A majority of AI-generated articles published on CNET in recent weeks are believed to be at fault.
The tech news site had used artificial intelligence to write articles from November. This practice was suspended a few days ago. And for good reason, the majority of the texts produced contained errors.
41 out of 77, the problematic ratio
With the emergence of AI, and the enormous popularity of ChatGPT for a little over a month, each profession is wondering if it is currently affected by this technology, which could tend to replace many professions. And journalism is obviously no exception, as the CNET case shows.
The media had indeed used artificial intelligence to write texts, before ending the practice. And today, we understand why. According to The Vergeof the 77 articles that used AI, 41 contained errors and had to be reviewed by the human eye.
Just as problematic, several corrected articles now also display the mention: “ We replaced sentences that weren’t entirely original. The authors thus suggest that artificial intelligence may have plagiarized content.
CNET will continue to use AI
CNET had launched its initiative without officially informing the public, and had started posting an indicative note only after the first polemics. However, the group does not intend to do without this technology in the future.
” We have taken a break and we will start using the AI tool again when we are sure that this tool and our editorial processes will avoid human error as well as that of artificial intelligence. explains in a note Connie Gugliemo, the editor-in-chief.
Meanwhile, again according to the journalist, CNET teams will continue to test the use of artificial intelligence for content creation. With the idea of making it an assistant working in constant collaboration with the human being?
Source : The Verge
4