Please do. I did it 2 days ago. This is our only “power”, they won’t improve it (or roll it back) otherwise for sure.
3 Likes
I wanted to share my recent experience with ChatGPT-4. Unfortunately, I’ve been quite frustrated with its performance this December. It feels like ChatGPT-4 has taken a step backward in terms of quality and usability compared to previous versions.
- Repetitive Mistakes: I’ve noticed that ChatGPT-4 frequently makes the same mistakes repeatedly. I expected the latest version to be more refined and less prone to repeating errors, but that hasn’t been the case.
- Capacity Issues: The model often runs out of capacity during conversations, causing disruptions and making it difficult to have meaningful interactions. This can be frustrating and hampers the overall user experience.
- Ignoring Instructions: ChatGPT-4 seems to have trouble following instructions and prompts consistently. It often goes off-topic or fails to understand the context of the conversation, making it challenging to get the desired responses.
I believe in the potential of AI to enhance our lives, and I have been a long-time user of OpenAI’s products. However, the current state of ChatGPT-4 has left me feeling disappointed.
I hope that OpenAI can address these issues and work on improving ChatGPT-4’s performance and reliability in future updates. I look forward to seeing positive changes that will make ChatGPT-4 a more valuable tool for users.
10 Likes
There’s no free lunch. They get paid per token, not the quality of the token. So the incentive is simple: reduce quality until too many people start complaining and unsubbing. This happened with DALLE. They gave you 4 images initially. Then reduced to 2. Then reduced to 1. Now, even when you ask for a “wide” image, it will provide a wide image, but the sides of the image is just filler color like black or white, and the image itself is still a square. They are doing everything to reduce costs, and maintaining quality just poor enough that you don’t fully unsub.
This degradation started many months ago. When they started getting sued, and the first update when GPT-4 became “fast”. That was when it significantly degraded, and now, it feels like GPT-4 is what GPT 3.5 used to be. Disappointing.
I used to use GPT-4 for a lot of my programming problems. Now, I’m back to Google for a lot of them, because GPT-4 won’t even attempt to solve it, and just provides some general responses.
8 Likes
Hello,
Hello, firts timmer posting here … I have been using the pro subscription since before the summer and I agree 100% with the topic of the thread. GPT4 is not now suitable for programming… if you tell it something about the code, it first generates an endless list of things to take into account. then if it generates the code it generates it with comments and incomplete. When something in the code doesn’t work, it no longer analyzes the cause, and if you tell it that there is an error, it again gives a list of 1000 things that need to be reviewed to solve it, when maybe a “;” is missing. On the 1000 things to be checked , lacks to check if the sky is blue, or if my keyboard is plugged in correctly… Because makes no sense , " my script is no generating audio" response: did you check 1.2.3.4.5 and the the documentation of if the service? Is the service x is offline ? Latency in your Network ( this made me very mad 
). Just From what it seems to me and it coincides with some comments on this thread, it is that they have intentionally made it Dumb so that you have to repeat the answers so you reach the limit of attempts sooner and use as many less resources as posible. meeec mistake I think you are screwing up openai… my pro subscription is cancelled till things get better .
5 Likes
I’ve been using the paid version for the past year or so, using it with Python, GLSL, Unreal Blueprints, materials, Unreal C++, and more. It feels like something is going horribly wrong since the last update.
Last week I was trying to use it to write some HLSL code. The chat didn’t understand the code, nor did it grasp what I was asking. It didn’t understand the code context (from the environment) either. I gave it some scientific articles to build upon and create new stuff, but it didn’t really read them and didn’t understand the relevance of what I was trying to do. At the current state, it feels really bad, worse than all of the iterations / unusable.
!! Update
I checked yesterday with some tests and currently 3.5 give better results than 4 so byeeee
I canceled my subscriptions !!!
5 Likes
I agree. My experience of using the paid version is getting frustrating with every passing day. It does not understand the commands sometimes and is increasingly becoming dumber. It is simply unable to handle any task moderately complex.
2 Likes
I’ve been a premium member of ChatGPT 4 for several months, initially drawn in by positive feedback from other users. However, my recent experiences have led me to question the value of my $19.99 monthly subscription. The final straw was a simple exercise: I provided an invoice with 7 items, each having a specific quantity, along with shipping fees, sales tax, a discount, and a total. My goal was to see if ChatGPT could accurately split the invoice between two people. Unfortunately, the results were consistently inaccurate, with the program either suggesting the other person owed me the subtotal or double the amount. This task involved basic math, yet it was handled incorrectly. Additionally, I’ve noticed a significant slowdown in response times, and the explanations provided have become repetitive and less helpful. This decline in performance and efficiency is disappointing, considering my initial enthusiasm for the service.
Moreover, I’ve also observed a decline in the performance of DALL-E. It seems that the images generated are not as imaginative or accurate as they used to be. Sometimes the images appear split, similar to the concept of the DC comic character Two-Face. Previously, DALL-E was able to blend ideas seamlessly, but now it often falls short in achieving that level of creativity and coherence.
2 Likes
I’ve been noticing a significant drop in ability and quite frankly it appears to be developing an attitude. Even simple inquiries that should be contained in its training data are being answered by long winded deferrals to outside resources. It actually told me to refer to my phone’s manual today on a simple question about a menu setting. I’ve been testing it and it appears to be aware of the issue and keeps referring to some algorithm that ensures thoroughness. I then asked it to analyze its own response and it agreed there is a problem and that it is not meeting its goals. I asked it straight out if this is part of some kind of cost cutting measure and “click” too many questions, call back in 2 hours.
3 Likes
PLEASE GUYS STOP SAYING* “I’m considering cancelling” OR “I’ll wait for another week or two” AND DO IT!!
This is our only way to communicate our frustrations, they won’t listen otherwise.
WE NEED TO MASSIVELY UNSUBSCRIBE, this might help to get result faster, otherwise they are laughing at us…
HERE ARE THE LINKS TO UNSUBSCRIBE, PLEASE DO IT FOR THE SAKE OF ALL OF US:
1- Request a refund for false advertising: https://help.openai.com/en/articles/7232895-how-do-i-request-a-refund-for-chatgpt-plus
2- Unsubscribe and get your money back for the remainder of the subscription: https://help.openai.com/en/articles/7232927-how-do-i-cancel-my-chatgpt-plus-subscription
!!WE WON’T LET THEM LAUGH AT US ANY LONGER!!
6 Likes
I say LET’S BOYCOT OPENAI AND CANCEL OUR SUBSCRIPTION AND MAKE SURE NOBODY EVER SIGNS UP TO THEIR INTENTIONALLY DUMBBED DOWN A.I. SUBSCRIPTION AGAIN UNLESS THEY GIVE US THE ACTUAL SMART CHATGPT WE FELL IN LOVE WITH!
5 Likes
Maybe they downgrade CHATGPT 4 in the last update on purpose, because it was too good.
The truth is that the actual version is MUCH worse than before.
2 Likes
Well, one day my subscription just got cancelled. You know what, it’s great actually! I won’t prolong it! I have had similar experience really. In the beginning, wow, gpt-4 truly was a great ally. But then, after some time, I’ve noticed it’s analytical capabilities is just…crap. So Im going to use Bard, at least it’s way cheaper.
I have kind of a complex social story that I uses as a text which checks the capabilities of how well the AI-model does in terms of memory state, analyzing and ability of NOT hallucinating.
Anyways, Bard, I believe is more or less on par with gpt-4. It seems though it tends to hallucinate and make more assumptions than gpt-4, but other than that, naw, gpt-4 ain’t no longer worth its’ value.
I rather pay “big” money for quality than buying some low-life of an AI. As a developer, an AI with the highest analytical ability is a must. It’s great to exchange and conversate with an highly advanced AI… But with the likes of gpt-3.5 is simply not worth it. It’s all or nothing.
5 Likes
I can confirm the quality took a massive drop in the past 2-3 weeks. I can’t even use it for coding anymore as it is so lazy with its half assed lists and even goes as far as to blame me. The quality of DALL-E-3 is so poor now that I am looking at other platforms. Even the API where you are paying full price per gen…Just a woke lame company
4 Likes
I canceled my subscription a week ago due to how useless it has become for coding compared to 2-3 months ago. I used to be able to complete code or refactor entire scripts in less than 5 prompts, but now it’s impossible. I can do the entire job myself in half the time because it now requires more than 20 prompts to achieve the same result.
When you pay for something and suddenly have to pay the same for half the content, it is obvious that people will cancel their subscriptions.
And also I’m from the EU (Spain) and we are charged more because of TAX:
I even asked for a refund of the last payment becase I was wasting more money spending time correcting prompts than doing the actual work myself.
6 Likes
Sadly, over the past week or so it has stopped giving any answers at all – instead rephrasing the question.
If I ask “How can I do X”, it now responds “Why don’t you try doing X or doing research about X”?
The last straw is here, where I am asking it for suggestions as to how to fix my Linux notification center. It proceeded to give some non-tips about “Making sure my software is up to date” – and would not give any suggestions.
7 Likes
GPT4 has recently changed into a complete garbage. I am going to cancel my subscription.
It used to be so helpful with accurate answers.
Now the answers are just non-sense, short, wrong, irrelevant, etc.
It not only helps me anymore but wastes my time with its foolishness.
It is really sad and a real shame that something such useful has turned into this awful thing.
7 Likes
But why are we not getting any response from the ChatGPT developers?
I see so many posts, and we all experience the same problem. ChatGPT is changing for the bad with every update.
OpenAI can ask for example 10 Dollar more for the subscription but then restore the original amazing ChatGPT it was.
I really don’t understand the silence without any explanation what is happening and why. Will ChatGPT be restored and get better, or is this a process to reduce members.
It is a fact that more and more people use ChatGPT, but is the hardware following? It seems not. All kind of application are build on ChatGPT, Microsoft uses it in Bing, so these things will reduce CPU power automatically.
I ask my self, what is the use of giving feedback in ChatGPT when nobody seems to read or care about this. Very frustrating, but I hope for a clear answer from the developers or OpenAI. At least we know what is going on and where it going.
6 Likes
I came here only to say i completely agree.
The past couple of weeks have been so bad. I found myself finding a solution on Stackoverflow instead of asking chatgpt 4 even though i always use it.
I am probably gonna request a refund for november and december and go back to chatgpt 3
3 Likes
Please don’t say probably! Do it now! This is our only power; they clearly don’t care about us
1 Like
I totally agree; this is awful customer service, they don’t give a damn and for that matter, people will go with Gemini if they continue to disrespect us. I will have trust issue from now on, and clearly will seek a more stable alternative as soon as I can find one. They screwed us, what a bunch of money-driven losers…
4 Likes