American newspaper every morning CNETa major name in the technology journalism, publishes several articles. Almost all on tech topics: smartphones, computers, cloud, artificial intelligence. But in some of CNET’s news artificial intelligence, instead of being the subject of discussion, is the author: Several articles in the “CNET Money” section have been written using AI. Even if, reading the article, it seems to have been written by a real person.
CNET and the news written with artificial intelligence
In the CNET Money session, every day at 9 am East Coast time, two stories come out about mortgage interest rates and refinance rates. Just below the title, find the box “Written by: Justin Jaffe”, the editor-in-chief of the site’s finance section. But they don’t appear on his author page, but on CNET Money’s.
In the bio of this account, we read: “This article was assisted by a artificial intelligence engine and reviewed, verified and modified by our editorial staff“.
Futurism reported last week that the website has started to produce entire articles written by artificial intelligence, starting last November. But only with a few extra clicks can users understand that these news stories are not written by the editor-in-chief Jaffe, an expert in the financial sector, but by a bot that changes the wording of a few sentences every day but presents the same article.
This news has not only confused readers, but also some of the editorial staff of CNET, which has The Verge reports that since the private equity firm Red Ventures bought the site and started testing generative AI to create content, they don’t know for sure if the news is written by a colleague or a bot.
For readers or for SEO?
If you haven’t visited CNET in a while, you might be surprised to learn that he writes about interest rates. The magazine made its name reviewing MacBooks and writing about Silicon Valley news. With enormous success, so much so that in 2008 CBS I’m buying it for a whopping $1.8 billion. But after the changes of the last few years, in 2020 CBS sold the company to Red Ventures for less than a third: 500 million dollars.
Red Ventures has a clear business model: Publish content designed for rank high in Google search (by optimizing articles for Google with SEO) and monetize that traffic with profitable affiliate links. Specifically, Red Ventures has found an important niche in credit cards and other financial products. An industry that is also familiar with other proprietary sites: Bankrate and CreditCards.com, for example.
All of these publications have used artificial intelligence to write articles such as those that appeared on CNET. If a user clicks on an affiliate link and signs up for a credit card, the site receives figures that appear to be around around $250. According to reports from the New York Times, the figure for Red Ventures would even reach 900 dollars.
The CNET Case: Does Artificial Intelligence Help Journalists or Just Increase Profits?
Following the Futurism report, CNET Editor-in-Chief Connie Guglielmo wrote an article detailing the head’s approach to AI-generated news. Guglielmo first of all said that it is an experiment, asserting that “For over two decades, CNET has built its reputation testing new technologies and separating hype from reality, from voice assistants to augmented reality to the metaverse“.
In this ‘pioneering’ spirit, he explains that “since November, our editorial team at CNET Money began testing the technology to assess whether there is a pragmatic use case for an AI assistant for basic explanations on financial services topics. Like ‘What is compound interest?’ and ‘How to Cash a Check Without a Bank Account’. So far we have published about 75 of these articles.”
The editor-in-chief explains that she uses the term “AI assistant” because all articles are controlled by a human editor (assuming Jaffe, whose signature appears below the title) before publication. And he specifies that the goal is “to see if technology can help our busy staff of journalists and publishers to cover topics from a 360-degree perspective.”
In other words, AI should serve to remove the most ‘boring’ articles from journalists’ diaries, thus allowing them to work on more interesting insights. But The Verge’s sources report several layoffs starting in Novemberwith morale in the CNET newsroom appearing to drop lower by the day.
Transparency makes the difference
CNET isn’t the first news organization to use AI to publish news that doesn’t require a lot of editorial work. Also there Associated Pressan international press agency, uses the tool Wordsmith since 2014 (the same one used by CNET, it seems). AP began using it to write quarterly reports of large companies, then expanded in 2016 with sports scores. Currently collaborates with Data Write to create AI-generated articles for this content and also for betting.
Especially on such technical and short subjects, distinguishing between an article written by a person and an artificial intelligence like WordSmith or ChatGPT becomes difficult. For this reason AP writes that “The Associated Press created this story using technology provided by Data Skrive.” Some might argue that ‘technology’ is not as clear as writing ‘AI-generated’. But it is certainly more transparent than reporting the name of an author who limits himself to checking that there are no errors.
Here too, it seems that clearer rules are needed on how to distinguish between one type of content and another. Between a robotic author and one in the flesh, between an automatic search on Google and a careful work on the sources. In order to leave readers with more information about who wrote the news and how they wrote it.
CNET reporters, however, seem less scared of AI than the company they work for, according to The Verge. “They don’t fear AI any more than they fear the many layoffs that Red Ventures has insisted on. Everyone at CNET is more afraid of Red Ventures than AI.”