HumAIne has initiated significant technical and scientific activities aimed at building a reliable and adaptable platform for Human-AI collaboration. A key milestone has been the specification of a comprehensive Reference Architecture for Human-AI Collaboration in Dynamic Environments, encompassing an ecosystem for integrating heterogeneous AI paradigms covering the AI lifecycle. This architecture is grounded in detailed functional and non-functional specifications and is being refined through ongoing alignment with real-world use cases.
To support transparency and trust in AI systems, the project has been working on the specification of a library of Explainable AI (XAI) techniques, a benchmarking suite for evaluating Human-AI models, and human-machine interfaces. These tools are designed to facilitate interaction between users and AI systems, supporting both technical and non-technical stakeholders.
Progress has also been made in defining the core AI platforms that underpin the HumAIne approach. These include platforms for Active Learning, Swarm Learning, and Neuro-Symbolic AI—each designed to enhance adaptability, collaboration, and knowledge integration in AI-driven decision-making.
In parallel, the project has outlined an Open Integration Platform for Human-AI collaboration and hybrid decision making, and developed training resources through a dedicated Learning Management System. These efforts aim to support the deployment and adoption of the platform by developers, researchers, and end-users.
Finally, we carried out a first validation of six pilots across diverse domains, which can integrate the different components that are being developed within the HumAIne platform. These pilots cover the sectors of: manufacturing, smart cities, healthcare, ticketing, and energy, and they are providing valuable insights into the platform’s applicability and user experience, helping to guide further development and refinement.
