AI-generated content is posing a threat to the quality of Wikipedia, and its editors are taking action to protect the platform.
According to 404 Media, a group of Wikipedia editors has formed a task force called “WikiProject AI Cleanup.” Their goal is to tackle the growing issue of AI-created content that is poorly written or lacks proper sources.
While the group does not intend to prohibit all AI use, they aim to eliminate cases where AI-produced material is full of errors, misleading information, or lacks credibility. Their efforts focus on preserving Wikipedia’s role as a trusted, accurate information resource.
“The goal of this initiative is not to prohibit or limit the use of AI in articles,” states the Wikipedia forum of the proactive group, “but rather to ensure that the AI-generated content is accurate and beneficial. If the output does not meet these standards, it will be corrected or removed accordingly.”
According to editors who spoke with 404, misuse of AI on Wikipedia is sometimes easy to identify. Common giveaways include users accidentally leaving typical chatbot responses in entries, such as paragraphs starting with “as an AI language model, I…” or “as of my last knowledge update.” Editors have also become familiar with certain writing styles and “catchphrases,” helping them detect and remove poorly crafted AI-generated text.
“We noticed an increase in unnatural writing that clearly seemed AI-generated, and we were able to replicate similar styles using ChatGPT,” said Ilyas Lebleu, a founding member of WikiProject AI Cleanup, to 404. He further explained that recognizing specific AI catchphrases made it easier to find some of the worst examples of AI-written content.
However, spotting low-quality AI content isn’t always straightforward, especially when mistakes are presented with confidence within complex topics.
One instance highlighted by editors to 404 involved a well-written but completely fictitious history of a “timbery” Ottoman fortress, which never existed. Although entirely inaccurate, the content was convincing enough that, unless you were an expert in 13th-century Ottoman architecture, you likely wouldn’t have noticed the mistake.
As previously mentioned, Wikipedia editors have at times downgraded the credibility of certain news sources, like CNET, after it was found publishing flawed AI-generated articles last year.
Since it’s cheap to generate AI content in large quantities, controlling poorly-written AI text is a significant challenge. Combined with the fact that Wikipedia is a volunteer-based, crowd-sourced platform, tackling the flood of low-quality AI content becomes even harder.
Relevant News You May Like