Academic journals are awash in AI-generated letters to the editor.

From American Public Media, this is Marketplace Tech.

I'm Megan McCarty Carrino.

Dr. Carlos Chakur at the University of Navarra noticed something fishy about a letter to the editor the New England Journal of Medicine received shortly after it published a paper of his on malaria treatment in July.

The letter was riddled with strange errors.

Critiques supposedly based on other research

Chakur himself had written. So he and his co-author, Matthew Rudd, decided to dig deeper.

They analyzed patterns of letters to the editor over the last decade and found a remarkable increase

in what they call prolific debutantes,

new authors who suddenly had dozens, even hundreds of letters published,

starting right around the time Chat GPT came out.

Why would academics want to do this?

We asked Chakur to explain.

Academia is a very twisted place, full of very bad incentives, I would say.

Because if the culture is published or perish, then I will always choose publish.

So in the academia, this culture that you need to get published often and as impactful as possible is creating perverse incentives.

And this is just a new way of cheating, but people have been cheating since for a long time.

I don't know if you've heard of Elizabeth Pick and the work she has done with

image forensics. And what she does is she takes images of papers and in which they report,

for example, a PCR gel or some cells. And then she finds out areas that are duplicate. Basically, people are using

some image software to duplicate some areas and fake results. And that's been happening for years.

So there's incentives on the researcher side, this idea that you need to get promoted, you need to get tenure.

...