Watch this video on YouTube
GitHub Copilot is trained on a large dataset of publicly available code. How does the prevalence of commonly seen examples in the training data affect the code suggestions Copilot generates?