Data is essential to scientific discovery, but in many fields high-quality real-world data can be limited, sensitive or expensive to collect. Increasingly, researchers are turning to synthetic data.
Synthetic data is artificially generated information designed to reflect patterns found in real datasets without directly exposing original records.
This can be valuable in areas such as healthcare, where privacy rules may restrict access to patient information, or in autonomous systems, where rare scenarios are difficult to observe naturally.
Researchers can use synthetic datasets to train algorithms, test models and explore hypotheses more quickly.
However, the approach has limits. If synthetic data reflects biases or errors in the original sources, those problems can be reproduced at scale.
Scientists therefore emphasise validation: synthetic data must be carefully compared with real-world outcomes.
The growth of this field shows how science is evolving alongside computing.
In many cases, future discoveries may depend not only on collecting new data, but on generating better versions of the data already available.
