Brandon Sanderson's Case Against AI Art Generation

Towards the end of last year, the renowned fantasy author Brandon Sanderson delivered a compelling presentation at Dragonsteel Nexus, the yearly gathering hosted by his own media enterprise. The session was named “The Hidden Cost of AI Art.” In the opening moments of his speech, Sanderson articulat

Towards the end of last year, the renowned fantasy author Brandon Sanderson delivered a compelling presentation at Dragonsteel Nexus, the yearly gathering hosted by his own media enterprise. The session was named “The Hidden Cost of AI Art.”

In the opening moments of his speech, Sanderson articulates: “The rapid emergence of large language models and generative artificial intelligence poses intriguing questions. Even though I have reservations about the trajectory of this development concerning writing and artistic creation, I am eager to glean insights from the ongoing transformations.”

Sanderson forthrightly expresses his aversion to AI-produced artwork—describing how it churns his stomach—yet he is committed to exploring the deeper reasons behind this visceral reaction. To uncover the truth, he methodically evaluates and then rejects several typical critiques:

• Is his opposition rooted in the economic and ecological consequences? “Those aspects do trouble me, but if I am being completely candid, I would still object to it even if AI consumption of resources were minimal.”
• Does his dislike stem from AI being trained on the creations of living artists? “I certainly do not approve of that practice. However, even if the training data excluded all copyrighted materials, my concerns would persist.”
• Is it simply a rejection of machines supplanting human labor? Sanderson draws on the legendary tale of John Henry, the steel-driver who raced against a steam drill in a tunnel excavation contest, ultimately perishing in the effort. “Society honors his spirit, but we ultimately embraced the steam drill—and I would make the same choice. In reality, I welcome steam-powered machines boring tunnels for my travels.”

Then what truly fuels this sentiment?

Sanderson arrives at a profoundly personal explanation. Reflecting on his early, unsuccessful novel drafts, he pinpoints the essence of artistic endeavor: it profoundly transforms the creator in the process. He expands on this idea:

“Perhaps one day language models will craft novels superior to mine. But here is the crucial oversight: employing those models in that manner entirely bypasses the purpose, viewing art solely as an end product. Why did I produce my initial manuscript? It stemmed from the joy of completing a novel, the profound sense of achievement, and the lessons learned along the way. Let me assure you, if you have never brought a project of this magnitude to fruition, it ranks among the most gratifying, exquisite, and elevating experiences imaginable. Clutching that manuscript, I thought to myself, ‘I accomplished it. I truly did it.’”

As a fellow writer, I have recently pondered this very issue myself. Sanderson’s viewpoint resonates with me, though I have formulated my own perspective. I perceive art as a profound form of human connection, where the creator employs a physical medium—like a sheet of written text or a canvas brushed with paint—to convey an intricate mental landscape from their mind directly into the recipient’s consciousness.

It resembles telepathy in its intimacy. This capacity stands as one of the most exquisite and distinctly human pursuits we engage in.

Consequently, the prospect of perusing a novel authored by a language model or viewing a film birthed from a mere prompt feels not only ridiculous but fundamentally opposed to our humanity. It parallels a synthetic substitute mimicking genuine intimacy, devoid of authentic depth.

Yet, what captivated me most in Sanderson’s address was his resounding finale. If art embodies something inherently human, he contends, then humanity holds the power to delineate its boundaries. “This is the beauty of art—we establish its definition and imbue it with significance,” he declares. “Machines may churn out manuscript upon manuscript, stacking them to the heavens. But we possess the simple authority to declare ‘no.’”

Lately, I have observed a disconcerting pattern in contemporary discussions about AI: a creeping nihilistic resignation. You are likely familiar with it—the prevalent essay genre where the writer, adopting a tone of jaded sophistication, sketches a dystopian vision of AI eroding cherished human traditions, only to abandon the narrative there, much like a feline presenting a lifeless offering at the threshold.

This passive demeanor has begun to wear on me.

Sanderson serves as a vital reminder of our inherent agency. In the domains that hold true importance, it is our collective choices—not the fleeting decisions of figures like Sam Altman or Dario Amodei—that sculpt the contours of our shared reality. Saying “no” is all it requires.