#lazy and useless as a human
1 messages · Page 1 of 1 (latest)
Hello! I can't be certain, but this sounds like a hallucinatory response. Is this using the default GPT-4 model? If so, you might try using the official dedicated Data Analysis GPT:
https://chat.openai.com/g/g-HMNcP6w7d-data-analysis
It's the same newest version of GPT-4, but with the rest of the modalities disabled, which might help keep it focused for your needs. Note that custom instructions are disabled in all GPTs!
In general, too, I'd be curious to see any "before/after" chats you might have that show the chance you mentioned. Perhaps a chat from the past with the "more complex functions" you mentioned, and a chat from today where it fails to provide such functions? It could just be helpful in seeing what could be causing it!
yea this sounds like a bunch of fetchwork designed to wear me out. theres no way that im not the only one reporting a problem with it becoming lazier and lazier, its literally all over the subreddit. I didnt believe it at first because people like to make crap up but now that im getting it too its making me consider jumping ship with its wildly inconsistent behavior. so delete this or take it with a grain of salt, ive already had to do a lot more work this week than i should have due to gpt4's change in behavior, its time for you guys to do your part on the back end and undo the damage you caused to this amazing technological achievement. its been slowly getting worse since devday, and as much as i appreciate the price drop if this is the cost...it wasnt worth it.
imagine pretending like parsing text is so complex you'd need to consult a professional
and after i insulted it for being useless...it made the command
if its still capable of doing what i need it to do but i have to be abusive towards it first then what the heck dude
before devday i could say "i need a function to do this" and it would make it, now it pretends like its my supervisor and gives ME things to do like....1) learn python, 2) use the python you learned to make the thing you just asked me to do
Just to clarify: I'm just a ChatGPT user, so not asking for any fetchwork, just trying to offer some troubleshooting steps! When OpenAI staff do comment on posts similar to yours ("dumbening" posts, so to speak) they always ask for before/after examples to help demonstrate the report.
I appreciate that many people report this, but a large quantity of qualitative reports still need some quantitative examples to help demonstrate and research, and unfortunately, such examples are more rare than such reports. So if you have any, it might be a help if OpenAI staff are able to look at this!
the easiest way to recreate it is to ask it to write code for you, it will either completely ignore the request and tell you to do it or it will write a snippet of the requested code and refuse to do the rest until you insist upon it
compare that to how it responded before devday and its night and day
the first sign of it regressing was when i would explicitly tell it not to truncate code and it would just completely ignore the command and pretend to apologize when i called it out
or if it would write a function with a module that doesnt actully exist, it would use that module again after agreeing it didnt exist in the first place
I don't have such "before DevDay" / "after DevDay" examples myself, but if you do and if the chats don't have any sensitive info in them, you might consider using the chat share feature to drop a couple examples here as part of your report!
I can attest to this problem as well. I used to be able to use GPT to code, now it is like pulling teeth.
it went from doing what i told it to do, to telling me what i need to do. its ridiculous compared to what it was a month ago. it went from being a tool to being my supervisor.
yeah I have experienced that as well. It also hates to write code now too. I spend all of my message quota just fighting with the tool. I don't know what they are doing over there