College professors and administrators have been busting students for cheating with artificial intelligence for several years. Now, the professors and administrators are getting in trouble, too.
Dartmouth College last week announced it was investigating its provost, Santiago Schnell, after the college’s student-run newspaper found that some of his published writing appeared written by A.I.
Dr. Schnell, a mathematical biologist, is perhaps the most prominent in a growing cast of college professors and administrators who have been accused of using A.I. in a way that crosses ethical lines. There have been instances of opinion articles, research papers and even campuswide messages to students showing traces of A.I. influence. He issued a statement apologizing.
Administrators wring their hands, and sometime issue punishments, when students outsource writing assignments to a robot to save time or avoid the hard work of thinking.
Some students are calling out the examples of faculty doing the same thing as hypocritical.
“It just goes against all the values of our institution and the concept of academia,” Eli Moyse, a Dartmouth senior, said in an interview.
A working group at George Washington University said in a report in August that many students reported “doubts that faculty themselves are using A.I. ethically and transparently.”
And at Brown University, a survey found that close to two-thirds of undergraduate students who responded had some worries about faculty members relying on A.I. to develop course lectures and other content.
At Dartmouth, Dr. Schnell’s A.I. use in public writing came up in an investigation in August by the news site Semafor about A.I.-created writing appearing in the opinion pages of major publications. That investigation said an Aug. 9 commentary by Dr. Schnell in The Washington Post was written by A.I. The opinion article was about A.I. cheating in school.
(The Semafor investigation also found indications of A.I. writing in New York Times opinion articles.)
The follow-up investigation in September by the student publication, The Dartmouth, found extensive A.I. use in Dr. Schnell’s published work this year.
Dr. Schnell told The Dartmouth in a statement that he used A.I. “for language refinement, copyediting, improving clarity and organization and making writing processes more efficient.” The newspaper then ran Dr. Schnell’s statement on his A.I. use through an A.I. detector, which flagged the statement itself as apparently written by A.I.
Dr. Schnell did not respond to a request for comment.
In a note to her campus announcing the review of Dr. Schnell, Dartmouth’s president, Sian Beilock, said that being an early leader on A.I. was “messy.” Dr. Beilock recently called for more A.I. use at her university.
“There will be missteps,” she wrote, “and that’s part of how an institution learns.”
Dartmouth declined to comment beyond what Dr. Beilock had already written.
There should also be consequences, said Mr. Moyse, the Dartmouth senior, who called the A.I. scandal a “completely cartoonish” display of hypocrisy.
“He should resign immediately or get fired,” he said of Dr. Schnell. “Passing something off as your work when it isn’t your work is just plagiarism, even if it’s written by a robot.”
In a letter published Sept. 25 in The Dartmouth, the provost apologized for not always disclosing A.I. use in his public writing.
“Suffice it to say, I am more aware than ever of the need for robust intellectual discussions about the appropriate use of A.I. in the higher education landscape,” Dr. Schnell wrote.
The scope of A.I. use by university employees is not entirely clear, but one of the most comprehensive surveys, conducted last fall by Elon University and the American Association of Colleges and Universities, found that about a quarter of the 1,057 faculty members from across the country who responded were not using generative A.I. tools.
Meanwhile, polling suggests U.S. college students lean heavily on A.I. in their coursework. More than half say they use it at least weekly, and one in five says they use it every day.
Among faculty, nearly nine in 10 said they were not using A.I. for grading, according to the Elon survey. And just 6 percent said they used A.I. “a lot” when preparing their own research papers. Close to half, though, said they turned to A.I. at least somewhat as they created or changed “traditional assignments.”
At the University of California, Berkeley, the student-run paper, The Daily Californian, ran dozens of campus communications through A.I. detectors and found that some missives from top administrators were apparently written or altered by A.I.
Dan Mogulof, a spokesman for Berkeley, said in an email that university leaders and communicators were using A.I. as a tool within certain guidelines and “so long as the final product accurately and authentically captures and expresses the author’s thoughts, perspectives, tone and objectives.”
He added, “We have found that not to be possible without extensive revision, editing and/or embellishment of whatever an A.I. agent might initially produce.”
Student journalists also found that a long essay published by the Berkeley math professor Zvezdelina Stankova in The San Francisco Standard was substantially written or changed by A.I.
Professor Stankova, who declined to comment, told the student paper that she had used A.I. for editing, but the essay, about the shaky math skills of many new Berkeley students, was the result of “intensive human work and deliberation.”
Some schools have specific policies on A.I. use among employees. At the University of Southern California, for instance, faculty and staff members cannot use A.I. to “independently make final decisions” about matters like admissions and discipline. And U.S.C. says that artificial intelligence cannot be used in a way that defies any other university policy.
Still, a set of faculty guidelines for artificial intelligence at U.S.C. suggest the balancing act facing many professors: “There is a difference between what you can do with generative A.I. tools and what you should do.”
Whatever professors decide, the guidelines say, “faculty who require A.I. disclosure from students should generally hold themselves to the same standard.”
Even when institutions have A.I. policies in place, they — like most rules anywhere — are not always followed.
Stanford University has a detailed A.I. policy, for example. But recently, the university used A.I. to remove a Hispanic man from a photograph, replacing him with a Black woman, and to alter the appearances of two other students in the same image. After the changes were discovered, the university acknowledged there had been a policy violation.
Experts say the controversies — and the uncertainty over how or whether to respond — can be traced to how quickly the technology is evolving on campuses. Ethical standards have long governed matters like plagiarism and other forms of academic misconduct, but even written guidelines have not always explicitly kept pace with A.I., leaving the normal boundaries murky.
“The big thing is there’s no social contract right now,” said Marc Watkins, who founded the Mississippi A.I. Institute for Teachers at the University of Mississippi. “We don’t have these sorts of priors to say what’s appropriate and what’s not.”
The post College Leaders Are Using A.I. Students Call It Hypocrisy. appeared first on New York Times.
