To me that is just an iteration on first you makes it run then you make it right. And to make it run you start by the simplest approach. And building your own model is generally not the simplest however, it can be. There are some areas where you should start with ml. Most importantly Vison and some NLP, whenever a pretrained model for your task exists.
This is correct. But i guess the article argues about not already solved problems. NLP is in the most cases so powerful and easy to implement, that i would argue it can be viewed as a more complex version of a heuristic. My thought comes from the idea, that you need to act up on the data NLP-Algorithms bring to you.
Yea I get that but I have experience people working with all sorts of insanely complicated heuristic to get something like a NER system running when they could have much more easily used a Hugging face model. But I totally agree that the article holds true if you have to train your own model.
