Collaborative Intelligence Databases (CID): Harnessing AI for Privacy-Preserving Multi-Source Data Management

The proliferation of data in distributed platforms has brought enormous opportunities for collaborative intelligence and, at the same time, generated a major concern in the area of privacy. The concept that will be presented in this article is the notion of Collaborative Intelligence Databases (CID), which is a new model that uses artificial intelligence to facilitate privacy-conscious and secure data management across jurisdictions. CID structures enable institutions to combine the use of heterogeneous data sources to derive insights by applying a suite of sophisticated cryptographic techniques, including homomorphic encryption, secure multi-party computation, differential privacy, as well as a federated learning architecture. The survey provides an overview of the existing state of the art in privacy-conscious collaborative data management, defines the basic elements of technology, and classifies possible implementation plans in various sectors such as health care, banking, transportation, and industrial Internet of Things. As our analysis has shown, deep technical challenges still exist, but recent developments have proven that it is possible to establish large-scale collaborative intelligence and, at the same time, provide high privacy guarantees.