What did the author do in response to the 'ChatGPT hunter'?

As the wave of "ChatGPT hunters" spread across the internet, many writers began intentionally adding spelling errors, grammatical inconsistencies, or slang to their work to avoid being suspected of using AI.

The trend of amateur "ChatGPT hunters" is making writers feel uneasy. (Illustration: Wall Street Journal)

According to the Wall Street Journal , as AI-generated content proliferates the internet, more and more people are turning themselves into amateur "ChatGPT detectors." They scrutinize every hyphen, sentence structure, and writing style to determine who is using AI to write articles, post on LinkedIn, or send emails.

Deliberately misspelling words to sound... human.

Sarah Suzuki Harvard, a 32-year-old copywriter and journalist living in Brooklyn, New York City, says she never thought she would have to learn how to "write like a human" in every paragraph.

"I started using very informal language like 'hey yo, for real,' or adding a bunch of exclamation marks. It made me uncomfortable, but now if you want people to believe you're real, you have to write like that," Harvard recounted.

Sarah Suzuki Harvard is a 32-year-old copywriter and journalist based in New York City.

Harvard calls this phenomenon a "new version of witch hunters," alluding to the medieval witch hunts in Europe. "People are demanding that others prove something that is impossible to prove," she said.

The paradox lies in the fact that AI writes well because it is trained using a massive dataset created by humans. This includes countless classic literary works, commentaries, editorials, blog posts, and more. Language models are also optimized to write clearly, coherently, and avoid verbosity – principles that are considered the standard for writing.

Ironically, this very smoothness became a suspicious sign.

Garrett Marcy, a 28-year-old finance employee in Jacksonville, says he often recognizes AI text by its abrupt sentence rhythm, the overuse of long dashes, or structures like "not A, but B."

The fear of losing one's personal writing style.

It's not just ordinary writers; even the tech industry is looking for ways to "ruin" AI-powered text to make it less perfect.

Sean Chou, co-founder of several tech startups in Chicago, says he uses AI to draft LinkedIn posts but changes the long dash to two short dashes so the text "looks like it was written by a human." "It's my artwork," Chou joked.

Chou also tried to avoid overly strong statements, arguing that language patterns are often influenced by TEDx talks or Reddit debates, leading to a tendency to share information with great confidence.

Andy O'Bryan, co-founder of an AI-focused entrepreneurial community, says he sees an increasing number of people intentionally making minor mistakes in AI-generated articles.

"You're reading a blog post perfectly smoothly, and then suddenly a silly error pops up in the middle of a paragraph," he said. "It's as if they're trying to prove the article has nothing to do with AI."

Ivan Jackson is the founder of Writehuman, a startup that provides AI-powered services to "humanize" written content.

Ivan Jackson, founder of the startup Writehuman, argues that distinguishing between humans and AI is becoming increasingly difficult as language models continue to advance. Jackson's company specializes in editing AI text to make it "more human-like," and regularly updates its methods to evade AI detection tools.

According to this startup's analysis, AI currently tends to overuse phrases like "rather than" or "essential for." Paradoxically, many texts written by humans are also flagged as AI because people are unconsciously beginning to mimic chatbot writing styles.

Ryan Johnson, a 33-year-old financial advisor in Michigan, used to use ChatGPT to write blog posts promoting his business. While he appreciated the convenience of AI, he decided to stop last year because he felt his posts were losing their personality.

"It's like a restaurant starting to dilute its signature soup," Johnson said. "Customers don't leave immediately, but eventually they find the dish unappealing."

To maintain his unique style, Johnson often incorporates lesser-known quotes from the series "The Office" into his writing. Even so, he's frequently asked if he uses AI.

Johnson recounted that once, while reading a prayer, he noticed a long dash in the book. He instinctively thought, "Could this part have been written by AI?"

Source: https://znews.vn/tac-gia-lam-gi-truoc-tho-san-chatgpt-post1654242.html