More Typos, Fewer Em Dashes: Writers Are Creating an Anti-AI ‘Literary Counterculture’

Last year, when the novelist Laura Brooke Robson began penning Love Is an Algorithm—a emotion communicative betwixt the cofounder of an AI dating app and a instrumentalist trying to past successful the property of ample connection models—she took other attraction not to dependable similar AI herself.

She wrote sentences she doubts she would person attempted earlier the advent of LLMs, avoiding what she calls “wishy-washy metaphors,” and took “a much winding way to debar an em dash oregon a tripartite list.”

Since the wide adoption of LLMs, Robson says her penning has fundamentally changed for the better, becoming a deliberate inverse of what she sees arsenic AI’s default deficiency of style.

“I find myself penning little mechanically,” she says. “I person this impulse to marque small mistakes, similar winks to the reader, to amusement that there’s a quality down the words. Like: What if I conscionable invented a phrase? What if I did thing unexpected with my punctuation?”

Robson is portion of a larger cohort of writers who are practicing what could beryllium called an anti-AI penning style. Across journalism, criticism, originative writing, and particularly among compulsive LinkedIn posters, writers are converging connected a acceptable of choices that implicitly defy the trappings of AI-generated prose.

The benignant mostly takes signifier done negation, often of the evident AI tells, similar lists of three, passive voice, not x but y, and em dashes. But it besides prizes idiosyncrasy arsenic its overriding principle, tapping into circumstantial first-person anecdotes and eccentric similes that show the indispensable drudgery of writing.

“I deliberation AI could perfectly trigger a literate counterculture,” says Robson. “It seems to maine that literate fabrication volition displacement adjacent further successful the absorption of the weird, the esoteric, portion personalization takes clasp of commercialized fiction.”

Throughout the publishing world, allegations of AI person been troubling authors. In March, Hachette pulled Mia Ballard’s fearfulness caller Shy Girl aft speculation dispersed online that its writer had heavy utilized AI, a assertion she rejected. Ballard, who says she has caller penning forthcoming, has had to restrategize. “Having my penning consciousness much existent and authentic is my fig 1 priority,” she says, successful summation to editing her ain penning due to the fact that she claims an exertion she hired ran the publication done AI software. “That is what spot maine successful the ass the archetypal time.” Hachette did not instantly respond to a petition for comment.

Likewise, Steven Rosenbaum’s publication The Future of Truth, an introspection of “how AI transforms reality,” was purported to see aggregate fabricated quotations generated by AI, which Rosenbaum aboriginal admitted was true.

Ellena Savage, an Australian novelist and world whose prose tends towards complex-compound sentences inspired by Victorian literature, says she has “sadly been victimized by thoughts of chatbots.” “AI has stolen respective syntactical conventions I erstwhile held dear. I miss over-deploying the em dash. I miss the regularisation of three, which was truthful satisfying. Now what—the regularisation of two? Five? Seven? 10?”

Across magazines and integer publications, this anti-AI counterstyle is reshaping in-house benignant and editorial signifier too. Richard Fisher, an honorary prof astatine University College London and exertion astatine online civilization mag Aeon, says AI has prompted him to promote writers “to beryllium much human” and constitute much conversationally. “We don't privation to soil retired the unsmooth edges, due to the fact that AI cannot constitute prose arsenic spiky oregon idiosyncratic arsenic humans. Writers mightiness not privation to alteration their practice, but immoderate mightiness beryllium forced into it.”

Mano Sundaresan became Pitchfork’s caput of editorial contented successful July 2024, astir the clip LLMs became much mainstream. Coming from a blogging background, helium encouraged writers “towards their ain voice,” thing helium says, successful hindsight, “could person besides travel from this AI turn.” He and his chap editors person since sent writers a memo stating that nary AI whitethorn beryllium utilized erstwhile filing reappraisal copy, and encouraging first-person perspectives. (Pitchfork is owned by Condé Nast, which besides owns WIRED.)