Post by @lidsel@mastodon.gamedev.place

David Gerard
@davidgerard @circumstances.run
Public en edited

but of course AI works great for coding! if you apply this list of proper engineering practices your manager won't let you apply because the point is to go faster, not correctly

no you can't see my allegedly wonderfully engineered AI code

Lidsel
@lidsel @mastodon.gamedev.place

@davidgerard I was trying to humor a friend so I just prompted the thing to write me a Roslyn analyzer (a simple no context single file problem with 4 short bullet points of acceptance criteria) and even after re-prompting it to split it into sub-methods I can't read this shit.

David Gerard
@davidgerard @circumstances.run

@lidsel you must be prompting it wrong

Lidsel
@lidsel @mastodon.gamedev.place

@davidgerard One has to wonder what time I'm saving if it takes longer to figure out how to prompt it correctly than to just write the damned code yourself. 🤔

Nicolás Alvarez
@nicolas17 @treehouse.systems

@davidgerard @lidsel I avoided the "you must be prompting it wrong" and "yeah the free models suck but the expensive ones do work" objections, by asking someone to use his paid frontier-model subscription and his prompting skills, and make it write a Python program to do the simple task I had done the previous day.

First reaction "why is it using both A and B libraries for cryptography?", second reaction "wait lol it doesn't even run".

Lidsel
@lidsel @mastodon.gamedev.place

@davidgerard At my previous job they had a (human-made) 20 year old unreadable spaghetti codebase. It would regularly take me 3 hours of reading multiple files up a stack to figure out how to re-use a simple component. The revolutionary aspect of LLMs is being able to achieve this state in a month.