The online publication CNET will continue to use artificial intelligence to write articles. The editor-in-chief Connie Guglielmo stated.
According to her, the publication intends to keep pace with the times. Each AI-generated article is thoroughly checked by editors before publication. Moreover, the materials were created by a language model not from scratch, but based on a draft written by an employee, Guglielmo added.
She emphasised that the newsroom uses its own in-house development, not third-party tools such as ChatGPT.
In January 2023 стало известно, that CNET had been publishing AI-generated articles for several months. In total, 77 AI-generated pieces have appeared on the site.
Later, readers uncovered a glaring error in one of the articles. The editors promptly corrected it and audited the rest of the texts.
Shortcomings were found in other materials as well. As a result, the publication paused publishing new articles to refine the language model, the writing process, and the text-checking procedures.
During the audit, editors also identified other problems, notably with originality.
Guglielmo promised to fix all the errors. In addition, the editors will begin informing readers that they are reading text generated by artificial intelligence.
In response to widespread criticism, Guglielmo stressed that CNET will continue experimenting with automation tools.
As reported in January, Meta’s chief AI scientist Yann LeCun described ChatGPT не особо инновационным, but a well‑assembled chatbot.
Subscribe to ForkLog AI on Telegram: ForkLog AI — all the news from the world of AI!