I’m at my wits end. I subscribed to Plus so that I can use its picture input capability for it to look at my website and give me feedback, which GPT-4 refuses to do.
When I provide it with the link to the website it says it can’t access it, and when I ask it about a different way to let it take a look it says that it also can’t look at pictures! In some random cases even GPT-3.5 suggested that I send it an Imgur-Link, so in my case GPT-3.5 actually has image interpretation capabilities while GPT-4 doesn’t.
I am extremely disappointed and confused. What the hell is going on here?
If you read the press release from OpenAI, it states directly that the image process capabilities have not been released for
I see, thanks! Still don’t understand why 3.5 can already do it while 4 cannot.
Can do what exactly ?
I do not think
gpt-3.5 processes images as input.
Are you getting your info from a chatbot @Prince_of_Privacy ?
Look, I know GPT 3.5 is not supposed to be able to process images, but in some random cases it could. I attached a screenshot that proves it. And it didn’t hallucinate and actually gave a correct response.
That is NOT an image in your screenshot @Prince_of_Privacy
That is how the ChatGPT web UI formats code.
Are you serious right now? It’s not about ChatGPTs code formatting, it’s about it accessing the imgur link and giving instructions based on it.
Of course I am “serious” @Prince_of_Privacy
The screen show you posted has no images generated by ChatGPT.
What you “think” is an image is not an image.
It is a nicely formatted code block.
You are totally and 100% mistaken @Prince_of_Privacy …
It is 100% impossible that ChatGPT “accessed any link”.
You @Prince_of_Privacy are the only “thing” hallucinating with regard to ChatGPT.
ChatGPT (the underlying model) has zero capability to access the Internet. Zero. ChatGPT is using a pre-trained LLM which the “cut-off” time for data back in mid 2021.
ChatGPT has zero image processing capability.
This post was flagged by the community and is temporarily hidden.