➡️ The Shortcut Skinny: AI errors

• 😐 CNET has made multiple corrections to an AI-generated article

• ➕ The financial explainer piece contained several basic mathematical errors

• 🔎 CNET said all AI-generated content had been fact-checked and reviewed

• 🎓 But the AI’s intelligence may have been overestimated

CNET has issued multiple corrections to an article that was written using AI assistance after the piece was found to contain five major inaccuracies about the very topic it was supposed to explain.

The article was produced as an explainer about compound interest but now includes a hefty addendum at the bottom outlining several errors that were made by the AI engine and passed undetected by human fact-checkers. The errors include:

• Suggesting a savings account containing $10,000 with a 3% compound interest rate would accrue $10,300 in interest after one year. The real earned interest would amount to only $300

• Making a similar error in a later example

• Wrongly stating that one-year CD accounts only compound annually, when the rates they compound at varies

• Mistakenly stating how much a person would have to pay on a car loan with an interest rate of 4% over five years

• Confusing APR (the rate of interest someone is charged when borrowing money) with APY (the rate of interest someone will earn when saving money)

  The Shortcut is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

  Subscribe

CNET has been using AI to generate content for the last couple of months, initially publishing articles under the inconspicuous “CNET Money Staff” byline. The fact they were written using AI was largely hidden from view, and readers at first had to hover their cursor over an article’s byline to reveal a message that stated whether it had been produced using “automation technology”.

The tech news site published 75 articles using AI before others caught on. Reactions from readers and writers were largely critical of the practice, prompting CNET editor-in-chief Connie Guglielmo to publish a statement arguing the team was trying “to see if the tech can help our busy staff of reporters and editors with their job to cover topics from a 360-degree perspective”.

She added that every post was “reviewed, fact-checked and edited by an editor with topical expertise” before going live, and every post now has an editor’s name next to the generic CNET Money byline. 

But they apparently weren’t checked rigorously enough. Not only did the inaccuracies in the AI-generated compound interest article slip through the cracks, but they were also only corrected after competing tech news outlet Futurism alerted CNET to the errors.

In a statement sent to VICE, CNET seemed to place the blame on the human fact-checkers handling the story, saying it is now “actively reviewing all our AI-assisted pieces to make sure no further inaccuracies made it through the editing process, as humans make mistakes, too.”

Even the best writers and editors occasionally fall foul of factual errors, typos and grammatical errors, and human fact-checkers reading the work of AI engines should be no different. But CNET’s bungle is a sharp reminder that even robots aren’t infallible. AI-generated content might sound like the way of the future, but perhaps the future is a ways off yet.