For PhD students in machine learning, academia vs. industry is an increasingly strange choice. Getting a faculty job at a good university is still quite hard, but the pay is hugely lower than in industry (~ 120k for an assistant professor job at a top 10 school, vs. ~ 500k for a research scientist position at Google/FB/Amazon for a similar candidate).

What is the allure of academia? It comes with grant pressure, teaching huge classes, and huge opportunity cost in salary.