Plagiarism is not always easy to define or detect

Quite a few high-profile careers in higher education have been upended as of late amid questions of academic integrity. Marc Tessier-Lavigne, who had served as president of Stanford University for seven years, stepped down in 2023 after it was determined that he had co-authored several academic papers that contained manipulated data^[1]. The latest casualty is Claudine Gay, who resigned her presidency at Harvard after questions arose about her scholarship, in addition to her response to antisemitism on campus following the Oct. 7, 2023, attacks on Israel by Hamas.

In the following Q&A, Roger J. Kreuz, a psychology professor who is working on a manuscript about the history and psychology of plagiarism, explains the nature and prevalence of plagiarism and the challenges associated with detecting it in the age of AI.

What exactly is plagiarism?

Plagiarism is the use of someone else’s ideas without providing them with appropriate credit or compensation. This could be a melody, a design or some other creative work. Plagiarism isn’t a legal concept^[2], but those seeking compensation can sue the plagiarist for copyright infringement.

In higher education, plagiarism typically refers to the appropriation of words written by someone else. However, the same term is used to refer to a wide variety of behaviors.

Plagiarism can be the inadequate citation of a source^[3], paraphrasing without providing a reference, the wholesale copying of someone else’s work – or anything in between.

And what if a student submits the same paper in two different courses? Some instructors consider this to be unethical self-plagiarism^[4], whereas others do not. The concept is so broad that reasonable people can disagree^[5] about what constitutes plagiarism.

How pervasive is plagiarism in higher education?

Faculty members have greatly differing perceptions^[6] about whether plagiarism is a widespread problem. For example, instructors who teach large numbers of students from other countries, where there are different cultural perceptions^[7] regarding plagiarism, may encounter this issue more frequently.

The lack of a consistent definition for plagiarism means that different studies have arrived at different estimates of its pervasiveness. A 2019 study of university students in Australia found that 64% had engaged in some form of plagiarism^[8] at least once. These researchers found that this number has remained stable for a decade and was down a bit from 82% in 2004.

The actual rate of plagiarism may even be higher than this, however, because people tend to underreport problematic behaviors^[9] in surveys.

The decline in plagiarism over time may be due to the widespread adoption of plagiarism detection tools like Turnitin^[10], PlagScan^[11], and iThenticate^[12]. One study found that when students are given access to such tools, they are less likely to plagiarize^[13]. In another study, however, students were warned that their work would be checked, and this had no effect^[14] on the rates of plagiarism.

Do faculty face different standards than students?

Professors suspected of plagiarism may be given the benefit of the doubt by their institutions or afforded the opportunity to make corrections to their published work. University administrators assume that faculty members know what plagiarism is and how to avoid it, but that they were just a little sloppy with their citations or use of quotation marks.

Students, on the other hand, are governed by fairly strict university policies regarding academic conduct. Harvard, for example, warns that students may be forced to withdraw from the college^[15] if they plagiarize. Work written by students is more likely to be scrutinized and run through plagiarism detection tools than faculty theses or book chapters.

Not surprisingly, many students see this as a double standard^[16]. Concerns about plagiarism in theses written by politicians and university scholars have even led to student protests^[17] in several countries.

How is plagiarism typically punished?

Students can receive a failing grade or be expelled if they’re caught. These consequences are typically spelled out in the school’s academic policies.

At most institutions, however, instructors have a great deal of latitude in how they choose to deal with such cases. Some faculty may be reluctant to accuse a student of plagiarism outright, or they may be willing to give a particularly remorseful student a chance to revise their work.

An underappreciated aspect of academic plagiarism is that it can take an instructor a great deal of time to verify their suspicions.

Plagiarism detection tools don’t always provide a clear answer as to whether plagiarism has occurred. Instead, they provide a score that must be interpreted. For a long document like a thesis or dissertation, it can take several hours^[18] to weed out all of the false positives that such tools typically generate.

Is AI making plagiarism more difficult or easier to stop?

Ironically, while AI has made plagiarism easier than ever, it has also made it easier to detect.

Students can now use chatbots like ChatGPT to generate some or all of their text, and nearly 90%^[19] of students over the age of 18 in one survey admitted to doing so. But this is really more like ghostwriting than traditional plagiarism, since there is no original source – the language is being generated on the fly by the bot.

But chatbots confabulate – or “hallucinate”^[20] – and this can make their errors easy to spot. And plagiarism detection tools like Turnitin and iThenticate can make so-called copy-and-paste plagiarism easier to detect.

Some students use programs called text spinners^[21] to create paraphrases of plagiarized text so that it won’t be detected by services like Turnitin. There seems to be an arms race^[22] between efforts to conceal plagiarized text and the methods that detection tools employ to identify it.

This article has been updated to reflect that a review found Tessier-Lavigne co-authored several papers^[23] that contained manipulated data but found no evidence that he directed or knew about the manipulation in question.

References

1. ^{^} co-authored several academic papers that contained manipulated data (www.npr.org)
2. ^{^} isn’t a legal concept (www.law.cornell.edu)
3. ^{^} citation of a source (academicguides.waldenu.edu)
4. ^{^} self-plagiarism (files.eric.ed.gov)
5. ^{^} reasonable people can disagree (www.insidehighered.com)
6. ^{^} greatly differing perceptions (doi.org)
7. ^{^} cultural perceptions (doi.org)
8. ^{^} engaged in some form of plagiarism (doi.org)
9. ^{^} underreport problematic behaviors (doi.org)
10. ^{^} Turnitin (www.turnitin.com)
11. ^{^} PlagScan (www.plagscan.com)
12. ^{^} iThenticate (www.ithenticate.com)
13. ^{^} less likely to plagiarize (eric.ed.gov)
14. ^{^} no effect (doi.org)
15. ^{^} withdraw from the college (usingsources.fas.harvard.edu)
16. ^{^} double standard (bonniekgoodman.medium.com)
17. ^{^} student protests (www.plagiarismtoday.com)
18. ^{^} several hours (www.theatlantic.com)
19. ^{^} nearly 90% (www.forbes.com)
20. ^{^} “hallucinate” (www.nytimes.com)
21. ^{^} text spinners (todayslearner.cengage.com)
22. ^{^} arms race (www.nature.com)
23. ^{^} co-authored several papers (boardoftrustees.stanford.edu)