Machine Learning should be data-centric, not model-centric. Here’s why.

The text emphasizes the importance of focusing on data quality rather than model complexity in machine learning (ML). It argues that a "garbage in, garbage out" approach applies to ML as well, and improving data quality can lead to better outcomes even with simpler models. The author criticizes the industry's obsession with complex models and highlights how this tendency often overlooks fundamental data quality issues. They propose a shift towards data-centric ML, which prioritizes data cleansing, pre-processing, balancing, and augmentation over hyperparameter selection and architectural changes. The text also discusses the importance of monitoring data quality and improving it continually.