Decentralized AI refers to artificial intelligence systems in which data, computing resources, models, or decision-making are distributed across multiple independent devices or participants rather than controlled by a single central system. Techniques such as federated learning can allow models to learn from data held on separate devices while reducing the need to move that raw data to a central repository.
This approach can be useful when organizations or devices need to collaborate without fully sharing their underlying data. For example, several institutions could contribute to training a model while keeping sensitive datasets within their own environments. Decentralized AI is associated with federated learning, edge AI, distributed computing, blockchain, and privacy-preserving machine learning.
