform.
• Integrate AI tools in real-world media industry applications and workflows through seven use cases.
• Create a network of partners and associates in academia & industry and a vibrant ecosystem around the project.
• Create a Virtual Centre of Excellence that will disseminate the AI4Media research agenda, provide networking opportunities for R&I and serve as a nexus for the European AI & media ecosystem.
The media market is already benefiting from AI support across the value chain, including tools & services for information analysis, media editing, content generation, content optimisation, content recommendation, audience preferences analysis, etc. Society and politics are also affected by AI while concerns are rising that the combination of the power of major social media platforms and other AI technology companies with the large-scale automation capabilities of AI could prove detrimental to society and democracy.
AI technologies are disrupting the media industry through advances in content creation, analysis, and distribution, by enhancing user experience but also by offering deeper insights into complex social processes that unfold online or offline. AI technology could help shape the democratic role of media by enabling new ways of being informed, of entertainment, artistic creation, political participation, and decision making. In addition, AI could support the relationship between media providers and their audiences, helping to align with the needs of media users, and also cut down operating costs and free up resources for more in-depth and quality work. In addition, AI can create opportunities for the better realization of public values like media diversity, freedom of expression, and inclusiveness.
Motivated by these challenges, risks and opportunities, AI4Media aims to i) deliver the next generation of core AI technologies for the media sector, ii) ensure the development of ethical and trustworthy AI, and iii) reimagine AI as a human-centred, trusted and beneficial enabling technology for media and society.
To reach its goal, AI4Media has set out the following objectives:
• Define Europe's research agenda for media AI, monitor AI impact, provide policy analysis & recommendations, and establish an AI Media Observatory.
• Extend Europe's expertise in AI for media, focusing on four areas: (1) emerging ML paradigms, decentralised learning and AI at the Edge; (2) robustness, fairness, explainability, and privacy in AI; (3) AI for multimedia content analysis & creation; and (4) human- & society-centred AI technologies.
• Contribute to the development of a European ethical brand of AI for media applications.
• Establish a world-recognised PhD programme on AI.
• Support and enrich the European AI-on-demand platform.
• Integrate AI tools in real-world media industry applications and workflows through seven use cases.
• Create a network of partners and associates in academia & industry and a vibrant ecosystem around the project.
• Create a Virtual Centre of Excellence that will disseminate the AI4Media research agenda, provide networking opportunities for R&I and serve as a nexus for the European AI & media ecosystem.
Results achieved so far include:
• AI4Media Strategic Research Agenda on AI for the media industry.
• Overview of EU policies & regulation on AI, complemented by a technical roadmap on AI technologies & applications for the media, an analysis of the societal, economic and political impact of media AI, and relevant policy recommendations.
• Novel techniques for online learning, transfer learning, quality diversity, AI at the edge; for multimedia content analysis & production; and for deepfake detection, political debate analysis, detection of perceptions of hyper-local news, etc. Also, novel tools for AI robustness, fairness, explainability and privacy, complemented by an analysis of the legal & ethical framework for trusted AI, and a prototype platform for AI benchmarking. This research resulted in >330 publications, >30 open datasets and >90 open software.
• Contribution of >100 assets to the European AI-on-Demand platform and >30 models to the AI4EU Experiments marketplace; analysis of platform liability/responsibility for third party content; and organization of the AI-Cafe series.
• Definition of 7 use cases and their requirements to showcase how AI can transform the media. Use case demonstrators endowed with new AI functionalities are developed and evaluated, complemented by white papers offering insights on how to align AI research with media industry needs.
• Establishment of the International AI Doctoral Academy (AIDA) together with VISION, TAILOR, HumanE-AI-Net, and ELISE (78 members, 128 lecturers, 201 students, 60 courses, >140 educational resources) and launch of the Junior Fellows Exchange program.
• Funding of 20 innovative projects, after evaluation of >150 proposals submitted to two AI4Media Open Calls.
• AI Media Observatory and AI4Media Virtual Centre of Excellence.
• AI4Media Strategic Research Agenda on AI for the media industry.
• Overview of EU policies & regulation on AI, complemented by a technical roadmap on AI technologies & applications for the media, an analysis of the societal, economic and political impact of media AI, and relevant policy recommendations.
• Novel techniques for online learning, transfer learning, quality diversity, AI at the edge; for multimedia content analysis & production; and for deepfake detection, political debate analysis, detection of perceptions of hyper-local news, etc. Also, novel tools for AI robustness, fairness, explainability and privacy, complemented by an analysis of the legal & ethical framework for trusted AI, and a prototype platform for AI benchmarking. This research resulted in >330 publications, >30 open datasets and >90 open software.
• Contribution of >100 assets to the European AI-on-Demand platform and >30 models to the AI4EU Experiments marketplace; analysis of platform liability/responsibility for third party content; and organization of the AI-Cafe series.
• Definition of 7 use cases and their requirements to showcase how AI can transform the media. Use case demonstrators endowed with new AI functionalities are developed and evaluated, complemented by white papers offering insights on how to align AI research with media industry needs.
• Establishment of the International AI Doctoral Academy (AIDA) together with VISION, TAILOR, HumanE-AI-Net, and ELISE (78 members, 128 lecturers, 201 students, 60 courses, >140 educational resources) and launch of the Junior Fellows Exchange program.
• Funding of 20 innovative projects, after evaluation of >150 proposals submitted to two AI4Media Open Calls.
• AI Media Observatory and AI4Media Virtual Centre of Excellence.
AI4Media advances the state of the art by performing cutting-edge research on new learning paradigms and AI at the edge, trustworthy AI, AI for multimedia content analysis & production, and human- and society-centred AI, thus strengthening Europe’s position as an AI research powerhouse. The AI4Media Strategic Research Agenda lays out a strategic plan for AI for the media and is a useful reference for AI researchers, media practitioners and policymakers.
In addition, through 7 use cases focusing on different sectors of the media industry (news, video production & content automation, games, content organisation & moderation, human co-creation) but also on significant societal problems (disinformation, media study through social sciences), AI4Media integrates AI technologies in real-world applications thus creating new business models and opportunities for the media sector, while strengthening trust in EU media.
For the news industry specifically, AI4Media delivers key benefits across the whole value chain. Besides simplifying or automating tedious workflows, AI can enable new workflows and services that were previously not affordable or feasible, e.g. storytelling based on open data sources, efficient hyperlocal news, or advanced personalised services. This results in freeing resources for high-value tasks like investigative journalism, creating new knowledge and big data insights, allowing for better reuse of media archives, and improved efficiency and diversity of coverage. At the same time, the tools developed for disinformation detection support journalistic fact-checking and verification and counteract the serious threats that disinformation poses on democratic institutions.
Through the AI Media Observatory, European media, research and policy organisations can have access to high-quality expert analysis and recommendations on AI policy, technology, applications, and impact.
AI4Media strengthens the European AI community by a) making available the project’s research outcomes on the AI-on-Demand platform; b) organizing workshops on different AI topics; and c) funding with 1M euros 20 innovative project ideas through two open calls, supporting researchers and SMEs in developing novel AI technologies and applications for the media sector.
The establishment of the AI Doctoral Academy reinforces Europe’s research capacity in AI and enriches the education offer to equip ICT and media professionals with AI skills. AIDA fosters young European talent and provides them the opportunity to contribute to the development of ethical AI, conduct industry-relevant research, and establish career links that will allow finding employment in the EU.
In addition, through 7 use cases focusing on different sectors of the media industry (news, video production & content automation, games, content organisation & moderation, human co-creation) but also on significant societal problems (disinformation, media study through social sciences), AI4Media integrates AI technologies in real-world applications thus creating new business models and opportunities for the media sector, while strengthening trust in EU media.
For the news industry specifically, AI4Media delivers key benefits across the whole value chain. Besides simplifying or automating tedious workflows, AI can enable new workflows and services that were previously not affordable or feasible, e.g. storytelling based on open data sources, efficient hyperlocal news, or advanced personalised services. This results in freeing resources for high-value tasks like investigative journalism, creating new knowledge and big data insights, allowing for better reuse of media archives, and improved efficiency and diversity of coverage. At the same time, the tools developed for disinformation detection support journalistic fact-checking and verification and counteract the serious threats that disinformation poses on democratic institutions.
Through the AI Media Observatory, European media, research and policy organisations can have access to high-quality expert analysis and recommendations on AI policy, technology, applications, and impact.
AI4Media strengthens the European AI community by a) making available the project’s research outcomes on the AI-on-Demand platform; b) organizing workshops on different AI topics; and c) funding with 1M euros 20 innovative project ideas through two open calls, supporting researchers and SMEs in developing novel AI technologies and applications for the media sector.
The establishment of the AI Doctoral Academy reinforces Europe’s research capacity in AI and enriches the education offer to equip ICT and media professionals with AI skills. AIDA fosters young European talent and provides them the opportunity to contribute to the development of ethical AI, conduct industry-relevant research, and establish career links that will allow finding employment in the EU.