Though proving to be a daydream tool for many industries, ChatGPT is quickly becoming a nightmare for academia.
As of January 2023, four separate research papers have cited the AI chatbot as a co-author in a research project — forcing scientific journals to scramble to update their policies and regulations addressing possible ethical problems.
Read More: The Pros and Cons of Artificial Intelligence
Ethical Issues
The process of adding an author who made little to no contribution to a scientific paper is called honorary authorship, and it’s caused some serious ethical issues in the past. One of the earliest of these issues occurred in the mid-1970s and involved, surprisingly, a cat.
In 1975, a University of Michigan physics professor by the name of Jack Hetherington had just finished writing a rather influential paper on changing particle behaviors at different temperatures. The paper was due to be published in Physics Review Letters, and the deadline was looming.
Unfortunately, a colleague pointed out a problem: Hetherington had referred to himself as “we” in the paper, yet he was the only author, which could cause the paper to be rejected.
So, instead of retyping the whole paper, Hetherington simply added the name of his cat, a Siamese called Chester, as a co-author.
(Credit: Dr. Jack Hetherington, More Random Walks in Science, circa 1975.)
In his book More Random Walks in Science, Hetherington explains that he created Chester’s pseudonym, F.D.C. Willard, by adding Felix domesticus (the Latin name for domestic cats) in front of Chester’s first initial. Then Hetherington slapped Chester’s father’s name, Willard, on as a surname.
Cat’s Out of the Bag
“I did not ignore completely the publicity value,” Hetherington admits in his book. “If it eventually proved to be correct, people would remember the paper more if the anomalous authorship were known.”
Hetherington’s theory turned out to be correct. Not only did the paper become widely cited, but eventually the world found out about Chester’s authorship — arousing even more publicity.
While the University of Michigan leveraged this attention, even offering Chester a faculty position as a Distinguished Visiting Fellow, others were not as thrilled. The editors at Physics Review Letters, for one, felt misled and silly for publishing a paper co-authored by a cat.
The ethical controversy was mostly overlooked at the time, however, and Chester went on to co-author two more papers and one solo paper before passing away in 1982 at the age of 14. His Google Scholar profile shows around 104 citations of his papers.
To honor Chester’s legacy, on April 1, 2014, the American Physical Society announced an open-access initiative for all cat-authored papers.
Read More: 5 Cats Who Owned Famous Scientists
Pet Authorship
Chester’s story is just one of a handful in which scientists have added a pet or animal test subject as a co-author.
Nobel Prize winner Andre Geim co-authored a paper (not his Nobel-winning publication) with an author suspiciously named “H.A.M.S. ter Trisha.” While the paper did not disclose Trisha’s contributions, Geim still was able to add his pet hamster as an honorary author.
Others have not been as lucky. Immunologist Polly Matzinger published a paper with her dog, Galadriel Mirkwood, as an honorary author in the Journal of Experimental Immunology in 1978.
Upon finding out the truth, the journal’s editor banned Matzinger from publication until she died. She also became the subject of an internal investigation at the University of California San Diego.
Luckily, Matzinger was able to show that her dog had indeed contributed to her research and that no fraud had been committed.
Though this example supposes that animal test subjects have more of a right to be listed as co-author on a paper than a mere pet, Hansrudi Lenz of the University of Würzburg argues this practice is unethical.
“Logically, a pet or deceased relative cannot make a genuine and identifiable contribution to a scientific publication,” he says.
Honorary Authorship
The whimsy of these stories can easily obstruct the ethical dilemmas they cause, yet the process of honorary authorship — even beyond pets — continues, thanks to the pressure scientists feel to continually publish.
In a 2020 study, Mariola Paruzel-Czachura of the University of Silesia in Poland and her team found that the most common form of scientific misconduct reported is honorary authorship, with 52 percent of the study’s participants observing this process.
“It could be some kind of bribe,” Parazel-Czachura says, offering a possible reason for such a high percentage. “It could help a researcher get a better job, or funding for a conference. It could even be a partnership, where both researchers agree to add each other as honorary authors.”
This, however, causes problems with authorship inflation, in which an author’s number of citations is higher than it should be because they seem to have “published” more papers.
AI Writers
Now, with the information age and all it brings (looking at you, ChatGPT), it’s even easier for researchers to practice honorary co-authorship. Because of this, most scholarly journals are finding it more difficult to regulate AI co-authors.
“We’re trying to take the most cautious approach that we can,” says H. Holden Thorp, the editor-in-chief of Science. “We’ll start with something more restrictive and then loosen it up over time.”
In a January 2023 Science editorial, Thorp cites a couple of these policy changes — including a complete ban on the use of any part of the text, images, figures or graphics made by ChatGPT or other AI tools.
Thorp, like others, hopes that the National Academy of Sciences will address this issue soon, in order to set the precedent of dealing with an AI honorary author.
The path moving forward may lie, in fact, in a paper published by researchers from the University of Cologne in Germany, way back in 2017. They write: “The appropriate way of considering [others’] factual role in scientific publications should generally be in the Acknowledgements section.”
The authors then went on to thank their goldfish, Einstein and Heisenberg, in this very section.
Read More: AI and the Human Brian: How Similar Are They?
