Word on the street is that SRE/DevOps jobs are starting to ramp up, very slowly. Any truth to that?
Haven't seen anything significant yet in Canada, although most reliability teams were on fairly high staffing and high alert after the G7 that happened recently.
They don't think that far ahead. Only 2-3 quarters ahead
The assumption is that if AI has gone from being non-existent to junior level engineer in three years, then it'll be intermediate AI in another three years, whereas a junior employee might take four or five. That intermediate AI will scale to senior capability in another three years, whereas that intermediate employee might take five to ten years more.
Effectively, they're projecting its current capability growth as being at minimal, linear going forward, while hoping its in fact exponential. The fatal flaw here is the assumption that AI is actually at even Junior levels without an actually competent developer handholding it the entire way, which its not. But an mid level suits job is quite literally to analyze data and project trends and provide that information to senior leadership, so its kind of hard to blame them, they are doing their jobs with the information available to them. The whole space is just flooded with retarded misinformation meaning its impossible for them to make sensible decisions. Look at how many AI firms are turning out to be absolute trash, scams, or Actually Indians. For companies looking to use AI to provide a viable cost reduction to producing their goods, they're lining up to have a bunch of critical disasters, but I suspect for some of them, the alternative is guaranteed bankruptcy anyway, just slower. A lot of firms are basically zombie firms after the economic fuckery of the last ten years, and AI is a tiny, slim chance at resurrection for many of them.
But for the infrastructure giants looking to provide the AI technology, Microsoft and them, the long term is pretty easy to invest in even with full knowledge of the spaces problems right now. Even if AI only has a 1% chance of achieving what its promising in the next ten years, and needs billions of dollars a year to do it, that's a CEO being asked if he'd like to invest 1 dollar to make ten thousand dollars. Even if its got slim as fuck chances of success, getting just a fraction of what's considered possible will generate more money for them than anything they can currently invest in. And for the core infrastructure players, if AI turns out to well and truly be a failure, then all they've done is invest heavily in computing infrastructure, as companies that make their money renting access to computing infrastructure. It'd suck for them in the short term as they'd have to reduce pricing to encourage use of the capacity, but long term, if there's compute available someone will want to rent access to it. They're still left with a viable asset, they've just overpaid for it.
