Strugling using ai to much but also strugling using ai less

Since 2022, ChatGPT has been an integral part of my life in doing anything, not only gpt other llms like claude, deepseek, glm and others has been a really big deal in my life in literally doing anything.

When Github Copilot released the first time, i was still a student at school, i got access using copilot from the github student pack, now from the first time using it i’ve always been sceptical about using it, not only it generates to much information in code based of assumtion, even when i gave it instruction… it still returns nothing of use, people i know who at the time didnt have much time loved it so much, because it generates code, it does what they need and so its good… but then they hand it of to me to maintain which is not good.

Now after a while LLM has gotten so good at generating code, it really does, and its amazing… but not at everything, i’ve been working on an app for 3 months now, i vibed it at first which is nice, it does what i need, but i started to find flaws in the app that i strunggle to fix, even the LLM that generates it struggles at fixing it, the simpless way i can explain it, its like handing it over to a different person everytime because LLM are just stateless machines with its current limited context as its memories.

well i know that im suppose to manage this context, and agentic tools like claude code has features where it manages memories, but even with that it keeps missing the point, instead of fixing it like a machine with no feelings does, it keeps giving excues like humans who dont care about the product

Have you ever gotten that? or am i just crazy and making this up? now when humans does that, we can replace them with another human, but LLMS? we cant, its litterlaly the same stateless machine everytime that has no memory or the feelings towards something…

LLMS are so good and so usefull, but after a while its just anoying having to manage it, when people say they havent touched an ide since llms are around, i questioned their sanity, like how do you sleep at night without knowing that a product that has you as its sole creator and you being the only human responsible with it, being used by other people that keeps “maybe” their personal information, stores their critical data, is solely being managed by an LLM?

that really does keep me up at night, it literally holds my reputation at the edge of a cliff…

Im not hating on it, i love LLMs, i can make time to study rather then struggle for hours on a problem, but when the LLM also struggles at a problem they create that is a whole other thing.