Synthetic Data: Servicing Privacy

Originating in experiments with computer vision in the 1960s (Nikolenko, 2021), synthetic data have recently gained traction in the data industries. The reasons for the growing interest in synthetic data lie at the heart of well-known issues with data from real-world systems: gender and racial bias in data sets, collection and processing costs, and notably new policy requirements concerning the use of personal information. Unlike organic data, synthetic data are generated in algorithmic models. They are presented as an alternative to manually collected and processed data. Synthetic data can thus be created as novel output by systems of artificial intelligence...