For years I've been hearing that "One day AI will be smarter than humans and we'll all be doomed."

"Nonsense," I said. "AI is very stupid, and not getting noticeably smarter." And I was right.

But I didn't think about the fact that there were
two ways that prophecy could be fulfilled.
For some reason, my year-old pinned post is doing the rounds, so I might as well add a shameless plug in case anyone's hiring an artisanal, LLM-free software developer.

RE:
bytes.programming.dev/notes/ah0j7hoiexdglzvo

@cholling

How has industry attitude to your, quite understandable, red lines changed in 12 months.

@MatthewNewell As far as I can tell, industry is even more pro-AI, because management has no idea where the actual bottlenecks of software development are or what needs to be optimized, and thinks the main cause of low velocity is "developers can't write code fast enough" rather than "developers trying desperately to get management to give them the resources they need and prioritize what's actually important instead of moving goalposts every sprint to the next new shiny thing". Then the whole "agentic" fad, where they think that the solution to a language model that doesn't actually understand anything is giving that same model unfettered access to your whole computer or even deploy pipeline, which I'm sure will end well.

Maybe I should just open a bakery.

@cholling

I was hoping, from my very outside perspective, that the industry ardour for AI might have cooled.

Baking is good. But to be excluded from an industry one is trained for due to management failure to recognize the emperor's new clothes must be galling.

I can offer nothing apart from boosts and wishes of good luck