They included a multicoloured JAK-STAT signalling pathway diagram which experts likened to “some crazy level of Candy Crush” and said was not grounded in “any known biology”.
The paper, written by researchers at the Honghui Hospital in China, has since been retracted by the journal, which issued an apology and said it was working to “correct the record”.
Dangers of faked research
But several scientists have expressed concern as to how it was published in the first place, and warn it heralds a dangerous era of researchers faking their work using AI.
Adrian Liston, professor of pathology at Cambridge University and editor of the journal Immunology & Cell Biology, said: “Generative AI is very good at making things that sound like they come from a human being. It doesn’t check whether those things are correct.
“It is like an actor playing a doctor on a TV show – they look like a doctor, they sound like a doctor, they even use words that a doctor would use. But you wouldn’t want to get medical advice from the actor.
“The JAK figure is even worse than the rat figure, in my opinion. There are simply meaningless connections and lines that don’t associate with any known biology.”
He added: “The problem for real journals is getting harder because generative AI makes it easier for cheats.
“It used to be really obvious to tell cheat papers at a glance. It is getting harder, and a lot of people in scientific publishing are getting genuinely concerned that we will reach a tipping point where we won’t be able to manually tell whether an article is genuine or a fraud.”
Other scientists took to social media to describe the graphics as “absolutely shameful” and “devastating”, while some said they didn’t know “whether to laugh or cry”.
Prof John Tregoning, of Imperial College London, said the graphics were “objectively funny” but “have no place in science journals”.
A good laugh
Writing on the Science Integrity Digest, Dr Elisabeth Bik, the Dutch microbiologist who works spotting manipulation in scientific papers, said: “Of course, we can have a good laugh at these figures, and wonder how on earth the handling editor and the two peer reviewers didn’t catch this.
“But the paper is actually a sad example of how scientific journals, editors, and peer reviewers can be naive in terms of accepting and publishing AI-generated crap.
“These figures are clearly not scientifically correct, but if such botched illustrations can pass peer review so easily, more realistic-looking AI-generated figures have likely already infiltrated the scientific literature.”
Dr Bik has identified more than 1000 papers that have fraudulent imagery, most of which she believes was generated by AI.
Companies are attempting to develop software that will detect AI-generated content and in the future it may be watermarked, but currently, there is no way to spot it unless it is glaringly wrong.
A spokesman for Frontiers in Cell and Development Biology said: “We thank the readers for their scrutiny of our articles: when we get it wrong, the crowdsourcing dynamic of open science means that community feedback helps us to quickly correct the record.”
The Telegraph has approached the authors for comment.
