What is Synthetic Data? Plain-English Definition

In more detail

Synthetic data is training material generated by AI rather than collected from people — invented conversations, practice problems, realistic-but-fictional records. Increasingly, new models learn partly from text that older models wrote.

It matters for two opposite reasons. The upside: you can manufacture unlimited examples of rare or private situations — flight-simulator hours for AI — without touching anyone’s real data. The downside: models learning mostly from models risks slowly amplifying quirks and errors, like photocopying a photocopy.

📌 See it in action

A hospital wants AI that summarizes patient notes but can’t hand real records to a training run. So a model generates thousands of fictional patients — realistic symptoms, plausible histories, no actual humans — and the summarizer trains on those instead. Privacy preserved, patterns still learned.

Goes with

Training Data →Distillation →

Now put the vocabulary to work.

Meet the models →Which AI is mine?