AI-Ready Data and Trustworthy Pipelines for Europe’s Next Wave of Industrial AI
Europe has strong research capabilities, industrial expertise and a growing portfolio of initiatives addressing efficient and trustworthy AI. However, many AI solutions remain confined to isolated pilots and struggle to progress towards scalable, reusable and trustworthy deployment. A key challenge is the fragmentation of data, tools and infrastructures across organisations, sectors and stages of the AI lifecycle. Data may not be sufficiently accessible, interoperable or fit for AI, while the processes for developing, validating, deploying and monitoring AI models are often disconnected.
This session, organised by the AI-DAPT project, will explore how Europe can move from AI-ready data to trustworthy AI operations. It will examine the technical and organisational capabilities required to transform heterogeneous data into reliable AI solutions and to manage those solutions throughout their lifecycle. In particular, the session will present the AI-DAPT integrated approach to data preparation and AI pipeline orchestration, including its human-in-the-loop strategy. A real-life AI-DAPT pilot will illustrate how these capabilities can be applied beyond controlled research environments, providing practical insights into the challenges of integrating data and AI pipelines into operational settings, involving domain experts throughout the process and adapting solutions to evolving requirements.
The session will open with introductory remarks from Nuria de Lama, Secretary General of BDVA, followed by short presentations on AI-DAPT’s approach and pilot. A 30-minute panel discussion entitled “From AI-Ready Data to Trustworthy AI Operations” will bring together representatives from AI-DAPT and some of its sister projects funded under the HORIZON-CL4-2023-HUMAN-01-01 topic, namely MANOLO, PANDORA, TURING, RAIDO, and EXTRA-BRAIN. The panel will broaden the discussion, with panelists sharing their experiences of moving beyond individual models and pilots, addressing questions such as:
- To what extent are the approaches followed by the different projects aligned, and how can interoperability between their solutions be promoted?
- What mechanisms are needed to maintain data quality, model performance and trustworthiness throughout the AI lifecycle?
- How should human oversight be integrated into the development and operation of AI systems?
- Which technical and organisational barriers continue to limit adoption at scale?
The session is intended for representatives of European research and innovation projects, industry, research and technology organisations, data spaces, AI Factories, Testing and Experimentation Facilities, Digital Innovation Hubs, policymakers and organisations developing or adopting data-driven AI solutions. The ultimate goal is to identify practical pathways for turning high-quality data and experimental AI models into scalable, trustworthy and reusable AI operations with measurable value in industrial and societal contexts.
Structure of the session:
- Introduction and setting-up the scene
- The AI-DAPT approach with short presentations covering:
- AI-DAPT solution,
- The human-in-the-loop approach
- Health Pilot and lessons learned.
- Panel: “From AI-Ready Data to Trustworthy AI Operations”
- Audience discussion and conclusions
Session slides
