Egregoros

@feld For sure. This is where the bean counters are being stupid thinking they can replace people with this stuff.

You need CHOPS to understand what "good" and "structurally sound" look like.

Because LLMs are AWFULLY damn good at producing very convincingly shiny garbage that emits very nearly all the right beeps and pings.

Except for that totally critical beep 5 of 7. If you're missing that beep the machine EXPLODES in a shower of sparks, leaving the clueless bean counters wandering confused amidst the rubble.

I'd say "This will be fun to watch" except it won't be given all the harm that'll happen in order for us to GET there.

@drwho This is another for realz problem with LLM use in code generation:

They will ALWAYS, even if you tell them otherwise, fall back on 'native' abstractions and choose brute force code gen to solve a problem that may well be better or more readably solved by using a library instead.

This is the kind of thing that makes me queasy about the current LLM wave, even though I wind up talking a lot about how I use them for work and otherwise.

We really ARE cheapening the overall quality of our output when we use these things. We're often making a devil's bargain exchanging velocity for quality, because that's what the majority of what the industry wants.

There are bright spots though, I've been reading about a few people who reject the narrative of velocity and are using these tools to generate better quality code.