I've got a small corpus of human-generated writing (my own and a few collaborators) that various "AI-detectors" flag as "AI-generated" at very high rates. I'd like to see what Turnitin's AI Detection does with this corpus, but none of us who are involved in this work have access to Turnitin. Would anyone who DOES have access be willing to run our corpus through it?
Boosts welcome!
@writerethink I would rather train people to mimic the bots and do other things to turn all chatbots into Tay.
Our main line of defense now is data poisoning.
Maybe we could train the chatbots to attack the tech billionaires. It may be our only chance to stop this idiocy.
