Large language models now enable the generation of synthetic training data that can augment or replace real-world datasets, addressing challenges of data scarcity and privacy.

• Prompt-based generation techniques create task-specific examples
• Retrieval-augmented pipelines enhance data quality and relevance
• Iterative self-refinement improves synthetic data accuracy
• Educational applications include creating diverse learning materials and personalized practice examples

For education providers, this research offers cost-effective solutions to develop customized training datasets, generate varied assessment materials, and support personalized learning at scale.

Synthetic Data Generation Using Large Language Models: Advances in Text and Code