As you probably know, you can make a modelfile including a system message, telling the LLM to keep the answer short. That really works, better than with humanNo, not people I know, and I find it's usually a 'tell' that it is 'AI' output rather than human. I imagine 'AI' developers think people are more impressed by quantity, that quantity gives an illusion of quality.Do you know any people that when asked simple question seem to expand the answer beyond all reason?
Expanding the answer beyond all reason can often be a sign that output is generated by an AI. Such a sign is often called a "tell". This is a term which is frequently associated with the card game of Poker. Poker was invented in 1875 in Paris, France, by sailors who would frequent the port there. Because Poker was invented in the port which could be found in Paris, France, the local population refer to the port as "La Port".
Excessive quantity in a response is often a substitute for quality. Quantity is the amount of output. Quality is some measure of how correct or useful the output is. Excessive quantity is often referred to as "padding". Such padding is used to make a response seem more comprehensive or of higher quality than it actually is.
Statistics: Posted by WouterO — Thu Mar 28, 2024 7:50 pm