340: By 7:1 Margin, Students Expect More Cheating in Next Three Years

Plus, a professor rationalizes away student misconduct. Plus, a LinkedIn video you should see.

Issue 340

Subscribe below to join 4,324 (+4) other smart people who get “The Cheat Sheet.” New Issues every Tuesday and Thursday.

If you enjoy “The Cheat Sheet,” please consider joining the 16 amazing people who are chipping in a few bucks via Patreon. Or joining the 46 outstanding citizens who are now paid subscribers. Paid subscriptions start at $8 a month or $80 a year, and corporate or institutional subscriptions are $240 a year. Thank you!

---

Survey: Cheating is Getting Worse, AI is Making Cheating Easier, and More

Academic publisher Wiley issued a new survey report on academic integrity.

The cover says Wiley updated the survey in 2024, though I am not sure when exactly. It covers “850 instructors and 2,067 students” and is well worthy of your consideration.

According to the survey, 45% of students said they used “generative AI tools” in their classes in the past year. Thirty-six percent said they used AI to “write essays and papers” — the most common use. Curiously, “ChatGPT” is listed as its own use in classes, and scored 14%. Not sure how to interpret that.

When students were asked to “briefly explain why you feel negative about the use of Al,” the highest response, at 33% was:

Makes it easier for students to cheat

Of the students who said they did not use AI in their classes in the last year, a plurality (37%) said they were:

concerned their instructor would think they were cheating if they used AI

Worth noting. The fear of being perceived as cheating is, it seems, a top motivator not to use AI.

Then there’s this finding, which deserves a bullet. Teachers and students were asked, “Do you feel more or fewer students will cheat in the next 3 years?” The choices were a five-option menu from “significantly fewer” to “significantly more,” with “about the same” as the middle option.

Care to venture a guess about the responses?

Most students (55%) said more students would cheat in the next three years, with 18% saying “significantly more.” Just seven percent said fewer would. Two percent said “significantly fewer.” Fifty-five to seven. Eighteen to two.

As for instructors, it was largely the same — 55% thought more students would cheat, while just 2% overall thought it would be fewer. For teachers, the “significantly fewer” was zero — none.

None of that is good.

Big credit to Wiley for saying clearly and without hedging:

Cheating is getting worse, especially in online classes, and instructors are more worried students won’t be prepared for the real world as a result.

Cheating is getting worse. Especially in online classes. Just rare to see someone say so.

If you’re able and inclined, pop over and take a look at slide nine in the report. Wiley asked students, “Why is it easier to cheat compared to last year?” Here is the response chart:

Nearly half (47%) say cheating is easier because of increased use of generative AI, 35% say use of ChatGPT. I am not sure what the difference is. The answer is clear either way. Interesting to me that “increase of online classes & exams” is third. And I literally did laugh out loud at “The internet” being a choice, like it’s 1996. And Chegg. Seriously. What’s funny is not that Chegg scored two percent on why it’s easier to cheat, it’s that Chegg was a choice. But Chegg is totally not a cheating company.

Cough.

There were also some open-ended, narrative responses shared in the Wiley survey, such as this one, from a student:

“AI makes it easier to cheat, and especially so when professors encourage you to explore AI in classes. Also, professors use repeated work that allows for cheating.”

Sure, it’s a sample of one. But I find it very compelling that a student has linked ease of cheating directly to professor encouragement to explore AI. It kind of makes sense. Also, the “repeated work” issue is a real problem.

Worth your time as well is slide 10, which asks students why cheating is more difficult now. Here’s the chart:

No surprise, “in-person classes and exams” is the top reason students say it’s harder to cheat — and by a massive gap. It’s harder to cheat when people can actually see you. Imagine that.

I am also not surprised that stricter rules, proctoring, and “detection software” made the list, all at about the same rate — between 16% and 14%.

At some level, this is not complicated. If you supervise and/or secure your assessments, cheating is harder.

Also on slide 10, Wiley asks students whether they agree with statements about AI and cheating. Here are the statements and the agree/disagree:

• I am worried about the consequences of cheating — 65% to 14%

• The increased use of artificial intelligence makes it easier for students to cheat — 69% to 11%

Repeat and underline. Generative AI has made it easier to cheat. Consequences matter.

Later, the survey asks teachers why they think more or fewer students will cheat in the next three years. That too is insightful, though I won’t share all of it for space and time. I will share only that, on the question, to teachers, on why they think fewer students will cheat, the top answer at 58% is:

Al tools and software will make it easier to detect cheating

For all the literal investment some people have made in convincing the public in general, and teachers in particular, that AI detection does not work, or that AI cannot be detected, that’s an eyebrow-raiser of a response.

Further in, Wiley writes:

Almost 60% of instructors felt that AI had no impact on cheating over the past year, but most feel it will have a negative impact in the future.

That’s curious because that “almost 60%” is actually 56%. But fine. Shocking, but fine.

Still, when you look at the teachers who do think AI has had an impact on cheating, 40% of teachers said that impact was negative, four percent said it was positive. Zero said it was “significantly positive.” That’s ten to one in the negative.

When asked about the impact of AI on academic integrity over the next three years, teachers were less ambiguous — 68% said it would be negative, 11% said positive.

To consider further, there’s slide 13 which asks teachers what percentage of their students they think are cheating and compares answers between 2021 and 2023. The difference is considerable. Here is the chart:

I mean, you can see the difference. In 2021, for example, an unbelievable 28% of teachers thought no one was cheating. As of 2023, that percentage is down to four. The percentage of teachers who think at least half of all students are cheating tripled, albeit from two percent to six percent.

This, combined with the other data on teacher’s views of cheating now, versus what they expect in the future, shows teachers are really behind the awareness curve — a conclusion that’s supported elsewhere in the Wiley data. For example, by 54% to 14%, students said cheating this year — 2023, I assume — was worse than previous years. Teachers responded to the same question 26% to 15%.

I also love that Wiley directly asks whether teachers and students think cheating is more likely in online classes, also comparing 2021 answers to 2023. Students went from 59% more likely to 80%, while teachers moved from 77% to 86%. Here, at least, teachers seem ahead of the curve. About which Wiley again, straight says:

Previously, concerns were around online classrooms post-covid, but cheating is still getting worse.

A thousand times yes.

Slides 16 and 17 get into what students and teachers say are the why of cheating. Nothing too surprising there, although it is interesting to see where the two survey groups part company. For example, 43% of students say “workload from instructors” is linked to cheating. Just 9% of teachers think so. Likewise, students think relevancy of course and materials correlates to cheating, teachers are less inclined to think so.

I say again, if you can, go review the whole thing. I’m doing my best to highlight but I am skipping a considerable amount, much of which is potent.

In another section, students are asked agree/disagree on a series of statements related to integrity and misconduct. The two highest scoring agree statements, with the percent that agree and strongly agree are:

• I am worried about the consequences of cheating — 65% agree, 32% strongly

• The increased use of artificial intelligence makes it easier for students to cheat — 69%, 30%

I’m not sure here, or previously, that “consequences” was defined. So, it could be taken as punitive outcomes (e.g., getting caught) or academic outcomes (e.g., falling behind, not learning). Still, important to digest, I think.

I am just not sure how much more evidence we need that AI is making cheating easier and more common. But we keep getting it.

The survey goes on to ask teachers, “What are the biggest challenges/concerns academic misconduct created over the past year?” As before, it compares 2021 answers to those in 2023. I won’t go too deep here but point out only where the challenges/concerns went up between the two rounds — increasing concerns/challenges. The largest jump, up 14% over the two years, was:

It's time consuming to deal with students once they cheat

Second, up 9% was:

I'm concerned students won't be as prepared for the real-world/careers

I share that concern. Both, actually.

Slide 20 is also worth your time is it includes information on what teachers are doing to “mitigate cheating.” Some are borderline useless, such as “Required all students to sign an honor code contract” (18%), while some are likely quite significant, such as “Used cheating detection/prevention software” (16%). Others, such as “Added more open-ended questions to exams/assessments” (35%) make no sense whatsoever.

There are open-ended responses from teachers to this question about mitigation as well. I’m sharing just one:

“I downloaded photos of exam questions from Chegg and other cheating websites and matched them with randomized question/answer pools to identify cheaters. I also contacted such cheating websites to obtain IP addresses of cheaters who submitted questions during an exam. I also reviewed ProctorTrack videos of students cheating on exams, and then reported students suspected of cheating.”

My grade: A.

Skipping a ton more … slide 23 seems to show, quite unbelievably, that 22% of teachers say:

My school does not offer specific resources to support/create a culture of academic integrity

I’m sorry, what’s that now? That cannot possibly be right. As in accurate. It’s definitely not right.

There’s a ton of stuff I had to move past, but this survey is excellent and well worth a personal review. That it exists is pretty amazing. Wiley deserves credit for investing in this topic, asking these questions, and sharing the results.

To me, the takeaway is clear: cheating is up, AI is fueling it, it’s unlikely to get better.

Professor Rationalizes Cheating

Salmaan Khan, Assistant Professor at Toronto Metropolitan University, recently offered his thoughts on academic integrity over at The Conversation.

Most of what he thinks we should know is that public funding cuts to higher education, and the follow-on required commercialization of teaching and credentialing, create the environment in which cheating is common, perhaps even expected.

He’s right, in my view, that the zeitgeist of education being a product and students being customers has made cheating seem more rational to some. But that’s at best, part of the problem. And laying rampant unethical behavior at the feet of political policy is convenient — and recklessly exculpatory of those closest to the issue, namely teachers, administrators, and students. It’s therefore counterproductive.

It’s also difficult for me to take his insights too seriously when the second paragraph in his article is:

GenAI models such as ChatGPT offer students untraceable and economic means of churning out answers and term papers on any given subject.

ChatGPT is absolutely not untraceable. Beyond being untrue, I don’t know what else this is.

In reflecting on the recent survey from academic publisher Wiley we covered above, the professor writes:

Of the students surveyed, a majority noted the role of emerging technologies, such as ChatGPT, in making it easier to cheat than before. When asked why more students may turn toward cheating, almost half responded that because education is so expensive, there is an added pressure to pass or attain certain grades.

Thirty six per cent of students said they are more willing to cheat because it is hard to balance going to school with work or family commitments.

Students did say they are more likely to cheat because “education is so expensive” and “it is hard to balance going to school with work or family.” But let us all be clear that those are rationalizations for informed choices, not necessarily drivers of misconduct. The cost of tuition does not close the door on doing the work of learning.

Our professor goes on:

When larger numbers of students are batched into lecture halls, there are fewer opportunities for active student-teacher engagement, characterized by dialogue, which is a key ingredient in fostering engaged and critical learning. In this context, should we be surprised if students feel disconnected?

In the same Wiley report, students noted they are more likely to resort to cheating if they do not sense the significance of the course material to either their own lives or to the real world.

Rational. Eye. Zation.

I agree that large class sizes and distance between instructor and student, especially in online courses, can facilitate cheating. But that in no way justifies it.

And should we be surprised? Is that a serious question? If it is, it’s the wrong question. The question ought to be, what are we going to do about it? Pro tip: the answer is not blaming “neoliberal” policies and budget cuts when the reality is that no one has more influence over academic misconduct that the classroom teacher.

I also feel a compulsion to point out that, when you ask people why they are doing something they know is wrong, they will pick a reasonable justification. In order to cheat, a self-justified reason is virtually psychologically required. If you give a list that includes blame-shifting such as “tuition is high,” or “my teachers do not engage me,” or “I won’t use this class in my job” — they will pick those answers of absolution every time. It in no way means any of that is true. It means they feel better about being able to make someone else responsible.

The piece more or less concludes with:

The crisis spurred by concerns with student ethics or of the use of genAI to cheat on assigned work must be understood within this larger context, as opposed to being seen as emerging from features specific to genAI.

I mean, sure. Whatever. Just one more person who sneaks in that the crisis we’re dealing with is “concerns with student ethics” instead of student ethics itself, while also giving solid effort to rationalize and explain away the problem as someone else’s fault and, we assume, responsibility.

LinkedIn Post/Video: Watch AI Agents Take an LMS-Based Course

Our friend (and The Cheat Sheet’s 2024 Person of the Year) Joseph Thibault has a recent post on LinkedIn with a video showing just how “easily” an AI operator will take an entire course nested in an LMS.

It’s worth a view and, unfortunately, just the tip of a very big iceberg related to integrity vulnerabilities in LMS — open doors that, so far at least, LMS makers have been reluctant to address.