The Provenance War: Scalzi and Stross on the Automation of Creativity

In the world of speculative fiction, the arrival of large language models (LLMs) is being framed not as a technological leap, but as a predatory raid on human creativity, as The Register first reported. For authors Charles Stross and John Scalzi, the conflict with AI labs is a fundamental clash over whether writing is a proprietary human act or simply a data pattern to be automated for profit.

Reporting from *The Register* highlights a visceral disdain for the current AI business model. Charles Stross, a party to a class action settlement against Anthropic AI for the alleged pirating of eBooks, describes AI companies as "absolute scum." According to Stross, these firms are engaging in a cycle of theft: stealing copyrighted material to train models that will eventually compete for the same revenue as the original authors. He likens the practice to a burglar who steals a victim's possessions only to sell them back to the owner the following morning. Stross specifically pointed to Sam Altman in this context, questioning the wisdom of having moved away from 18th-century punishments for such crimes.

While the legal battle centers on copyright, the cultural battle centers on provenance. John Scalzi, known for the *Old Man's War* series and *Redshirts*, argues that the integration of AI into the creative process creates a contagion of doubt. Scalzi notes that the primary issue is not necessarily whether AI material is copyrightable, but that because LLMs are trained on copyrighted works, any generated prose could be contested by the original owners.

This creates a precarious environment for new writers. Scalzi predicts that publishers will scrutinize works for signs of AI use, which will unfairly disadvantage debut authors who lack a pre-ChatGPT body of work to prove their human origin. He further criticizes tech giants Microsoft and Google for "frantically stuffing" AI tools like Copilot and Gemini into Word and Google Docs, respectively. Scalzi maintains that the mere presence of these tools "hovering around" his writing is detrimental from a provenance standpoint.

For these authors, the "creative spark" is an irreplaceable human effort. Stross argues that the value of a well-trained human brain far exceeds that of a tool that does the work the "easy way." He views LLMs as "stochastic parrots" and warns that feeding a story outline into ChatGPT or Claude could lead to the model reproducing that work for other users.

However, the authors distinguish between corporate automation and personal utility. Stross admits he would value a local model—running on his own hardware without cloud services or "copyright-thieving grifters"—to perform technical tasks like deriving scene-by-scene timelines or mapping character frequency distributions. He views such a tool as a "decision-support tool," akin to a spell-checker, rather than a substitute for the hard work of writing.

Ultimately, the resistance is about reputation and autonomy. Scalzi reveals that he requires every aspect of his book production to be handled by humans in his contracts. This is not merely a gesture of honor toward human input, but a defensive measure to ensure his work is not "tarred with the same 'AI' brush" as generated content. In this view, the AI labs aren't offering a partnership; they are attempting to automate the very essence of the craft.