…have a look at this article:
ChatGPT is a fancy auto-complete: what it does is finding the most likely continuation of text for the text it has already printed out (or as a reply for the prompt, when it is starting printing stuff out), according to the large corpus of text it has been trained with.
Pixy Misa at AoSHQ put it this way: “ChatGPT cannot think, only speak.” That statement got me thinking about a few folks I’ve known who can’t do either. You know, the blokes who spout all manner of nonsense without ever forming a coherent sentence. Some are drug burnouts; others are schizophrenics; still others are victims of a substandard upbringing and a “public” education.
And some, determined to find a trade where their disabilities would be assets, have gone into politics:
“Hire the Handicapped! They’re fun to watch.” – Originator unknown
A month ago, NPR did a test where some real, human rocket scientists were asked to judge the AI application by asking it rocketry questions. It failed spectacularly.
The problem is, it’s trained to make up rules for its responses – the autocomplete analogy is good. In an area like rocketry equations it knows nothing and has no abilities to go look up equations and interpret how to use them. They’re missing knowledge of which things they read or get asked about are rock hard physical law and which are more like suggestions, or that can be made secondary.
It’s much ado about nothing. Oh, it can write a few paragraphs or pages that’s supposed to sound like some writer. BFD. Who cares?
“Hire the handicapped, they’re fun to watch.” Was on Radio Dinner by the National Lampoon, I believe. Along with a mocking of Paul McCartney: “Give Ireland back to the Irish, give Lapland back to the Laps, give China back to the Chinese, and Yoko back to the Japs!”
Thank you! One does learn something new every day — if he’s paying attention, at least.