[array_key_first]

The payout is very, very long term and honestly you probably don't come out ahead.

Starting your career 10 years earlier (effectively) and immediately progressing in your career probably gets you further. Especially if you look towards management, which is where the money is (usually, million dollar AI research salaries is a new thing).

[aprilthird2021]

I don't think this is always true. AI research big payouts aren't even a new thing. It was common 10 years ago for big name professors to get hired by Google etc. and bring the whole lab with them.

Also a PHD is not 10 years long? You start your career 4-5 years after a typical BS in CS

[michaelt]

Well, the big money is in becoming a leader in a field before it's hot - making the big payoffs very speculative.

In the mid 2000s, natural language modelling was a joke and the best performing ML systems for sentiment analysis would lose in benchmarks to emoji counters. Today, people with a PhD in ML language modelling and years of experience delivering projects in industry are finding the PhD in ML really pays off.

But what about someone who spent the mid 2000s working on formal methods for static analysis? Or a compiler responding to the challenges created by Intel's Itanium architecture? Or trying to fit FPGA accelerator boards into the niche CUDA fills today?

Well, honestly their career's probably still going fine, they're a smart person and there's been many years of high demand for competent programmers. But industry isn't beating their door down; the beneficial effects of the PhD will be a lot less obvious.

Honestly a lot of PhDs go to people from cultures that prize education; and to people from upper-middle-class backgrounds who've been brought up to do well in school, follow their passions, no need to worry too much about money.