No, it really doesn’t, nor does it function like human cognition. Take this example:
I, personally, to decide that I wanted to make a sci-fi show. I don’t want to come up with ideas so, I want to try to do something that works. I take the scripts of every Star Trek: The Search for Spock, Alien, and Earth Girls Are Easy and feed them into a database, seperating words into individual data entries with some grammatical classification. Then, using this database, I generate a script, averaging the length of the films, with every word based upon its occurrence in the films or randomized, if it’s a tie. I go straight into production with “Star Alien: The Girls Are Spock”. I am immediately sued by Disney, Lionsgate, and Paramount for trademark and copyright infringement, even though I basically just used a small LLM.
You are right that nothing is created in a vacuum. However, plagiarism is still plagiarism, even if it is using a technically sophisticated LLM plagiarism engine.