I reminds me of study methods that involve "memorizing" as opposed to thinking about the problem, and applying various methods to solve it. It can solve some al

I reminds me of study methods that involve "memorizing" as opposed to thinking about the problem, and applying various methods to solve it. It can solve some algorithm problems at a certain level, but struggles at a detailed answer in some scenarios. I guess it's not surprise, given ML is essentially exploring the local solution space of some abstract problem; it reminds me of a Taylor series expansion or a breadth-first search of an abstract problem space.