| >with all the prior knowledge ChatGPT had gained As currently built (as I understand it), it doesn't "gain" anything. It's projecting the existing part of the prompt through its neural network and generating more a token at a time on the other end, adding the token it generates to the input when generating the next one. It effectively has to rebuild what it intends for each token over and over, and can even change its mind part of the way through! (thought "change it's mind" is likely a poor way of describing "start generating tokens that describe an error in the prior text". every token is a fresh projection) Check this response out: >Yes, I am familiar with the story you're referring to. The title of the short story is "The Machine That Won the War." It was written by Isaac Asimov and first published in 1961. The story is a conversation between three men who played major roles in a war against an alien race, and they discuss the role of a machine called Multivac in winning the war. > >However, it seems that I've made an error in my recollection, as the specific detail you mentioned—refusing to work until the engineer says 'please'—is from a different short story, "Sally," also by Isaac Asimov. In "Sally," autonomous cars stop working until a command is given courteously, using the word 'please.' The above is just a single response I received when seeing if gpt-4 could help me remember the name of an old Isaac Asimov story I liked. After it had generated the tokens for the first part, it self-corrected and gave me a second answer. (which was still not what I wanted, but that's aside from the point. Asimov was prolific, no surprise that even an AI can't keep track of all of it :D ) |