AI code is only impressive to begginers.AI code is bad, because is trained on public internet code made by amateurs and pajeets.
>>110079728>trained on public internet code made by amateurs and pajeetsbut enough about the Linux kernel
Now let's see your code now.
AI apparently can't refactor and simplify, only make more and more bloated.
>>110079728Yeah while Claude is amazing for a lot of things there are many aspects of my work it simply fails at. That said, if trained on our proprietary code base for long enough then it would be amazing again.
>>110079728using ai gen code "as-is".
>>110079728>AI code is only impressive to begginers.>AI code is bad, because is trained on public internet code made by amateurs and pajeets.At this point only actual noncoders and self-taught intermediates say this.
>>110079764>>110079733foss code is just amateur shit, 99.9% of foss shit is just amateur projects.
using your phone's sms autocomplete to write code is like eating curry and rice off the bare ground with filthy handsso I guess that's why it's so popular in India
>>110079728I pooed
You can see how retarded the internet has become now that everyone can post their 2 hours slop on any social media, for them its like they have landed on the moon or they feeling like they work at nasa but in their basement now.
>>11007972866% of devs use coding agents and this increases to 70+% with more years of experience.https://survey.stackoverflow.co/2026/ai/dataSince the source is a stack overflow survey, it's probably an underestimate (heavy users of AI aren't visiting stack overflow at all anymore).I agree that AI-generated code tends to be of noticeably lower quality than handwritten code, but there are measures that can be taken to avoid the worst of the slop, and they're getting better with each model release.If the only argument for handwriting vs generating is that human code is somewhat clearer or more elegant, that's bad news for anyone who wants to keep getting paid to write code by hand.If you're a professional SWE who doesn't use AI, you're living on borrowed time. If you lose your current role you risk being unemployable.
>>110079946>they're getting better with each model releaseThis isn't true. GPT-6 Astra's code quality is noticeably shittier than earlier GPT models and was memed as being unreadable.
>>110079946No skill is required to prompt an LLM to make code. You are deluding yourself that you are valuable in an AI world.
>>110079977OK fair, there've been regressions. But if you're unhappy with the quality of the current best available model, just wait 3 months and try the best model available at that time. I'm pretty happy with opus 5.5 right now.>>110080011Right now that's kind of like saying "no skill is required to use an autopilot". It's basically true that it doesn't require any real skill compared to hand flying, but you still wouldn't hire a pilot without making sure that they're experienced with using it, that they know all the different modes, that they can use judgement to decide when to use it and when not to, etc.
>>110080244My point is that model makers don't give a shit about the code quality, only about whether the models are capable of a coding task or not. Opus is no different than the rest of Anthropic's models in that department but it's definitely better at successfully carrying out something and is therefore higher on the benchmarks.Either it's something that they consider unimportant, or they keep it to foster dependency.
>>110079728Luddite cope thread
>>110079728bumping an epic cris thread!
>>110080244>but you still wouldn't hire a pilot without making sure that they're experienced with using it, that they know all the different modes, that they can use judgement to decide when to use it and when not to, etc.and yet tech companies have been hiring cheap foreigners with phony credentials for decades now and still haven't learned
>luddite copedo a better argument instead, show the dataa.
>>110079728You are unemployed
>>110085575>libtard arrestshow your shower soap, otherwise filtered.
