»AI Spaces«

Content

The paper introduces AI Spaces as a framework that enables AI systems to collaborate across organizational boundaries while addressing constraints such as data sovereignty, confidentiality, regulation, and misaligned incentives. Many real-world challenges, including those in supply chains, manufacturing, healthcare, and smart buildings, cannot be solved by a single organization because relevant data and knowledge are distributed. AI Spaces therefore enable the secure and sovereignty-preserving use of distributed data without requiring centralization. The framework describes three main forms of collaboration: joint model development, for example through federated learning, the use of distributed data for AI-driven reasoning at inference time, and the coordination of autonomous AI agents across organizations. The paper emphasizes that sustainable implementation depends not only on technical feasibility but also on well-designed incentives, interoperability through standardization, and robust quality management.

 

Authors

• Boris Otto (Fraunhofer ISST) 
• Tobias Moritz Guggenberger (Fraunhofer ISST)
• Julia Pampus (Fraunhofer ISST) 
• Takahide Matsutsuka (Fujistu Research) 
• Janosch Haber (Fujitsu Research) 
• Noboru Koshizuka (The University of Tokyo) 

 

Partners

• Fujistu Research 
• University of Tokyo