Insider: ChatGPT is "Mainly a Tool for Cheating"
Plus, a TV segment from Utah you should watch. Plus, a note from The Netherlands. Plus, a piece in Psychology Today.
Issue 248
To join the 3,664 smart people who subscribe to “The Cheat Sheet,” enter your e-mail address below. New Issues every Tuesday and Thursday. It’s free:
If you enjoy “The Cheat Sheet,” please consider joining the 14 amazing people who are chipping in a few bucks via Patreon. Or joining the 24 outstanding citizens who are now paid subscribers. Thank you!
Insider: ChatGPT is “Mainly a Tool for Cheating on Homework”
Last month, Insider ran a piece with the headline:
The summer is over, schools are back, and the data is in: ChatGPT is mainly a tool for cheating on homework.
In the article, Insider notes that traffic or usage of ChatGPT fell suddenly, which, it correctly notes, is odd since:
ChatGPT was supposed to be the fastest-growing tech product in history
Insider writes:
One hypothesis stood out: Millions of students went on summer break, so they didn't need ChatGPT to cheat — er, I mean research.
Although I find it neither conclusive nor compelling, it is noteworthy. Especially considering that, as Insider also notes, traffic to ChatGPT regained its upward trajectory once schools returned in the fall. And, that:
Back when summer began, online search interest in [popular video game] Minecraft jumped, while ChatGPT interest declined. Now school is back, the trends have reversed.
Hmm. Not something I would want to try to make a legal case on, but interesting. And the charts in the story are worth checking.
Since ChatGPT showed up, I’ve asserted that it’s a cheating engine. That a major portion of misconduct taking place over the past year has been with or through generative AI. But it had not occurred to me that skirting academic work was actually also a big part of ChatGPT’s customer base.
Seen from the ChatGPT side, Insider writes:
However, if usage is only recovering because students are back, that may be a bad sign because it suggests there's a limited range of use cases for ChatGPT and other AI-powered chatbots.
Mark Shmulik, a top internet analyst at Bernstein, made this point at the start of the summer, when usage fell.
"If it's school kids, that's a real yellow-red flag on the size of the prize," he told me back in July. "This idea that if the ChatGPT drop-off is due to students on summer break, that implies a narrower audience and fewer use cases."
In other words, if a big part of ChatGPT growth is driven by cheating students, this means the technology, or at least the chatbot format, may not be the dominant computing platform of the future.
Again, interesting.
Because I assume that ChatGPT knows more than I do about who is using its system and why — and more than a writer at Insider does, for that matter — even if it’s partly true that ChatGPT’s overall traffic can be dented by students not cheating — it would explain why the company has been so hostile to academic integrity policies and interventions (see Issue 241).
Insider gets a thump for repeating OpenAI’s nonsense that AI detectors don’t work (see Issue 241 again). On that, Insider also writes:
In fact, these tools sometimes suggest that human-written content was generated by AI, including Shakespeare and The Declaration of Independence.
I’m including this canard as evidence of how so many people are thinking so little about this. Of course, AI detectors flag them. They are some of the most repeated, most quoted English writing — at the very top of predictable word choices. And making predictable word choices is what AI does, and what AI detectors detect.
Try this - what word is most likely to come next: “We hold these truths to be [what?]” Or here: “that all men are created [what?]” Or, “Now is the winter of our [what?]” The next word is overwhelmingly predictable, so it’s no wonder AI detectors flag it. Flagging highly predictable word choices is what they do.
If this is the evidence that AI detectors don’t work, you’ll need better evidence. If anything, it proves they do work.
By the way, back in that exceedingly popular Issue 241, I wrote:
[There are a] litany of academic studies and press stunts that show uniformly that AI text-classifiers work. Unlike the one from OpenAI, the good ones work quite well. Not a single study that I know of shows anything to the contrary. If you know of a study that shows they flatly do not work at all, please send it.
Unsurprisingly, I have not gotten any. Not one.
Sorry, caught myself on a tangent there. The Insider piece is interesting and the underlying idea — that academic use/misuse may be a significant part of ChatGPT’s user base — is quite intriguing.
Utah TV Station Reports on Cheating
KSL, a TV station in Salt Lake City, recently reported on cheating. The 3:35 segment is worth watching.
The station’s investigative reporter was called regarding a flier that a viewer received in the mail offering cheating services, saying:
We do homework for college students
I’ll try to make three points very quickly. Well, I’ll make them. I will try to do so quickly.
First, this does not surprise me at all.
In what I hope is one of the core messages of “The Cheat Sheet,” cheating providers are not hiding. This is not a dark art anymore, with illicit documents and services traded in dark alleys. Cheating companies advertise, they hire influencers, sponsor athletes, host and sponsor conferences, give scholarships, and pay teachers. They have investors. They get fawning press coverage. That some random company is sending ads through the mail does not surprise me one inch.
Second, I am stunned that a cheating company is sending ads through the mail.
That’s because using the mail service has specific legal consequences. Unfortunately, academic cheating and selling cheating are not illegal in Utah, where the TV story is and where we know a flier went. The station’s investigative reporter says the cheating company is based in Georgia — where cheating and selling cheating are also not illegal.
But if a flier landed in Utah, I’d wager a few donuts that one or two went to Colorado or Florida or California or Pennsylvania or another of the 17 states where cheating is against the law.
And by mail. Yikes.
If only someone in any of those states cared enough to — do anything. If you’re holding your breath for that, call a doctor first.
Finally, third, I am baffled that an official at Utah Valley University, who was interviewed in the TV segment, would say:
here you have a company that’s encouraging [students to cheat], I was shocked actually.
That a school official can be “shocked” by an ad for a cheating company is simply unbelievable. It’s little wonder that the likes of Chegg, Course Hero and Quizlet can operate so brazenly, so utterly uncontested - entirely unnoticed by education leaders. And everyone else, for that matter.
News About the Dutch Case Finding Proctoring Software Did Not Discriminate
As covered in the last Issue, a quasi-judicial panel in The Netherlands found that remote proctoring provider Proctorio did not discriminate against a student based on her race or skin color.
A few news outlets in The Netherlands picked up the story. Though, with no surprise at all, I’ve yet to see any US or primarily English-language coverage of it, even though several covered it when the case was filed.
I’m sharing the news link above because it has this, attributed to the student who filed the discrimination case:
she had expected a different ruling and that she is disappointed the council did not make a statement about possible discrimination via anti-cheating software in general.
So, even though the Council said there was no discrimination in her case, she still wanted it to “make a statement” about “possible” discrimination by similar software “in general.”
Oh. I see.
Opinion in Psychology Today on Cheating Disconnect
About three weeks ago, Joseph E. Davis Ph.D. pecked out an opinion piece for Psychology Today about a “disconnect” seen in academic cheating.
The first “key point” is the skeleton of the story:
Cheating is rampant, yet young people consistently affirm honesty and the belief that cheating is wrong.
True. And true.
Davis writes:
In a 2010 study by the Josephson Institute, for example, 59 percent of the 43,000 high school students admitted to cheating on a test in the past year. According to a 2012 white paper, Cheat or Be Cheated? prepared by Challenge Success, 80 percent admitted to copying another student’s homework. The other studies summarized in the paper found self-reports of past-year cheating by high school students in the 70 percent to 80 percent range and higher.
At colleges, the situation is only marginally better. Studies consistently put the level of self-reported cheating among undergraduates between 50 percent and 70 percent depending in part on what behaviors are included.
The sad fact is that cheating is widespread.
Because you’re reading this, you likely already have an awareness of these levels of misconduct. I still repeat them, and the truckload of similar stats, every chance I get.
What’s most enlightening about the article however is not about cheating or that students overwhelmingly say it’s wrong. Or the “mixed message” that we tell students — be good and don’t cheat, but do what it takes to get ahead. This, I found most interesting:
In noting the mixed messages, my point is not to offer another excuse for bad behavior. But some of the messages just don’t mix, placing young people in a difficult bind. Answering the expectations placed on them can be at odds with being an honest person. In the trade-off, cheating takes on a certain logic.
The proposed remedies to academic dishonesty typically focus on parents and schools. One commonly recommended strategy is to do more to promote student integrity. That seems obvious. Yet, as we saw, students already believe in honesty and the wrongness of (most) cheating. It’s not clear how more teaching on that point would make much of a difference.
I do not think that the author is an expert in teaching or learning. But I think his point is worth considering and I tend to agree with it. Though I would add — alone. I don’t see more teaching on integrity as a fix if that’s all that is being done.
Anyway, the piece is worth reviewing.