Are there machine learning subfields that are becoming irrelevant (or is irrelevant)? [D]

A discussion on machine learning subfields questions the relevance of Neural Architecture Search (NAS). Despite over 3000 new models proposed within five years, the transformer, a significant development, was not discovered through NAS, leading to its quiet decline. This raises concerns about continued research in NAS, contrasting with emotional arguments that all approaches, like SVM, LDA, or Markov chains, might eventually regain prominence, akin to the unlikelihood of vacuum tubes becoming popular again.