Advancing Trustworthy AI in Milan
On March 13-14, 2025, the AI-DAPT consortium gathered in Milan for the 3rd Plenary Meeting, bringing together experts from across Europe to push forward our mission of trustworthy, explainable, and efficient AI-driven automation. Over two days, we engaged in interactive discussions, hands-on demonstrations, and deep technical sessions, focusing on key advancements in AI pipeline design and execution, hybrid science-AI models, and real-world applications in health, robotics, energy, and manufacturing.
Key Discussions & Technical Advancements
Day 1: Project Management & Use Case Demonstrations
The meeting kicked off with a welcome session and project management updates by ATHENA, ensuring all partners were aligned on progress and next steps.
The first major focus was on demonstrator use cases, where partners presented ongoing work and discussed key requirements for each domain:
🔹 Health Demonstrator (CHARITE): Detailed discussions on data availability, pipeline scenarios, explainability methods, and hybrid modeling approaches to ensure the AI systems are robust and transparent in medical applications.
🔹 Robotics Demonstrator (MADE): Showcased automation strategies, AI lifecycle considerations, and real-world implementation of robotics-driven AI solutions.
🔹 Energy Demonstrator (ZENITH): Explored AI-driven energy management solutions, emphasizing scalability, security, and real-time decision-making.
🔹 Manufacturing Demonstrator (BIBA): Addressed AI deployment in industrial automation, focusing on pipeline design, execution and validation for ensuring efficiency and reliability.
These interactive sessions allowed all partners to refine technical strategies, align on AI pipeline requirements, and address challenges related to data, security, and deployment.
Ethics, Outreach & Administrative Updates
In the afternoon, the spotlight was on AI ethics, communication, and dissemination, led by UNINOVA (WP6) and ATHENA (WP7). Discussions revolved around:
- The ethical implications of AI automation and ensuring fairness, transparency, and accountability.
- AI-DAPT’s communication and dissemination efforts, ensuring that our work reaches the right audiences and contributes to global AI innovation.
- Administrative updates to maintain strong project coordination across all partners.
The day concluded with a networking dinner at Don Lisander, allowing participants to engage in informal discussions and strengthen collaborations.
Day 2: AI Foundations & Technical Integration
The second day focused on the foundational AI components, platform integration, and hybrid AI models:
🔹 WP2 (SUITE5): Clarified interaction between AI components, addressing integration challenges within the AI-DAPT core platform.
🔹 Hands-on Demos (UNINOVA & ATHENA): Live demonstrations on integrating external components, user-interactive services, and AI pipeline workflows using Airflow-based platforms.
🔹 WP4 (UBITECH): Final decisions on system architecture, tech stack, and continuous integration strategies.
🔹 WP3 (ATHENA): Focused on advancing hybrid science-AI models, ensuring explainability and robustness in AI decision-making.
The plenary wrapped up with a discussion on next steps, reinforcing AI-DAPT’s commitment to building an AI ecosystem that is efficient, ethical, and impactful.
What’s Next for AI-DAPT?
As we move forward, AI-DAPT will:
✅ Continue refining demonstrator implementations and integrating AI components.
✅ Strengthen AI ethics, governance, and security frameworks.
✅ Develop explainability and fairness mechanisms to enhance trust in AI systems.
✅ Expand outreach and engagement to showcase our work on trustworthy AI.
✅ Update the AI-DAPT Reference Architecture, ensuring a robust, scalable, and standardized framework for AI pipeline development.