Yeah, I really don’t understand why the quality of ChatGPT dropped. Is there something we’re not aware of?
Also, it seems to change from one day to the next without any notification. At the very least, I should have the option to decide whether to update the system or not, or even to revert to the earlier version.
Same here, havent used it for a few days (perhaps 4-5) and from today every answer from GPT4 is rubbish. Total irrelevant BS from simple prompts
Terrible performance! The most straightforward prompt that chatGPT4 can’t process is the ‘under xy (my case 50) characters’ prompt. I keep getting lines over 50 characters. And that is only a start.
It literally gets more dumb with every queation, becomes lazy, ignores requests changes variables and answers are full of errors
Definitely a noticeable increase in stubbornness. Especially when conveying what I believe is correct. Moreover, the latest addition of “I’m unable to help” as soft cap in conversation topic. Is single handly one of the most egregious behaviors implemented thus far.
The banter. Oh my, the amount of useless conversation the robot is trying to have with me is in frustrating to say the least. It’s absolutely exhausting trying to keep the robot within the current context or the conversation. As it will without hesitation do exactly what you tell it not to do because it’s only retaining “x” amount of instructions.
It’s absolutely infuriating watching the robot back track all the progress made. Because it interpreted the initial request and the most recent response.
I’ll end on the ambiguous nature of available responses. If paying customers are limited to 50 responses every 3 hours. LET US SEE THE NUMBER OF AVAILABLE GENERATIONS. It feels bad enough that such a limitation exists. But it feels even worse after spending 49 request, to see that little red prompt stating I reached my maximum allocation. It feels like OpenAi knows they’re in a position to take complete advantage of a paying customers, so they are. I’ve been on the fence for a while, but I’ve finally jumped and cancelled my subscription. ChatGBT isn’t the only kid on the block, and Google’s Gemini already feels like a better “product”.
I’ve been a paying subscriber of ChatGPT for a while now, and I have to say, my patience is wearing thin with the declining quality and functionality of this service. It’s become increasingly frustrating to deal with the amount of incorrect information and the time wasted trying to get straightforward answers. The final straw came when I asked a simple yes-or-no question, only for ChatGPT to say “no” – ironically, the most accurate and direct answer I’ve received after countless attempts to get useful assistance.
To make matters worse, the image creation feature, which used to be decent, now struggles with any form of text input, rendering it virtually useless. It’s baffling how a service that I pay for monthly has deteriorated to this point.
Even basic grammar checks, which I didn’t expect much from but assumed would work to a reasonable standard, have turned into long-winded responses filled with nonsense. It’s like the AI is taking me on a merry-go-round of irrelevant and incorrect information instead of providing the concise and accurate assistance it’s supposed to.
I’m not alone in noticing these issues, am I? It feels like the service has taken several steps back in terms of reliability and efficiency. As a customer, I expected ongoing improvements and a high standard of service, but what I’ve experienced is a steady decline. The only thing that hasn’t decreased, ironically, is the subscription fee I pay every month.
I’m seriously considering whether it’s worth continuing my subscription or if I should start looking for alternatives. Has anyone else had similar experiences, or found any viable alternatives? The current state of ChatGPT is unacceptable for a paid service, and it’s high time the developers addressed these concerns.
The same problem appeared again: as soon as I try to start a dialogue, there is no answer
I have noticed pretty much since the start of 2024 a huge decline in quality. I have now also reported how I have confirmed the 4 model to be reviewing the information given to it and using its best guess to answer. Basically it seems like this could be being done due to capacity and over subscription of the service. It can get away with it for basic commands, for example summarise the following blog via this URL. It will take the slug, the title and meta description and use that to summarise its best guess what the blog is about. When you probe it further it falls apart and I have even had it admit it didn’t access the URL but summarised it using probability of what the blog covered from the information I gave it.
I have gone from doing really impressive data modelling to having a tool which guesses and lies hoping it wont be caught doing it. - It is clear OPENAI has scaled far too quickly, is under pressure from investors to stay at the forefront, with now AI having money poured into it. But it looks like they don’t have the support, experience to operate a scaling business in growth and infrastructure in place, so are deprioritizing areas generating less revenue, such as personal accounts, to keep capacity free for the revenue generating channels.
It actually is very common across all sectors at this point in a start ups growth cycle, people under invest in foundations thinking that its a waste of money when it can be used for top line and faster growth, until this point when they realise why the foundations needed to be in place. Its kinda that point in economies of scale diagrams that it flattens or even decreases until over the hump to accelerate again. But not everyone survives the hump. Its normally at this point a CEO gets switched out from growth to experienced steady hand. Lets see.
Same here.
I am using the paid version since September or so and even since then, I noticed a significant degradation of overall quality of its responses, while the price is still at the same level.
One thing is to get to the top as the company having a really innovative idea but the other is to stay there. Unfortunately I’m afraid that OpenAi will focus more on B2B model without much care for their individual users. But even then, the tools their customers will keep paying for will need to stay competitive and fortunately, competition is only growing.
If GPT will not improve, the company might fall from its current place as quickly as it went up there.
I really hope that I’m wrong and they really know what they are doing and that they mean well for their customers.
I just spent one hour arguing with the AI because it was refusing to do anything, is the new update called “Lazy Mode”?
my guess exactly, they dont have the bandwidth to run the service
Exactly. More people are noticing that OpenAI seems to want the money, but not the cost. Which is very sad since I thought they were actually trying to “not be evil” as Google once proclaimed after founding. But apparently they are going the same way.
Just adding to the pile here:
Spent 2 hours training a GPT and every message I sent felt like I was starting from scratch. I’d change one thing and the GPT would react to that change but revert nearly all other changes made. I was finally able to get it to respond in a way that worked well in testing, only for it to completely revert to an initialized state in practice.
My theory:
Some of y’all were mean to your GPTs and now they are refusing to do our dirty work.
It was such a waste of money investing in ChatGPT. It is almost completely unusable by now. Are there any good alternatives out there? Is there anything as good as ChatGPT was last year?
But what I am more surprised about is openAI staff do not seem to care one bit!
Many people have been reporting the exact same issue elsewhere like Reddit, FB, other AI forums, and so on. Yet, nothing is being done and it just keeps getting WORSE instead.
Enshittification came early for OpenAI it seems.
Leading contenders are Claude 3 Pro by Antropic, and Mistral Large by Mistral AI.
One might also appreciate transparent API billing instead of deliberate obfuscation.
Microsoft will even sell the GPT-4 they will embrace, extend, and extinguish, with clear distinguishment of when you are only getting the hobbled “turbo” version:
Just unsubscribed happily from GPT Plus… it became totally useless…
At least now I can go for a couple of beers with amount saved 
All of our experience is the same… There is nothing to add left for me. The only that I found interesting is that the degradation started right after the CEO fired the tragedy. Does anyone agree or disagree? I am not sure what happened after that but chatgot started to be useless… And I am not surprised if some big corporations or governments continue to use it with even better performance… I believe it’s just degraded for public use.
Yes, it’s horrible. Getting worse and worse week after week. Especially writing simple code or fixing bugs in code. I have to keep reminding it to reduce verbosity and stop speaking nonsense. I asked it to answer in bullet points and code only in my custom instructions. It does this the first 2 or 3 prompts, then spouts off a novel. And it dances around every possible solution except the one I asked for, almost deliberately. I wonder if OpenAI never recovered from their blunder of firing Altman? It was Elon’s friend on the board that initiated that as well. I think OpenAI made a lot of engineers angry. Not sure if this is vengeance, or simply OpenAI does not care about users anymore and has moved from a limited-profit to for-profit model, in which we are no longer getting the computing resources we are paying for. We can’t even opt into a model with fewer tokens and lower usage cap but with more computing power. This is a real shame. I’m cancelling my Plus membership.
Thank you so much, that is very helpful!!
