Two pieces this week, from opposite ends of the cultural spectrum, are quietly arguing about the same thing. The Atlantic published a defense of ghostwriting, arguing that if you have a story to tell but not the skills, hiring a professional is not only fine, it is good. Simultaneously, a 2026 paper in arXiv CS.CY by Dongsoo Han empirically investigated how AI-augmented peer review systems are changing scientific productivity across countries, finding that AI intermediation is reshaping who gets credited for intellectual labor at the institutional level. Put them together and you get the central anxiety of the current moment: if the voice is borrowed and the review is assisted, what exactly is the byline certifying?
The Craft Argument and Its AI Problem
The Atlantic piece is careful to distinguish ghostwriting from fraud, arguing that the ghostwriter's skill is real labor deserving compensation, and the subject's experience is real content deserving a platform. It is a reasonable argument that has worked for decades in memoir, business books, and celebrity publishing. But the piece notably frames AI as the thing ghostwriting is not, the soulless alternative that proves human collaboration has value. This is a fragile distinction. A 2026 arXiv paper on fine-tuning LLMs for epistemic reasoning through Navya-Nyaya logic frameworks by Sharath Sathish found that LLMs struggle with systematic reasoning precisely because they flatten the provenance of claims, they cannot track who knew what and when. That is exactly the problem with AI ghostwriting: not that it produces bad prose, but that it cannot carry genuine epistemic responsibility.
The Soderbergh Problem: Fabricated Oeuvres and Real Credit
Artnet this week covered the art created for Steven Soderbergh's new film, which involved fabricating an entire artist's oeuvre for the production. Fictional artists with real-looking work. The piece asks, entirely earnestly, how that art was made and whether it constitutes a legitimate creative practice. It is the ghostwriting question in a different costume. The film needed an artist. They made one. The work is real. The artist is not. For startups navigating similar questions about AI-generated content and intellectual property, TurboFund's list of 25 AI seed-stage investors tracks which funds are actively betting on the companies trying to resolve exactly this tension. The byline, in every field, is now a provisional claim.