fluid current
Bug Report:
Steps to reproduce:
- functioning python script with question for adjustments in GPT3.5
- functioning python script with question for adjustments in GPT4
Expected result:
a piece of code that's not jibberish (doesn't need to work necessarily)
Actual result:
GPT3.5: "what can I help you with"
GPT4: screenshot
Additional information
Browser: Chrome (issue persists in Edge and Firefox)
OS: Windows 10
Extra info:
I've been using ChatGPT for a few days now to help me with some functions in Python here and there (quite new to Python myself). So far it has been helping wonderfully until about 14 or 15 hours ago. I thought id give it a nights rest but today still the same odd behaviour.
I state that I am working on the python script for a Discord bot. I then submit part of my code (1300 words long give or take) and after parting the code i ask in a few lines for the adjustments.
In gpt3.5 it keeps replying with "how can I help you?" or similar results or occasionally when it feels daring it tells me I am working on a script for a Discord bot and that it can help if I submit my code (a lot of things wrong in that reply)
In gpt4 the answers are a lot more interesting (attached screenshot).
The script I feed the AI is a fully working script from top to bottom that has some of the following functions (using the gpt API to generate messages on command or as a greeting. fetch and store data on a google spreadsheet and some other minor functions)
I have tried the following:
- Try in a new chat
- Use a different Model
- Rephrase my questions
- Clarify the project I am working on first before sending the code
- Use the script from an earlier backup that it DID work with perfectly before to ensure there's nothing in my current script triggering this behaviour
- Try 3 browsers
- Gave it a nights rest
Nothing seems to work.
In the screenshot given I have opened a new chat, GPT4.
In my first message i declared to project that I'm working on and I need some adjustments. It replies in an normal manner and asks for the code.
I supply the code followed by the adjustment needed and the response is what you see in the screenshot. I have had that response a few times using GPT4, every time its a bit different but the "Run the bot" in large bold letters is always there.
Not sure if this is a bug or if I do something wrong. But as it did work before I do not see what I am doing wrong now and I am inclined to believe it is a bug or the AI is having a heatstroke.
Sorry if this post is not according to the "Bug report format" and if the supplied extra info was unnecessary.
