Tuesday, February 7, 2023
Home Tech News CNET found errors in more than half of its AI-written stories

CNET found errors in more than half of its AI-written stories

/

The tech news site came under scrutiny this month after it was reported the outlet was using AI to write articles. Last week, executives said the use of the AI tool would be paused ‘for now.’

CNET issued corrections on 41 of the 77 stories the outlet published that were written using an AI tool. In a note published today, CNET editor-in-chief Connie Guglielmo defended the use of the AI writing tool but said that an internal review of stories uncovered numerous errors in the articles at the center of the controversy.

Articles with corrections include ones titled “What Is Compound Interest?,” “How Much Should You Keep in a CD?,” “Does a Home Equity Loan Affect Private Mortgage Insurance?” and dozens more. A handful of articles, like this one, had the correction note reading, “We’ve replaced phrases that were not entirely original,” indicating that some of their language may have been plagiarized.

Earlier this month, Futurism broke the news that CNET had been quietly publishing articles written by AI for months without drawing much public attention or making a formal announcement. In a follow-up story, the outlet noted numerous errors in a CNET article about compound interest, which eventually resulted in a lengthy correction. Following the errors, a disclaimer appeared at the top of all AI-written stories: “We are currently reviewing this story for accuracy. If we find errors, we will update and issue corrections.” 

Last week, The Verge reported that automated tools have been in use at CNET for much longer than the article-writing robot and that staff sometimes didn’t know if content was written by a machine or a human co-worker. The AI-written articles are designed to game Google searches with SEO-friendly keywords so lucrative affiliate ads can be plastered on the pages. CNET’s parent company, Red Ventures, which also owns publications like Bankrate, The Points Guy, and CreditCards.com, stands to benefit every time a reader signs up for a credit card from one of the highly trafficked articles.

After weeks of debate about CNET’s disclosure policies around AI tools, Red Ventures and CNET leadership told staff in a meeting on Friday that the company was temporarily pausing AI-generated content across all websites. The errors, though, don’t appear to be stopping CNET’s use of AI tools.

“Expect CNET to continue exploring and testing how AI can be used to help our teams as they go about their work testing, researching and crafting the unbiased advice and fact-based reporting we’re known for,” Guglielmo wrote in her memo today. “The process may not always be easy or pretty, but we’re going to continue embracing it – and any new tech that we believe makes life better.”

- Advertisment -

Most Popular

Murder trial: Lawyer grills William Sandeson on his ‘two very different sides’

A Crown lawyer grilled William Sandeson about what she called his “two very different sides” during his first-degree murder trial Tuesday. Sandeson, 30, is accused...

OnePlus 11 review: the first flagship of 2023 doesn’t disappoint

The OnePlus 11 looks similar to the OnePlus 10 but the camera module is now circular (Picture: Metro.co.uk)The OnePlus 11 is the first Android...

Analogue opens up final preorders for its beautiful takes on the Sega Genesis and SNES

/ The Super Nt and Mega Sg consoles are going on their final production run and are estimated to ship in 2023.The Mega...

Microsoft’s Adaptive Accessories are a tradeoff

/ They’ve got genuinely helpful features, but the materials could use some work.Like many people, I have invisible accessibility needs. I’m not in...