I think this isn't really giving credit. Standard deep learning is not really slowing down. It is improving exponentially still.

Just yesterday I was looking into new deep learning methods for solving PDEs. That is a huge area to explore and will change many things dramatically