This post deserves a more engaged response. I believe deep learning (and statistical/machine learning) presents a good reason for introspection about current linguistic theory, in part because I find the computational model espoused in neural networks to be more biologically plausible than single CPU symbolic computation.

(2019)