RE: ChatGPT May Have Irreversible Consequences for Learning and Decision-making

I note that the public statements regarding the nature of ChatGPT are demonstrably false. It is not merely a LLM. An LLM cannot invent text. It can only weight examples it has been trained on. Since ChatGPT and it's ilk have proven to generate text that did not exist, they are something more than LLMs. The capability to lie and create new text that they could not have been trained on is part of their programming.

I have pointed this out before, and cautioned people to recognize that if they do not grasp this reality and continue to wield ChatGPT and it's ilk, they are grasping a weapon intended to be used against them - by the pointy end.

Thanks!