I'm trying to use ChatGPT to get information from a PDF file for a product I'm building, but it doesn't seem to work. Whenever I upload the PDF, it just makes up information instead of providing accurate details from the document. I even tried uploading screenshots of the PDF, hoping it would understand those better, but it still seems to ignore them and generate nonsense responses. I thought ChatGPT had improved since earlier versions, so what's going on?
1 Answer
I haven't run into issues myself. I upload my files to the Projects, and when I start a new chat each day, I can hear it scanning the files. It usually gives me back relevant parts of the files when I ask about them, sometimes even referencing the file directly. I also upload images into Projects, and those are read accurately too! Maybe you could try that?
That's interesting! I was using it right before a big update, so maybe that was affecting performance? I had better luck with PDFs before that update, and I was pretty shocked at how off it was this time.