To be fair, it's a review article in nature -- so meant to advertise deep learning to scientists in other fields. His actual research is still publicly accessible.