I never in my wildest dreams thought code would be the first thing to get automated. I thought copilot was a glorified autocomplete for years. I thought vibe coding last year was a more general but still a yassified autocomplete that required users to do a lot of manual work, doomed to build upon broken incorrect code.But, here we are. I guess the reason why none of us recognize this as cyberpunk or futuristic is that it's just not something most thought possible, until last year, really. And because the densest introspective fog isn't over the past, or the future, but the present.We'll likely end up with a superhuman coder by the end of this year, if it isn't already here, and I think it is, just lacking a harness. At that point, writing code manually will be harmful to a project. Human feedback and taste would stop helping if you can just burn more tokens and let the AI arbitrarily rewrite code if it gets stuck in a local minimum. At that point, maintainers will likely be more of a harm, and reviewing will become impossible.Sure, we can write code for fun. But any public code ends up being more training data. And agents ran by those with nothing better to do will find a way to contact you to nag you about your low-quality code.Human training data will become irrelevant in no more than a year, when synthetic data from majority AI-led projects becomes used in production, at which point anything proven to last a month or two with no bugs is good enough material to use in training.
>>108875150The first thing to be automated is """AI""" marketing spam.
Unfortunately, LLM tech is fundamentally nondeterministic, which makes it unsuitable for applications that require accuracy and mathematical truth.
>>108875150if you think i'm reading all of that you're retarded.that's a man btw.
>>108875272>that's a man btw.Source or shut up you fucking asshole.
>>108875150OK but what did you make with AI coding?
>>108875150>But any public code ends up being more training data.There won't be public code anymore.Only open source retards who eagerly code themselves into unemployment by training bots
>>108875150if AI were to replace programmers it would certainly replace retards with a cuck mindset like you. AI will never be better at programming than a determined and persistent human being. did you forget we built all this shit? get a grip you fucking faggot.
>>108875150
>>108875260Humans are also non deterministic. Programmers make mistakes all the time. Judges have a much higher conviction rate right before lunch.
>>108876886>Humans are also non deterministic.True, which is why code generation has been delegated as much as possible to deterministic tools, such as compilers, assemblers, macro processors, etc.LLMs as code generators are a step back, unironically. The only people who believe otherwise are ignorant about computing history as a whole.
>>108876925How would making the whole stack deterministic work? Even if the person writes the spec in a formal language, the person or thing that generates the formal spec has to do so from a fuzzy set of goals.I think LLMs could allow you to make more of the code generation deterministic, because you can give an LLM formal requirements and ask it to iterate until ti produces code plus a proof that it meets the spec, in a more consistent way than if you had a human there writing the spec for you. Technically speaking you should be able to give the same prompt and same seed and have the LLM generate the same output every time. I don't know if you would consider a PRNG "deterministic".
>>108876886>but le humans alsoSpambot rhetorical pattern.
>>108876886>Humans are also non deterministicHumans are capable of reason. Reason is deterministic.
Even working with claude is still not enough. The code it spits out is sometimes very good but quite often it is utter garbage. I like to use it as a guideline for my own development rather than a replacement for me entirely. Vibe coders still are shit.
>>108877034it really is like a slot machine
>>108876999>I think LLMs could allow you to make more of the code generation deterministic, because you can give an LLM formal requirements and ask it to iterate until ti produces code plus a proof that it meets the spec, in a more consistent way than if you had a human there writing the spec for you. Technically speaking you should be able to give the same prompt and same seed and have the LLM generate the same output every time. I don't know if you would consider a PRNG "deterministic".Sounds very much like a thing called "compilers".
>>108875150>I never in my wildest dreams thought code would be the first thing to get automatedWhy? Until hardware and comms got good enough to make mass manipulation posible and the software industry got pumped with money, it was seen as a meme ocupation. It's not like you're making rockets here.I think codefags just overestimate the complexity of coding.
>>108879094>It's not like you're making rockets here.Not all programming is webdev.
