I want to ask whether AI labs measure their model capabilities by Lines of Code, PRs per day, or whatever like whether the model escaped the “sandbox”? You guys are naive. No arguments. I’ll see this post one year later. (A calendar event scheduled)
@bradwestfall Everything is fine as long as you don’t read code and this shit is happening every day. It cannot be resolved with some Skills or prompts. As a young man who loves trying new things and chose CS as majority in 2024, I’m becoming increasingly disappointed at LLMs for coding.
I love Fish-shell so much so I wrote a language server for it as well as docs about Fish scripting languages as if reader never heard of Bash. The general idea is simple, just explain the builtins and syntax of pipe, redirection, expansion, etc. But writing itself is hard.
@didymus77@ChShersh That’s the point. I never seen they spent time talking about maintaining but how *fast* it was built. I don’t care how fast the code was generated, and I do care if the software carefully maintained in next 10 years or more.
@didymus77@ChShersh Yeah. I assumed that a codebase created from scratch by AI, and I didn’t have legacy codebase considered, my bad. But currently it seems that lots of new codebase are not that case and we can make it properly maybe or not at an early phase.
@didymus77@ChShersh Yeah we cannot avoid a bug in a software, but the bug occurred within “proper software” is under your control because you managed the complexity, otherwise you have to fix it like implementing a new feature or other something new to us, which costs our more time and effort.
@vettive_@fjzeit For example, the idea that WASM as first class in web platform hasn’t been completed and we still need Javascript clue, but browser vendors don’t want to move forward it.
https://t.co/HEn3WyxdOp
I’m sure it will have huge impact on web, and AI coding once it has been achieved.
@vettive_@fjzeit Help our users make money.
A product made just its author thought “wow it solved my problems so it has market and i’ll make money” maybe didn’t solve the problems that matter in the field. Those problems are “invisible” bc they’re normalized.
@ChShersh No one will always be young, but there always be someone young. Just give everything to the present.
When the day comes and I’m about to die, I’d rather brag about what I’ve accomplished and laugh at what I’ve fucked up than regret what I never did.
@Dearme2_ Time.
I do whatever I like, e.g. biking, running, fitness, football, watching movies etc; try something new, interesting or important, like cooking, writing novels, systems programming. Anyway, I have time to do silly things even just for fun. And I appreciate that.