Is the Guardian skating on thin ice with AI?

‘It wasn’t done by AI, our writer is just very bad,’ is a curious explanation for a media organisation to have to give in response to concerns about one of their articles – but it seems to be the reason the Guardian has offered to explain a very oddly written piece about figure skating on its website.

Under the headline ‘Anatomy of an upset: how Ilia Malinin lost Olympic figure skating gold’, Bryan Armen Graham reports on the American’s surprise defeat in the Winter Olympics at the weekend, deploying many of the tropes often associated with articles written by large language model (LLM) AI programs such as ChatGPT.

Common AI tics such as ‘not X but X’, ‘this isn’t X, it’s…’ and ‘this wasn’t just…, it was…’ are littered throughout the article, along with others rarely used except by AI, such as an explanation ‘sitting inside’ something or someone ‘quietly’ winning Olympic titles. On X, the writer Ian Leslie showed how he had put the article through the online AI detector Pangram, which gave it a 100% chance of being generated by AI.

Suggested Reading

Toothless Ofcom lets GB News off the hook, again

Rats in a Sack

“What made Ilia Malinin’s Olympic defeat so shocking was not simply his years-long dominance entering Friday night,” the article read. “It was how completely the competition had tilted in his favor before he even stepped on the ice.

“What made the result feel even closer to inevitable was what happened around him.”

The Guardian has denied that AI was used in writing the piece, saying: “Bryan is an exemplary journalist, and this is the same style he’s used for 11 years writing for the Guardian, long before LLMs existed. The allegation is preposterous.”

But many on social media have shared previous pieces by the writer, showing a much more florid style than the robotic one published this weekend. Have the robot overlords finally taken charge at King’s Place?