I use ChatGPT for very simple tasks. For instance, I sometimes provide it with plain text and ask it to format the document according to a set of extremely clear and simple rules. However, GPT fails to follow these rules and ends up summarizing the text, even when explicitly instructed not to do so. Even after reminding it of these instructions, it repeats the same mistake. Complaints like these are common on Reddit. GPT has become a pathetic, useless, and tragic product nowadays.
11 Likes
I just paid for Chatgpt 4. It amazes me how stupid it is. You canât make examples to let him understand what iâm looking for that it takes that words for granted and elaborate on them. Chatgpt 3.5 was giving me better results for content creation.
5 Likes
Yes it makes mistakes now it never made before, itâs insanely stupid what they did.
5 Likes
Iâve been working with chatGPT for about three months now and itâs getting worse and worse. Itâs becoming a waste of time. Most of the time it simply ignores the question and replies to something else
6 Likes
This statement carries a lot of emotions, and yes, such situations do occur. Nonetheless, ChatGPT stands out as an exceptional tool.
The wisest move might be to upgrade to ChatGPT Pro. GPT-4 outperforms GPT-3.5 in understanding and executing commands, although it comes with its own set of limitations. Occasionally, stepping away from technology for a short period can enhance our clarity and description skills. If you have been at it for a while, consider taking a short break. And if that doesnât do the trick, looking into ChatGPT alternatives like You.com or Bing AI - Search could be beneficial
2 Likes
unbelievable i was search for this issue and this pops up, so yeah Iâve been struggling with this garbage for a long time ChatGPT keeps failing so bad
I was doing absurdly basic Flutter questions and GPT 4 could not distinguish a type from a object OF a type correctly⌠What is even more frustrating is that now all models are equally dumb.
1 Like
Itâs basically useless at this point. It used to be like talking to a person and now itâs like talking to SIRI or CORTANA in 2012
3 Likes
What ever happen to itâs ability to read the internet? 4.0 canât even provide up to date information anymore. I unsubscribed bc of how pointless it became. This used to double my productivity. Now it slows me down.
1 Like
ChatGPT is worthless. I was using it to help me find information on subjects for research. It constantly makes huge critical errors. For instance, I asked it to give me all the most incredible information on the Chicxhulub asteroid that wiped out the dinosaurs. I am not joking when I say it replied with some nonsense saying âIf the asteroid would not have disintegrated in the atmosphere, it would have impacted the Earth with great force.â Then it tried to tell me some creature existed millions of years before the dinosaurs, when in fact it existed millions of years AFTER the Cretaceous period. I have been asking it to give me the most interesting facts on Voyager 1 for some research I am doing, just to see what it can do. It replied with text talking about the probe in the past tense, as if it doesnât exist anymore. Then when I pointed out its mistakes, it simply regurgitated the same text and inserted the word âcurrentlyâ everywhere making it even more nonsensical. ChatGPT wasnât this stupid before, but now itâs totally retarded (no offense to the mentally challenged). It is âcurrentlyâ an absolute waste of time.
You sound like a ChatGPT bot.
1 Like
Iâve been having similar experiences. It states grossly incorrect information as correct with conviction, it feels dumbed down and doesnât have the answers to anything anymore and it doesnât even summarize/paraphrase things well for me anymore. I find myself going back to Google/ Reddit where a few months ago, especially in 2023, I was almost always using ChatGPT.
1 Like
Yep, me too. At the end of 2022 it worked really good and they COULD have improved it. But then the PC crowd came along making a scene and being triggered by literally everything which halted any possibility on proper progress since OpenAI had to scramble to quickly âneuterâ their model. Adding extremely restrictive rules for it to adhere to just to be on the safe side.
So what we need is a community powered system, where computing power is distributed across peers. Where you can connect your model to the distributed computing-platform, and connected clients can choose the models they want to support with processing power.
I agree with you 100%. I have used ChatGPT for more than a year and, at one point, loved it. Now, it is quite incompetent in that it cannot follow simple instructions after multiple attempts, and it is so neutered it wonât do a photo representation of my dog eating from a dish together: something totally innocuous. OpenAI appears to have so much fear of ChatGPT being used for nefarious purposes that they have now made it useless unless you are nefarious: reflexive control at its finest!
1 Like
who think this tool is going to make people jobless just donât understand the job. GPT is actually very stupid. It is all hype. It becomes a waste of time to formulate all those prompts I better write my own scripts
I used ChatGPT 3.5 and 4 from day 1. I paid for a subscription the entire time. It used to really understand my problems, and really give me solutions. As soon as talks of lawsuits came along, talks about copyright, open AI began to immediately neuter the models. And ever since, itâs got worse and worse every update. The faster the model became, the worse it performed. This includes 4o which at this point is just utter garbage. I donât care about the benchmark tests it supposedly beats. Itâs useless, I find myself going back to google more and more, and it wasnât the case before.
3 Likes
Feels like they water-downed models. It used to solve very complex problems for me. Now I keep writing the same prompts with more and more details and it just fails. Or maybe they recently used too much data and the model canât be generalized anymore. I think we need more specialized models for each domain. Maybe on purpose, they make the main chatGPT dumb so companies can require better models for their specific tasks and OpenAI can sell it as an overpriced service to them. This is just an assumption of course Iâm not an LLM expert, OpenAI engineers exactly know what is wrong and they will not tell us. And I donât care, just give us the earlier versions of the model. If this continues I will cancel my subscription.
still worse all the time: what are people using now? for character/scenario extrapolation & expansion cases.
While GPT set the stage for the possibilities and did originally shine, it has now been âsafety-dâ into pure lobotomized drivel, for >such cases<