ChatGPT 4 is worse than 3.5

Add me to the list of severely disappointed customers. This tool simply does not work for me. Subscription canceled.

1 Like

Yeah, I cancelled my subscription too, with the way it is behaving that amount is entirely unjustifiable.

1 Like

I also faced the same problem. ChatGPT 4 keeps answering very generally and uselessly. It consumed a lot of my time to convince it to give me the answers as good as it did on the first day, but I felt helpless.

I also canceled my subscription. I felt like I paid for a service much worse than the public version.

1 Like

I’m going to have to agree with this. It’s getting blatantly bad. From constant errors, to ignoring its own capabilities. For example, I uploaded a file for it to analyze and it told me it was an AI language model and couldn’t access uploaded files. I had to convince it that it could. Same with browsing the web, either by requesting that it search online, or giving it a specific URL to visit. Just kept saying it wasn’t able to do these things. I had to argue with it before it was like “Apologies, let me do the things you asked me to do.” Ridiculous. I was left shaking my head at the absolute garbage this has become.

1 Like

I also confirm that chat gpt 4 is currently broken. 3.5 is working better. But this happened in the last two months. IT was working fine before.

2 Likes

I’ve had the same issues. Nonsensical responses.

I asked gpt4 of historical female figures that embodies strength. It responded with maximilian, no, not the emperor. It invented some guy, and said he was a close friend of alexander the great.

What am i paying for? Gpt4 is fine by itself, its this absolute lobotomization you guys have done on the chatgpt version that’s ruining everything. You want to save costs? Okay? Good for you openai, i’m paying the premium cost for BETTER service, and youve actively made it worse durinh my patronage.

1 Like

I’m cancelling my subscription too, I was facing this issue since December but I thought they’ll fix it sooner, but all they did is making it worse.
I think I’ll just use 3.5 untill they fix it.

Chatgpt was awesome last year, but now it’s absolutely dumb, it forgets your conversation after three messages, looks like it’s learning on user’s prompts and prompts are dumb, so chatgpt becomes dumb too😅 Hope they’ll fix it or anybody else will develop something better

1 Like

All the same problems here, too. I cancelled a few hours after buying. I guess the free version gets a ton more training data? I don’t know. But anyone looking for development advice will be far better off with GPT3.5.

Agreed with all of the above, this is such a disappointment. I’ve been a pro user for a few months and now can’t even get basic tasks done that were easy a few weeks ago. If my prompt has more than one element (like read this message and recreate the tone) it can’t handle it, just gives me a summary of what the message was about. It’s a major bummer, but I’m canceling my subscription too. Somebody email me once they’ve gotten their shit together :frowning:

ChatGPT: Pretty dang bad. Turns working code into non-working code. Report the problem with the changes the AI made and the need for a different solution, gives me plausible things I can fix in its code. Says it would need more information, the very schema and example I placed three turns before. Complain enough to then activate memory of the specification, and get a code snippet 1/3 the size that does nothing.

Why can ChatGPT not use retrieval properly, which requires multi-tool use only trained into the latest gpt-4-turbo models, which also have knowledge through 2022?

I think this is your answer.

image

1 Like

Bruh it’s 30 message cap not 5!

I too am very disappointed. Both models just got progressively worse. Subscription canceled.

1 Like

Hi! I’m here to share the same problems I have with 4.0 recently. The issue I have with it is that even if I ask a VERY specific question, or ask to do something VERY specific (like to create a table with specific formatting and content) - it starts to generate kilometer-long commentaries, add related information and so on. Basically wasting my time, because its response time is much much slower than was for 3.5. Oh, and after all these long texts it fails to complete my request or answer my question, so all I get is a wall of useless text.

Also it tends to plain ignore the request details and makes something unrelated from time to time. And requires 10-15 cycles of arguing and detalization to finally achieve what I need, while 3.5 works flawlessly in the same conditions and does what is requested fast and precise.

Oh and it switches to English and back randomly.

I would not care much about the slow response speed, but given that I have to literally argue with it and tell several times in a row something like “You ignored my initial request, please read it once more” - it becomes a real burden.

So true, 1 year of plus sub, and décided to unsubscribe. Gpt 3,5 is more efficient in lot of prompt lol WTF?