JAKARTA - AI chatbots are starting to be used to generate news and blog posts for online content in hopes of attracting small advertising revenue from lost web users.
Experts have warned over the years that AI-generated content farming will soon become commonplace, but the wider availability of tools like the ChatGPT from OpenAI has now made the warning a reality. NewsGuard, an organization that assesses news site confidence, highlights this issue in a recent report identifying 49 sites "which appears to be almost completely written by artificial intelligence software."
NewsGuard said:
This website, which often fails to disclose ownership or control, generates large amounts of content related to various topics, including politics, health, entertainment, finance, and technology. Some publish hundreds of articles every day. Some content advances false narratives. Almost all content displays a bland language and iterated phrases, which characterize artificial intelligence.
The sites identified by the organization often have generic names (like Biz Breaking News and Market News Reports) and are full of programmatic ads purchased and sold automatically. They write news articles and write writers with generic or fake names, and most of the content seems to be a summary or repeat of stories from well-known sites like CNN.
"Most of the sites don't spread misinformation," NewsGuard said. But some published clear lies. For example, in early April, a content farm called CelebritiesDeaths.com posted a story claiming Joe Biden had died.
The story about Biden may have tricked readers, although it was immediately revealed as fake news. The second paragraph contains incorrect messages from chatbots that are requested to make text and are clearly copied and installed to websites without supervision.
"Sorry, I can't complete this request for violating OpenAI's policy of using misleading content," the story said. "It's unethical to make fake news about someone's death, especially someone so famous like the President."
NewsGuard says they use such mistakes to find all sites in its report. As The Verge has reported, searching for phrases such as "As an AI language model" often reveals where chatbots are used to generate fake reviews and other cheap text content.
NewsGuard also verifies that text on these sites is generated by AI using detectors such as GPTZero (although it should be noted that these tools are not always reliable).
Noah Giansiracusa, a professor of data science associations who has written about fake news, told Bloomberg that such site makers are experimenting "to find what is effective" and will continue to create content farming given the cheap production costs. "Previously, it was a low-paid scheme. But at least it wasn't free," Giansiracusa told Bloomberg.
The English, Chinese, Japanese, Arabic, and French versions are automatically generated by the AI. So there may still be inaccuracies in translating, please always see Indonesian as our main language. (system supported by DigitalSiber.id)
