

#274: Real Talk About Synthetic Data with Winston Li
Jun 24, 2025
Winston Li, founder of Arima and a former data science leader at PWC Canada, dives deep into the fascinating realm of synthetic data. He discusses how synthetic data can revolutionize analytics by safeguarding privacy while providing robust alternatives to traditional datasets. The conversation covers its generation, applications in marketing mix modeling, and the balance between statistical integrity and privacy concerns. Li also shares insights on its utility in consumer research and the modern implications of AI in education and career choices.
AI Snips
Chapters
Transcript
Episode notes
What Is Synthetic Data
- Synthetic data is algorithmically generated to statistically mimic real data patterns.
- It is not fake but derived from learnings from real-world datasets.
Synthetic Data as Legal Alternative
- Synthetic data enables carrying out analyses restricted by privacy or legal issues with real data.
- It acts like a photocopy, offering similar utility without privacy risks.
Maintain Models To Avoid Bias
- Continuously maintain and update synthetic data models to avoid bias and staleness.
- Use current and trustworthy data sources as ingredients for generating synthetic data.