Very funny. I put in 3 screen captures of a (long) document, and it did relatively well. But when I proof-read it, I realized the AI has made up passages that were not there!
The reason is probably due to the nature of screen capturing, some sentences or paragraphs were cut short. That probably kicked off the “fill in the blank” nature of the LLM and it could not resist to leave these paragraphs stand unfinished :LOL. It even put in a short conclusion paragraph that was not in the original document at all!
replies(1):