A nice Video from the very Andrew Ng on data quality: A Chat with Andrew on MLOps: From Model-centric to Data-centric AI. Spoiler: Data quality is more important than model quality in ML. How does one find datasets of high quality? Perhaps through FAIR principles?