I use chat gtp for creative storytelling. Its like your own cutome story. But over the past few months the respnses have been getting worse. Its memory is worse i have to keep reminding it of facts or correcting it. Its responeses are shorter and less creative and dynamic. Once a file becomes to long it crashes and i cant fix it just have to give up the file. And now my pro subscription witch was unlimited is now not unlimited. And the unlimited is $200 compared to the $20 im currently paying. How do they expect a regular user to pay $200 a month thats insane. If it keeps getting worse i may just have to give up gtp and go back to Kindle unlimited and read kther peopleās storys. Its so annoying because this is the future of entertainment and they are running it.
In just the last few days the chatgpt mini-o4-high has become unuseable. It is the one recommended for coding and I have used it pretty regularly. It would analyze code and find issues pretty quickly and the suggestions might not work on the first go but there was a troubleshooting process that would get you there. But since around Friday it has become hopeless. The answers sometimes refer to lines of code that donāt exist, the suggestions make no sense, it wonāt give complete answers, it will give you back the same code you just uploaded, it gives invalid code snippets - and it never improves. It went from being a useful tool to a complete waste of time. What happened?
Iāve definitely noticed a significant drop in intelligence in the later models. o4-mini-high takes the same time to think as 03-mini-high but the depth and quality of the answers has suffered noticeably. For example, in planning the build of a new business, I told the new model that I plan to start with a single work van and then later upgrade to a second van. It responded as if I already had the second van.
This happens with almost every response now. Small but important details get overlooked and often replaced ideas that were clearly hallucinated.
I honestly had a fear this would happen. I was using ChatGPT for another project recently and decided to take a break to handle this current project. My worry has always been that OpenAI might somehow screw things up. Seeing as my prior project relies on accuracy of details, I am now worried that when I return, I wonāt be able to really use ChatGPT how I used to.
As far as whatās behind the decline in intelligence. My mind goes the conspiracy route (and often I am right in one way or another). As more and more users sign up looking to make their lives a little easier (and since OpenAI still has a cashflow problem). I would imagine the team is scrambling to āoptimizeā resources. Lowering the intelligence of the newer models is one way to conserve resources and distribute to all the new users. Thus, democratizing the platform further at this risk of decreasing its usability. Sounds reasonable to me, though itās not okay.
If not that, then I would go full blown conspiracy and say that giving ambitious individuals access to infinite employment of a highly intelligent machineāto which can put that individual on par with some of the greatest corporations of the dayāis not something OpenAI (whose dual structure is still 49% owned by Microsoft) is NOT authorized to do. ĀÆ_(ć)_/ĀÆ
Whatever the case is, its highly annoying and leaves an opening for other AI providers to dominate. I still use DeepSeek and though its not quite on par with ChatGPT (previous models)āit is dangerously close and less annoying, retaining details much better and even appearing more cynical in a very useful way.
Me too, Plus subscription. Iāve used it for programming. It started off great, very useful, clear, good memory, clear and precise workflow. All that ended a few days ago. Now it gives bot-like responses, makes mistakes, suggests things that donāt work or that itās already suggested before, and itās getting worse. At this rate, theyāre going to lose what little they have left and weāll lose a great tool that they ruined so disastrously.
Iāve also been experiencing problems for at least the past week. I noticed mine started having memory retention problems from after 10th April. Unfortunately the problems have become progressively worse ever since. Iāve reached out to support 6 times already this past week (via email) but keep getting fluff replies and no help, escalation route or assistance. Iām rather cross to be honest, as Iām a paying plus user and they wont even help me try to resolve it.
