A few thoughts:<p>* the economics of staying in academia are just <i>so amazingly awful</i>, the decision to go into industry is a no-brainer for most anyone along the academic path -- grad student, postdoc, instructor, whatever -- who stops, looks around, and realizes (better late than never), that they'd like to actually make a decent income before they reach old age, maybe actually start a family, etc. Academia is effectively pushing talented people out, and industry will happily put their skills to use selling advertisements (instead of them researching cancer at a university or whatever).<p>* For ML/AI work, you need to have a strong quantitative skillset. Almost any graduate degree will provide a sufficient quantitative/statistical background (whether it's physics, sociology, whatever), hence employers find it a useful filter for such roles. If one only has a bachelor's the ML/AI market right now will be tougher but I wish those folks success.<p>Personally, as to whether someone should pursue grad school, regardless of future plans, my feeling it's usually a rip-off, but it depends on one's own situation.