The article discusses the learnability of different types of functions in a model of learning. It shows that learning Disjunctive Normal Form (DNF) expressions is a challenging problem. The researchers prove that learning DNF expressions is difficult even with the ability to ask questions. They also present a new algorithm for learning DNF expressions efficiently. Additionally, the article explores the learning of parities with random noise and shows that it is a key problem in learning theory. The researchers demonstrate that learning monomials is also a challenging task.