HN Explorer

Ask HN: ChatGPT Is Now PagliacciGPT?

2 points by Racing0461 ↗ HN
Lately, it seems ChatGPT 4 has been very lazy in it's responses. Telling me to change the code to my needs even tho i asked it to, doesn't implement functions and just adds a comment to add logic here, tells me to see a specialist or even go read the documentation.

Has anyone else noticed this? Is OAI just trying to save resources by nerfing chatgpt even more? If so why can't they do this bs on chat gpt 3.5 and leave 4 alone?

I can't help by think of Pagliacci when chatgpt tells me to go see a specialist for my problem. You are the specialist.

6 comments

[ 2.7 ms ] story [ 20.1 ms ] thread
smoldesu
> I can't help by think of Pagliacci when chatgpt tells me to go see a specialist for my problem. You are the specialist.

This has been a thing since the beginning, no? Arguably it's the right response too, an LLM is not getting any more credible as time goes on. Making an LLM smarter might just entail fewer uncertain answers.

Racing0461
At the beginning it was a disclaimer, now it doesnt even try.
smoldesu
Caveat emptor.
Racing0461
> Caveat emptor

Doesn't apply here. I would have loved to freeze the chatgpt4 model the day i bought it.

It's more like i bought a car, then 4 months after Sam Altman broke into my garage and smashed the car windows and broke the engine then left.

smoldesu
Maybe that would be an apt analogy if you didn't agree to an EULA limiting OpenAI's liability as terms of using their service. Your only path of recourse as their customer is leaving the platform.
catchnear4321
sama owns the cars. not in a financial sense, of course.

you pay rent for the opportunity to use the new best version of a car, probably.

it has been decided / determined that the windows increased drag, supposedly for the sake of safety and comfort, and that isn’t viable in the current economic climate.

(the engine was quantified as redundant.)

that will be another $20.