The Philosophical Significance of Deep Learning

Abstract

    Artificial intelligence research has made impressive progress in the last ten years with the development of new methodologies such as deep learning. There are several implications of the progress both for philosophy of cognitive science and philosophy of artificial intelligence, but none are conclusive. Though its success seems to support connectionism in cognitive science, there are several features of human cognition that remain to be explained. Also, though it is often said that deep learning is the key to build artificial general intelligence, deep neural networks we now have are specialized ones and it's not clear how we can build a general artificial intelligence from such specialized networks.