Preventing Plagiarism and Cheating in the Age of Artificial Intelligence

Literature review

According to Jolicoeur and Pagé (2015) [in French]:

• Plagiarism is the act of not respecting the rules for citing sources. Plagiarism may be intentional or unintentional (due to ignorance of the rules or false beliefs, for example).
• Cheating and fraud are linked to deliberate actions (or negligence).

So when the rules for citing sources are intentionally disregarded, it’s both plagiarism and cheating.

Still according to Jolicoeur and Pagé (2015), the preferred intervention approach to plagiarism is prevention. When it comes to fraud and cheating, the focus is on dissuasion.

I’ve summarized my research into plagiarism and cheating in a concept map. I present the various elements below.

Concept map summarizing my research on plagiarism and cheating[in French]

Factors surrounding plagiarism and cheating

Pedagogical factors

Test weighting

Farland and Childs-Kean (2021) argue that students are more likely to cheat if the stakes of an evaluation are higher.
Since graded evaluations are associated with higher stakes than ungraded ones, the authors suggest that ungraded work should be favored (as is the case, for example in the ungrading approach).

For graded assignments, several evaluations with lower weighting (or an assignment divided into several steps) undoubtedly reduce the temptation to cheat [in French].

Type of evaluation

Gremeaux (2019) argues that open-book exams (or exams for which a memory aid is allowed) limit the temptation to cheat, compared with closed-book exams based on memorization.

In an open-book exam, students have a sense of applying knowledge, rather than just having to learn by heart, and they appreciate this. However, while open-book exams reduce the incidence of cheating, they don’t eliminate it completely.

Information skills

For Peters and Cadieux (2019), students’ informational and writing skills and their ability to cite their sources (knowing when, how and why to cite their sources) are key to avoiding plagiarism. These skills need to be taught (Monney et al., 2019 [in French]). Liu et al. (2021) showed that workshops to develop information and research skills were beneficial to the international students who attended them.

Similarly, students’ unfamiliarity with the plagiarism detection tools used by teaching staff can increase their anxiety (through fear of being the victim of a false positive from the tool, or of unintentionally plagiarizing and being punished too severely because of detection by the tool) (Zaza and McKenzie, 2018).

Students’ language level

A study by Perkins et al. (2018) carried out at a university in Vietnam where courses are taught in English to non-native English speakers shows that students who are less proficient in English are more likely to plagiarize. If the same is true in the college network, then helping allophone students to improve their mastery of the language of instruction could reduce the number of cases of plagiarism.

As an ESL teacher, I’ve observed that the students I’ve caught cheating over the years often claim that their initial placement is inaccurate (i.e. the level of the course is too high for them) or that they have a sense of low competency (which amounts to the same thing). This suggests a link between cheating and their experiences.

Institutional factors

Institutional response

Brooks, Marini and Radue (2011) [PDF] argue that breaches of intellectual integrity are acts of incivility and that their impact on the learning environment is often ignored. It is also observed that the institutional response is uneven, both from one institution to another and from one teacher to another.

Mulholland (2020) writes that institutional policies related to plagiarism and cheating often associate plagiarism with a moral problem (in a way equating it with cheating) rather than with a lack of education on the part of students. She argues that this is detrimental to students, who instead need to acquire the skills they need to avoid plagiarism (research skills and the ability to cite sources). In Mulholland’s view, Canadian post-secondary institutions need to shift their paradigm from a punitive to a rehabilitative vision of higher education in response to plagiarism.

Jolicoeur and Pagé (2015) [in French], meanwhile, identify 5 intervention targets in the fight against plagiarism at college:

• restrict the ease of cheating and plagiarism
• act against students’ ignorance of plagiarism
• reduce the possibility for plagiarists or cheaters to use unclear instructions as an excuse or argument
• act on the level of risk perceived by students
• mobilize teachers and encourage them to report cases of plagiarism

Workload

Among the many factors influencing plagiarism, Espinoza and Najerà (2015) emphasize the importance, for teaching staff, of working as a team to ensure that the workload of students in any given program is manageable.

Intrinsic factors

Perception of the seriousness of the act

Gremeaux (2019) [in French] discovered that helping each other during an exam is judged less severely by students than an individual cheating alone, without the help of someone else. Similarly, students who admit to cheating are more tolerant—or even indifferent—to this behaviour than those who don’t cheat. Honest students judge cheating much more harshly.

Perception of impunity

According to Choo and Paull (2013) and Bennett (2010), students cheat for many reasons, including the perception that the risk of significant negative consequences is low. For Choo and Paull, teachers have an obligation to act to change this perception.

Time management

According to Michelle Bergadaà (quoted in Perreault, 2014) [in French], for some students, plagiarism is simply a way to save time, to balance a schedule that’s too busy for them.

Narcissism

Laily, Ermayda and Azzardina (2021) have shown a correlation between narcissism and plagiarism or cheating. Celik and Kanak (2021) have established a link between narcissism and self-sabotage.

Language anxiety

Abasi (2008) [PDF, in French] analyzed the impact of language anxiety on the plagiarism practices of English as a second language students, as well as the impact of coercive communication strategies on their anxiety.

The author explains that students experience anxiety at the thought of being caught as plagiarists, not least because they don’t always know exactly what it is to plagiarize; what does and doesn’t constitute plagiarism. For example, when teaching a second language: is using a turn of phrase plagiarism?

Desire to pass the course

Al Darwish and Sadeqi (2016) believe that what leads students to cheat is the feeling of not having the ability to succeed on their own, combined with the desire to pass the course to get their diploma.

Extrinsic factors

Peer and cultural influences

Fontaine, Frenette and Hébert (2020) write that peer influence is the most influential factor on a person’s propensity to cheat on exams.

Zhao et al. (2022) confirm that peers play a role in the decision to commit ethical misconduct and that strategies to promote academic integrity must consider the cultural context.

Similarly, Scrimpshire et al. (2016) found that cheating, and in particular minor cheating (copying homework or collaborating when forbidden — as opposed to major cheating, which would be cheating during an exam), is a social and personal act:

• Students tend not to cheat alone, but to cheat with friends.
• Students are more likely to help friends cheat than strangers.
• Cautious people are less likely to take part in an act of cheating.
• Those who have cheated before are more likely to help others cheat.

Normalizing plagiarism

The normalization of plagiarism on social networks has been described by Amigud and Lancaster (2020) and Bailey and Trudy (2018). Much of this involves contract cheating, i.e. “the submission of work by students for academic credit which the students have paid contractors to write for them” (Clarke and Lancaster, 2006 [PDF]).

In this respect, one has to wonder if ChatGPT isn’t a contract cheating provider… that works for free.

Google’s informational infrastructure

Boubée (2019) [in French] hypothesizes that Google’s overly simple and non-selective access to information sources stops students from fully acquiring theoretical knowledge through information retrieval. Paraphrasing this knowledge, i.e. reformulating it through clarification, a major “anti-plagiarism” skill, then becomes difficult.

Employment status

Bennett (2010) found that the students most likely to plagiarize are those with a casual attitude to plagiarism and who hold part-time paid jobs that they feel interfere with their studies.

A complex situation

Thus, there are many factors to consider when it comes to preventing plagiarism and cheating. As teachers, we sometimes forget that our students aren’t just brains on sticks. They are complex beings, and their behaviour in our courses depends on more than just what we say in our lectures:

• if students have a 40-hour job and 8 courses a week, they are more likely to cheat
• if students are experiencing a high anxiety level they are more likely to cheat
• if students come from a culture where plagiarism is normalized, they are more likely to cheat
• if the exam is closed-book, students are more likely to cheat
• etc.

To combat plagiarism or know how to react when it occurs, it’s useful to be aware of all the factors involved.

The role of the teacher

As teachers, we certainly have a significant role to play in preventing plagiarism and promoting intellectual integrity.

Ignore plagiarism?

In a study by Coren (2011) [in French], 40% of participating teachers admitted to having ignored one or more cases of cheating. The reasons teachers gave for ignoring misconduct included:

• insufficient evidence
• triviality of the offense
• lack of time

However, the author observes that teachers who had ignored misconduct:

• felt more stress when discussing cheating with their students
• preferred to avoid emotionally charged situations
• said they would be less inclined to talk to a student if the person was likely to become emotional

Obviously, if a teacher ignores one or more cases of plagiarism, this has an effect on students’ perception of impunity.

Administrative isolation

As Jolicoeur and Pagé (2015) [in French] write, mobilizing teaching staff and involving them in reporting cases of plagiarism must also be goals actively pursued by the institution.

Having the institution take responsibility for part of the overall prevention strategy and for overseeing the reporting process helps to reduce the isolation experienced by teachers with respect to plagiarism and cheating. In the medium term, it can be assumed that teachers who feel supported when they report a case of plagiarism or cheating will be less tempted to adapt the sanctions or not apply them at all. The same applies to teaching staff who have been able to consult one or more resource persons in such a context.

—Jolicoeur and Pagé (2015) [in French]

A caring attitude

Mulholland (2020) suggests that if plagiarism prevention were treated as a learning objective rather than a code of conduct, then institutions, programs and teachers would have to take greater responsibility for developing methodological skills (including basic and advanced research skills—including how to cite sources). They should also ensure that these skills are meaningful to students. This would mean creating more personal learning experiences, valuing the voices of our students in their work, not just the voices of experts.

Still according to Mulholland (2020), to achieve these goals, institutions should:

• minimize the size of student groups
• offer release time to people developing evaluations and pedagogical materials to ease the transition
• limit the insecure employment status of teachers

Prosocial messages

Bruschke and Gartner (2009) studied the impact of prosocial messages on classroom management. A prosocial message is one that shows concern for others, a positive attitude and a willingness to help. As Grant and Hofmann (2011) write, prosocial messages are ideological messages. They enable individuals to understand how their contributions will benefit others.

Smith (2020) describes 2 categories of messages that can create winning conditions for student success when communicated by teachers:

• messages that convey inspiring expectations

“I have high expectations for you to learn and grow, and want to support you in that growth.”

• messages that convey broad regard

“I regard you as a full person, with a range of identities, values, and interests.”

Smith writes that students who receive these messages from their teachers do better. Could these prosocial messages also have an impact in preventing plagiarism and promoting integrity? Personally, I think so!

At the very least, these messages seem to be effective in the workplace, as observed by Hildebrand and Barclay (2022). They concluded that the relationship between anxiety and unethical behaviour can be mitigated by prosocial messages that direct attention to the considerations and needs of others.

In practice: prevention or cure?

Detecting plagiarism in written work is often difficult and time-consuming. A free tool such as Compilatio’s artificial intelligence detector currently does a good job of detecting ChatGPT-generated text, even when it is paraphrased by WordTune or translated with DeepL. But for how long? Above all, it’s important to bear in mind that plagiarism detectors give a probabilistic result; without an administrative guideline, their use is tricky and more of a negotiation with the student than a smoking gun; they can’t be presented as irrefutable proof.

Therefore, I think prevention is the best way to curb the problem.

If we focus specifically on the goal of stopping students from using ChatGPT, some of the strategies I used in Winter 2023 are no longer useful, as the most recent iteration of GPT-4 is more powerful than its predecessors. For example, I used to ask my students to write a text that meets 5 criteria. ChatGPT would not understand complex instructions very well. It might have produced a text that met only 2 of my criteria. However, this is no longer true as GPT-4 can now understand complex instructions flawlessly.

In January 2023, Rudolph et al. (2023) made some short-term recommendations, including:

• asking students to analyze discussions that have taken place in class
• giving oral exams

Some of these ideas may not work for long. In the longer term, Rudolph et al. (2023) suggest:

• evaluating in class
• focusing on evaluations involving presentations, performances, etc. (Rudolph et al. (2023) also suggested web page creation, but GPT-4 can already do that.)
• designing authentic assessments
• exploiting peer-evaluation and peer-teaching (teach-back method)

And, of course… to make students aware of the importance of intellectual integrity.

Indeed, it’s clear that preventing plagiarism means promoting intellectual integrity. In any case, if our students undertake graduate studies, we can’t expect them to hand write their dissertation under face-to-face supervision… It is therefore paramount to act now to make them aware of the importance of intellectual integrity!

What could a plagiarism-proof written evaluation look like?

In light of my research, in my opinion, a plagiarism-proof written evaluation:

• takes place in the classroom using pen and paper, or in a computer lab where access to the Internet or certain software programs can be blocked
• is open-book, or with access to a memory aid
• has a low weighting
• has been preceded by teaching aimed at developing students’ information skills
• has been preceded by actions to prevent contract cheating

My classroom experience

For my part, since returning to the post-pandemic classroom, the vast majority of my evaluations take place in the classroom, on computers without Internet access. Students leave their backpacks at the front of the room and sit where I tell them to sit. Above all, I rely on prosocial messages to promote intellectual integrity. I communicate my expectations to my students and let them know how much I value and respect them. I let them know that I believe they are able to do well, and that it’s worth their while.

What’s more, I let my students choose the subjects for their assignments, so that they are more motivated by the task.

I also give them a lot of feedback, which helps them with their learning. Students hand in their formative assignments on Moodle and I comment on them. The 1^st draft is formative and heavily annotated. Their corrections (and therefore the learning that takes place in the use of the feedback provided) are the subject of a 2^nd draft in an exam situation. This is self-assessed by the student before I give a mark and further feedback.

Since this strategy emphasizes learning rather than performance, I think it helps minimize the risk of plagiarism.

I really feel that self-evaluation makes a difference to the quality of the work. For people who were already highly motivated to begin with, that’s obvious. But even for those who don’t perform as well, the self-assessment grid I provide helps them make sure they haven’t forgotten anything, and that’s very useful, especially when it comes to structure and content.

In Winter 2023, in the program-specific English course (B block), I had my students read a book on ethics (A Practical Guide to Ethics). At the beginning of the session, I asked the students to think of an ethical issue related to their program and that interested them. For example:

• A group of students in Administration took a closer look at harassment in the workplace.
• Future firefighters studied the prevention of cancer as an occupational disease.
• A Civil Engineering Technology student looked at collusion and corruption.
• etc.

I think the fact that ethics was at the heart of the course helped to create a classroom climate that was conducive to integrity.

Throughout the session, students were given tasks related to their topic to prepare them for the final exam. For example, one week I asked them to draw up a (very basic) research plan:

• name an action to be taken to verify their hypothesis or further investigate their question
• describe the sample of people they were going to interview
• detail if they were going to use a tool to help them during their investigation (and describe it, if appropriate)
• etc.

Partly to take advantage of the relationship between evaluation stakes and plagiarism, I decided that only formative work would be done at home. This work could then be used, after feedback, in exam situations.

At the exam, students were given a series of questions to answer. These questions included those they had already been asked among their weekly tasks, as well as a new one: to provide an update on the final progress of their project. The students didn’t know in advance what the exam questions would be, but they couldn’t really be surprised…

Thus, the summative evaluation took place entirely in the classroom, in a controlled context. However, it was on a subject that the students had mastered and had had ample time to prepare. What’s more, since each student had a different question, copying from a peer was impossible.

I also asked students to evaluate themselves before handing in their exam. I believe that this metacognitive exercise had a positive effect on the quality of their work and, by refocusing attention on learning, may have helped limit the urge to plagiarize or cheat.

At the end of the session, the majority of students said they liked the fact that all the writing had been done in class. But a few said that, had they been able to do the work at home, they would have been able to spend more time on it and thus produce a richer, higher-quality assignment. Since these comments came from people whose sincerity I have no reason to doubt (i.e., I don’t think it was the possibility of cheating that made them praise the virtues of out-of-class work), I have to admit that it got me thinking. In other words, there’s no ideal solution…

I also discussed ChatGPT with my students in another course. The students saw the usefulness of generative AIs in everyday life, but they also said they wanted to learn. In this respect, the students understood that using conversational robots could be counterproductive.

A crisis; an opportunity…

We are currently undergoing an abrupt technological shift. This gives us a responsibility to ensure that our teaching continues to foster the intellectual growth of our students. In this sense, we are in a privileged position. We can still preserve our students’ autonomy with regard to technology, so that once they enter the job market, they can use it wisely. So, when they use artificial intelligence, it will be by choice, not by dependence.

After all, preventing plagiarism is also about promoting learning: a person who cheats doesn’t learn. In the words of Tricia Bertram Gallant (quoted in Poitras Pratt and Gladue (2022)), rather than asking “How can we prevent students from cheating?”, we should be asking “How can we ensure that students learn?”.