ChatGPT Can Fool Humans Even When It's Wrong, Backs Up Assertions With Fake Quotes

https://ift.tt/NFqSAEY
ChatGPT, OpenAI's powerful chatbot that can generate sentences to mimic human-written text, is capable of fooling users, even when it is wrong, according to experts. The chatbot was able to generate a news story on Microsoft's earnings for 2021, and while the earnings figures were not accurate, it also generated a fake quote from CEO Satya Nadella that appeared to be ...
Originally posted here: https://ift.tt/fmnuGEP