AI Definitions: Synthetic Data

Synthetic Data – Instead of giving real data to LLMs for training (some experts say we are running out of original human data) there’s an idea that LLMs can be told generate data, synthetic data, on which it can be trained. If synthetic data can be made to work, it could negate the problem of using copyrighted material for training.  Sceptics say this will lead to a degradation of the data, weakening the performance of the model.

More AI definitions here.