There is a simple way to idiot ChatGPT
Ascannio/Shutterstock
A intelligent use of capital letters could possibly be a simple way to flummox synthetic intelligences like ChatGPT, letting individuals distinguish them from humans in dialog.
The concept is reminiscent of the Turing test, first proposed by pc scientist Alan Turing in 1950. He stated that an AI could be thought of actually clever as soon as we couldn’t distinguish its solutions from a human’s. But now that giant language fashions (LLMs) like ChatGPT can sound convincingly human, Hong Wang on the University of …