/cs/ - Computer Science

>>182
pretty much

>>182
Chat-gpt can't do shit in codebases larger than 500 lines of codes. It is only helpful when writing trivial stuff that you can probably already find on stackoverflow.

Yeah. Alot of the web 3.0 guys are as braindead as anyone else. It'll continue to be like this as long as tech exists and is profitable.

Self-driving cars are still retarded - this has nothing to do with generative models.
The sheer doomer autism with LLMs obfuscates the true AGI paradigm of reinforcement learning.

I've decided I like the "AI will replace programmers" thing because it lets me know instantly that the person saying it either isn't a programmer or has such little experience that ChatGPT can ACTUALLY replace him.

I use ChatGPT (or similar LLMs) regularly when I want to do some boilerplate, but most of the time it's useless if the thing I want to do isn't a completely self-contained simple function. LLMs are actually way more useful for learning new programming languages because you can convert code from a language you can understand to the new one or tell you the idiomatic way of doing what you want to do.

Any programmer who has any confidence in his own skill will see these technologies are helpful tools rather than competition.

no

>>194

>>197
>Mentions reinforcement learning
>Pretends its more than just cope
omegakek