I hadn’t even read the paper when I read the title: I knew I was wrong on the spot. I now believe Deep Learning almost certainly works.
An exact mapping between the Variational Renormalization Group and Deep Learning
https://arxiv.org/abs/1410.3831