Clojurians Log v2

Firing all the python devs also sounds suspect. Tech cos fire 5 to 10 % of all hires, is this just one way of making it seem smarter than it is?

https://news.ycombinator.com/item?id=41991291 > I work for Google, and I just got done with my work day. I was just writing I guess what you'd call "AI generated code." > But the code completion engine is basically just good at finishing the lines I'm writing. If I'm writing "function getAc..." it's smart enough to complete to "function getActionHandler()", and maybe suggest the correct arguments and a decent jsdoc comment. > So basically, it's a helpful productivity tool but it's not doing any engineering at all. It's probably about as good, maybe slightly worse, than Copilot. (I haven't used it recently though.) It's marketing imo. Investors want to hear 'forefront of ai'.

I'd like to talk to an insider about how much this is spin vs. reality, and which use cases really benefit (e.g., mostly the boiler plate types of dev?)

They have a lot of boilerplate that AI can generate so I’m not surprised they’re mentioning it. But Google tends to exaggerate, especially with AI, so I’d assume they’re gaming the number.

Sundar recently qualified that statement in an interview at the NY Times conference where he said, "To be clear, I was talking about code completion for small snippets of code, not large scale code design," or something to that effect

And that allegedly adds up to 25% in some way. I'd imagine it's more like 10% to 20%. I have no doubt it'll get better quick though

Yeah, sounds like that aligns with what the engineer is quoted saying above

If we're talking 25% of all characters produced across changes I could see it. I just thought it was interesting to see google show this off when obviously producing code != producing value. It's like someone saying "half our code is produced by auto complete", so what?