The professor whose student worked on this problem is a network theorist. He's never done string theory and has no need to "rebrand as machine learning expert". So yeah I don't buy it either.