The speaker stresses the importance of train models, fine-tune them with your dataset, and run experiments to gain an intuitive understanding of inputs and outputs within machine learning. They have seen a positive reaction from others regarding their ML course, which further validates this methodology.