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.
This is super cool and very helpful for visualizing how ChatGPT executes completions to "write" content.
If you have anyone in your life struggling to understand how generative AI chatbots works, see if this helps them understand it!