ML without when not to use
One common teaching for ML, and Chip start with this discussion, is about when to use and when not to use ML.
I don't know, but, I am thinking that, post-ChatGPT, this is distinct.
It is like just trying to solve everything with AI. Matter of fact, this is the quest now. But LLMs perhaps is not the only way to achieve this.
Reading the Arithmetic chapter from The Advanced Montessori Method and the learning, the initial learning is sensorial, intuitive, manipulating objects. Montessori says that grouping to distinct counting is not so effective, but can the model learn in this way? Interesting for research.