Art Only Blog Of Rabia FQ

Require it to be labeled. that's it. require every instance of AI output to be clearly and obviously labeled that it is AI. Every decision made with AI has to be disclosed in plain phrasing, every email, every paper, every image and video must be clearly and obviously watermarked. Make removing the label/identifier a procecutable offence.

I am an archeologist. I specialize in a somewhat obscure but by no means boring or meaningless Neolithic culture in Germany.

It has a Wikipedia page. A well curated, surprisingly extensive Wiki page that encapsulates all the important information about the culture, including literature references for further research.

One day, we asked Chat GPT about this culture. W were curious which details it would get wrong.

ALL OF THEM, except for the fact that it's a culture in present day Germany.

It didn't even get the chronological time frame wrong and called it a celtic culture.

When we told it it's wrong, it came at us with made up literature sources. Literally made up. It took two well known German archeologist who weren't even active at the same time, added a year - both were already dead - and sold that as source.

And it LITERALLY would only have had to quote Wikipedia to get everything right.

THAT is how unbelievably shitty and wrong all those AIs are.

They are making shit up. They are not sourcing information, they're just slapping words together by their most like relative occurance.