01 / The Model
Synthetic data
Artificially generated data that mimics real data. Used to train or test systems when real data is scarce, sensitive or expensive.
How you’ll hear it
“We cannot use real patient records to test this.”
“Then we generate synthetic ones with the same patterns.”
Avoid
Assuming synthetic data is automatically free of bias or privacy risk.
The longer version
Synthetic data can protect privacy and fill gaps, and it inherits the flaws of whatever produced it. If the generator learned from biased or identifiable data, those problems can carry through. Treat it as a useful tool that still needs testing, not as a clean substitute for reality.