"Can we run <LLM> on the entire codebase to just generate comments automatically?"
anon, how do you deal with this type of absolutely braindead AI managershit?
There is an inverse relationship between software development knowledge and trust of AI. So there is now the constant push-pull of "Hey, why did it take so long to do <task>? ChatGPT just gave me <some pile of flaming garbage>" versus you, the sniveling nerd, explaining to management why that's ackshually not gonna work.
I have been a software engineer for more than 10 years and have never seen the kind of blind faith silver bullet crap we are now seeing related to LLMs.
>"Can we run <LLM> on the entire codebase to just generate comments automatically?"
You can't, because normies think AI is literally magic. Jump ship if it gets too frustrating.
Can't, job is objectively good and the market sucks right now.
So what's the solution? Wait it out and let the results speak for themselves? I was actually tempted to suggest we do let an LLM auto comment the entire codebase, but in accelerationist type way.
> I was actually tempted to suggest we do let an LLM auto comment the entire codebase, but in accelerationist type way.
That would have been the correct answer. Always one up. Then do it, and watch it all burn.
>I have been a software engineer for more than 10 years and have never seen the kind of blind faith silver bullet crap we are now seeing related to LLMs.
Managoids are desperate to take away the power that software developers have and ai is the first thing that looks like it even slightly could
like its not even a bad idea anon ur just being close minded
Subhuman
ok boomers
Subhuman + moronic
see this is the problem. people like this who don't know enough to know why it's indeed a bad idea, and will go full-steam ahead.
Go read a book you Zoomer gay.
>Go read a book
ur right i should go read something with actual value instead of whatever u call posts, i should learn from failures like u and not waste my life on this website, thank u anon, ur so pathetic it inspired me to walk away
The craze will pass soon enough.
What makes you think so? I hope you're right but I don't see the timeline where that's true (at least "soon").
Trusting AI to write intelligent comments is stupid today.
But do you doubt AI will have it all figured out in 20 years?
Right now it looks like transformer based llms are slowing down so maybe. We really have no idea what the future will bring.
I think it'll largely work, but perhaps it is not mean to be in the main comments unless reviewed by a human but some supplementary docs database/file structure that can be regenerated as a whole if, for example, practical context size increases or the model becomes better...
>We can, but our engineering effort would be better spent delivering features.
>We can, but for the level of effort we spend integrating AI, we could just write the comments ourselves.
>We can, but it will distract from other priorities.
>We can, but...
Learn to office politick.
Yeah I'm just complaining. I know how to office politick. This stuff is just becoming a greater and greater annoyance.
I feel your pain anon, I have to "yes, but" my boss almost every day at this point.