Exercise 1: Why "AI and Copyright" Isn't One Question — Possible Solution ==================================================================== WHY TREATING IT AS ONE QUESTION IS A MISTAKE ------------------------------ Per this chapter, "AI and copyright" actually splits into three genuinely separate legal questions - whether training itself infringes (Question A), whether AI output can be owned (Question B), and whether AI output can infringe an existing work (Question C). Each has "its own, differently unsettled answer" and its own distinct legal reasoning - collapsing them into one question obscures that a statement true of one question may be entirely irrelevant, or even contradictory, to another. A CONCRETE EXAMPLE OF HOW COLLAPSING THEM CAUSES CONFUSION ------------------------------ Someone might hear that AI training is defended as fair use (an argument specific to Question A, resting on Chapter 4's transformative- use Factor 1) and wrongly conclude that any AI-generated output is therefore automatically safe from infringement claims too. But per this chapter, Question C is a completely separate analysis - "ordinary infringement analysis applies regardless of how the allegedly infringing material was actually produced." A model could plausibly have been trained under a legitimate fair use argument and still produce a specific output that infringes a specific existing work, because Question A's fair use analysis (about the training process) says nothing at all about Question C's separate substantial-similarity analysis (about a specific finished output). WHY QUESTION B SHOWS THE SEPARATION EVEN MORE CLEARLY ------------------------------ Per this chapter, Question B (can the output be owned) turns on an entirely different legal principle - human authorship - that has nothing to do with whether training was lawful (Question A) or whether a specific output infringes something else (Question C). A perfectly legally-trained model's output could still fail to be copyrightable at all if it lacks sufficient human creative input, a completely separate failure mode from either of the other two questions. WHY THIS MATTERS PRACTICALLY ------------------------------ Answering one question doesn't answer the others, and each requires its own separate legal analysis under its own separate body of reasoning (fair use's four factors for A, human-authorship doctrine for B, ordinary substantial-similarity infringement analysis for C) - treating a resolution to one as a resolution to all three would be a real, substantive legal error, not just an imprecise way of talking. WHY THIS WORKS AS AN ANSWER ------------------------------ It restates the chapter's own three-question framework, gives a concrete example of exactly how conflating two of the questions leads to a wrong conclusion, and explains why each question rests on a genuinely separate legal principle rather than being three phrasings of the same underlying issue.