I find it hard to understand how many people use chatgpt all the time I tried it a few time, and it gives answers in a very annoying way, often completely wrong, yet very sure of itself "akchually" correcting you with incorrect information
>>108764089Add something like "search the internet and double check all your claims" at the end, I always get very awful answers whenever it replies without performing a search
>>108764092so far I havent found much use for it, I can search the internet myself
>>108764089I don't use LLMs often but when I do I avoid prompting in ways that invite ass kissing responses. It's sometimes wrong and sometimes right, at the very least it can point me in the right direction. I only trust what I can verify. Treat it like an opinion, not the voice of God.
>>108764089skill issue
>>108764136fucking bot
>another AI cope thread just make a general at this point
>>108764092This is how I trained mine. Last night it started pulling answers out of its ass when I was trying to figure out regular expressions, which I have never worked with before, in KRename. Had to remind it of that one time it spent a full 2 minutes carefully searching the internet for some other thing last week and returned flawless results. We started getting good results after that.
>using the web chat in 2026 instead of running it in a harness like codexShiggydiggy, bro. You don't know what you're missing.
>>108764089Ai is over, it can only be trusted with casual answers that arent too important it gets wrong. The drunk at the bar is a more reliable source
>>108764224do you also have the impression it used to be better 2 years ago or so?
>>108764089i recommend opus. chatgpt has been this way (more than usual) for the last 2-3 months
You are thinking form a useful/skilled person perspective.AI is useless for you.But, for the majority of useless/lazy devs/workers, it is a revelation. Luckily the higher ups will realize such workers aren't needed with AI and fire their asses in the coming years.
>>108764463makes sense to me
>>108764463probably the management is keeping metrics on LLM usage by employee, and correlating it directly with right sizing potential
>>10876408999% of the world does not know the difference between the word "Browser" and "Internet Explorer" . Most people do not have the capability to think beyond 1 sentence ahead into the future.
it's basically a dumbed down google search (not talking about it's potential use cases for programming, I don't use that)
>>108765073its*
>>108764136NPC reply
