Using the chatgpt api, I ran the same completion prompt "Intelligence is " hundreds of times. Given a text, a Large Language Model assigns a probability for the word (token) to come, and it just repeats this process until a completion is… well, complete.
