[acdha]

I think that’s a bit harsh – but only so much, as I’ve had to explain basic stats literacy to several consultants – but there’s a common factor to many “AI” initiatives: some executives really want to be able to tell their buddies that they’re doing amazing things, and often believe there are huge untapped opportunities to make or save money which all of the employees they don’t trust haven’t mentioned.

In reality, of course, there probably just isn’t that much money at play so the high-budget approach is unlikely to break even, and more critically it’s unlikely that people are prepared to change how the organization makes decisions. I’ve known some people who worked for places where they had data showing solid improvements but it either hit politics or was simply small enough that their executives didn’t want to go with it because it wasn’t the magnitude they’d promised to justify the data science program.

[noirbot]

It's definitely harsh, and I want to be clear that I don't think it's innately a bad idea or destined to fail, but when it's easy to throw millions of dollars at a moon shot, trying to spin data straw into gold is worth a shot. It's similar to people trying to make a self-driving car, and how many of those are struggling/ending recently. It's not that there's nothing there, but when loans aren't free, you can't just throw money at it and hope.

It's a field that's just been over-hired because it's been a field where you can go "we're 2 years away from something big" for 10 years and the money was cheap enough to just keep paying people with questionable results or prospects. It doesn't shock me at all that people are having a lot of trouble getting work in the field. Even 8 years ago, going to college job fairs, I almost had to herd away Masters students in ML and Data Science who were already having a hard time getting a job and were resorting to applying for QA positions at web dev companies.

[ransom1538]

Eh. The real elephant in the room with "ML and Data Science" is checks and balances. They take over as the eyes of the company. Guess what department is seen as kicking ass? Guess which department is getting most money? Most hires? Data. Data.

It's like having the local police department check on the local police department.