A growing number of professional writers are deliberately altering their prose to appear more human and avoid accusations of using artificial intelligence, a shift that threatens to subtly reshape online communication and content creation. This reverse Turing Test involves using "aggressively casual" language, intentional typos, and avoiding overly polished sentences that have become hallmarks of large language models (LLMs). The trend comes as AI-detection software becomes more widespread and the volume of machine-generated content explodes, with one recent analysis estimating that AI-written articles now outnumber those produced by humans.
"It’s like the new McCarthyism," said Sarah Suzuki Harvard, a 32-year-old copywriter in Brooklyn who has started using more casual language to prove her human bona fides. "People are demanding proof of something that can’t be proven."
The push to sound more human has led writers to avoid common AI tells such as the frequent use of em dashes, lists of three, and phrases like "it's not x, but y." Some, like financial account coordinator Garrett Marcy, admit to using AI for drafting but then edit heavily, even re-inserting their own run-on sentences or typos to reclaim their voice. Software like Writehuman, a startup that edits AI text to sound more human, has emerged to capitalize on this trend, noting that its analysis shows current LLMs overuse phrases like "rather than" and "essential for."
This stylistic shift has minimal direct impact on financial markets but carries significant implications for the $250 billion content creation and media industries. As the line between human and machine-generated text blurs, the perceived value of authenticity is rising, potentially creating a premium for verifiably human content. This could impact everything from freelance writing rates to the perceived credibility of corporate communications and marketing materials, forcing companies to navigate a landscape where sounding too perfect can be a liability.
The fear of being falsely accused of using AI is a primary driver of this trend. Writers have seen careers damaged by such allegations, as in the case of the book "Shy Girl," which was pulled by publisher Hachette over claims it was partly AI-composed. This has led to a climate of suspicion where, as one writer noted, even reading historical texts with em dashes can trigger a momentary "is this AI?" check.
The irony is that AI models are trained on vast amounts of human writing, learning to emulate the clarity and style praised in guides like Strunk & White's "The Elements of Style." Yet, it is this very polish that can now be a giveaway. In response, some are fighting back with idiosyncrasy. Ryan Johnson, who runs a blog for young families, quit using ChatGPT for drafts because he felt it was "watering down the soup" and stripping his writing of its distinct voice, which includes obscure quotes from "The Office."
The Arms Race in AI Content
The phenomenon reflects a broader "escalating arms race" between those using AI and those trying to detect it, as described by Richard She, a stem-cell biologist at Nanyang Technological University. Studies are beginning to quantify the scale of AI's footprint. A recent analysis of nearly 7,000 manuscript abstracts submitted to the journal Organization Science found that submissions with more than 70% AI-generated text had more than doubled by February of this year compared to early 2024. Another preprint study screening over 124,000 manuscripts on arXiv found that computer science review preprints with AI-generated text surged from about 7% in 2023 to 43% in 2025.
This proliferation is not limited to academia. In China, the mass adoption of AI tools for everyday tasks has made it a vast testing ground, with over 600 million people using generative AI as of December—a 142% year-over-year increase, according to the China Internet Network Information Center. This widespread use, from booking travel to generating business websites, is rapidly closing the performance gap between U.S. and Chinese AI models.
For investors, the key takeaway is not about avoiding AI, but understanding its evolving role and the market's reaction to it. While venture capitalist Marc Andreessen champions a more "provocative, aggressive" AI free from "woke" constraints, the market for now seems to be rewarding a different kind of authenticity. The rise of "humanizer" tools and the conscious effort by writers to inject "flaws" into their work suggest a developing market for services that can either verify human authorship or convincingly mimic its imperfections. This dynamic could influence investment in AI detection companies, content marketing platforms, and the broader creator economy, where the definition of "value" is being actively renegotiated.
This article is for informational purposes only and does not constitute investment advice.
