Raw LLM Responses

It's really quite troubling how many people use it to literally just think for them. Without doing any checks on what it outputs because yes - LLMs do hallucinate, they do get shit wrong. Like, I use it for things like this: \-I ran a benchmarking suite I wrote to test 6 different models I had on my hardware \-Each model required 2 tests each - one quantized, one unquantized \-At the end it gives me a list with the results of each test - and it was an assload of data to compare. So I formatted the entire list with ChatGPT, and then I went over the results with a separate script to make sure all the values were as they should be, and then I took that list and fed it into ChatGPT again and had it do essentially what I would do myself - compare all the results, and then output the best performing model compared to the others. It did it(and again verified that it was correct) and it saved me about an hour and a half of monotony. ~~THAT'S what the technology is for - shit like that; augmenting your workflow and making you more efficient and faster. Not for literally fucking thinking for you. Or hell creative things too.~~ I don't know, I am not gonna tell you how to use it - but like, please for the love of god, don't use it to think for you please.