This is an interesting take on an over discussed topic. The transcript argues that traditional publishing has a double standard around AI: industry insiders may use AI in their own writing while discouraging authors from doing the same. It challenges the idea that AI-assisted writing is inherently illegitimate. Key points: - Authorship is about creative authority, not generating every word. Authors have long used editors, researchers, ghostwriters, and collaborators without losing authorship. - AI training is not the same as piracy. Courts distinguish between building libraries from pirated books and using lawfully acquired books for transformative fair use. Learning from existing works is also how human writers develop. - AI can support legitimate authorship, including for people with disabilities or those facing barriers to polished prose. Wealthy authors have long used human assistance; AI democratizes access to similar help. - Institutional definitions of authorship are problematic. Redefining authorship to require a certain amount of human-generated text gives publishers, agents, and organizations power to police creative processes. - Secrecy and compliance distort behavior. Many authors hide AI use due to backlash, while conscientious beginners may avoid AI unnecessarily. Publishers and agents may use AI themselves while enforcing strict bans on authors. - Distinguish creative authority from contractual authority. Publishers can set submission rules, but those rules should not define legitimate authorship universally. Authors should set their own ethical standards. - Copyright and authorship are separate. The U.S. Copyright Office distinguishes human-authored expression from AI-generated content, which creates practical problems for collaborative works, but copyright frameworks should not define who qualifies as an author. - Authors should retain creative control, use AI only when it serves the project, investigate contract clauses, and avoid letting fear of judgment hinder their development.