Summary

A college professor failed over half his graduating class for alleged use of ChatGPT on assignments, sparking controversy and highlighting challenges in detecting AI-generated content.

Abstract

A college professor accused twenty students of using ChatGPT to complete their assignments, failing them shortly after their graduation ceremony. The professor offered a chance to retake a paper-based test, threatening those who failed again with academic dishonesty proceedings. The incident gained widespread media attention, with outlets like The Washington Post and Rolling Stone covering the story. It was revealed that ChatGPT falsely claimed authorship over various texts, including pre-existing literature and the professor's own dissertation, leading to an update in its detection capabilities. Despite this, the AI's inconsistency in recognizing its own generated text raised concerns about the reliability of AI detection tools. The professor faced harassment and potential job consequences for his actions and the language he used with students. The college is working to resolve the situation, ensuring no student is denied their diploma. The incident has sparked a broader discussion on the effectiveness of essay assignments and whether they are the best measure of a student's understanding, particularly in fields like animal science.

Opinions

The professor's method of addressing suspected academic dishonesty is seen as harsh and potentially flawed, given the limitations of ChatGPT's detection capabilities.
There is a belief that the incident could have been handled more effectively, with better communication and without public shaming.
Some argue that the focus should shift from punishing cheating to reevaluating the relevance of traditional essay assignments, especially in non-writing-intensive fields.
The incident underscores the need for educational institutions to adapt to new technologies like AI, rather than simply penalizing students for using them.
The media coverage and public reaction suggest a broader societal concern about the balance between technological advancement and academic integrity.
The professor's actions, including his use of profanity, have been widely criticized, leading to sympathy for the students affected and calls for better training in using AI tools ethically.

A College Professor Just Failed Over Half His Grads Because of ChatGPT

Two students confessed to using ChatGPT on assignments. But maybe we’re missing the boat on the real problem here.

You couldn’t make this up if you tried.

It was the day after graduation. Twenty college students had just walked across the stage in cap and gown to proudly receive their diplomas.

The next morning, they got an ominous email from their professor.

The email said he’d run the last three of their assignments through “ChatGTP” not just once, but twice. If it claimed their work, they got an 0.

They would have one more chance to pass. All they had to do was write a paper, due Friday.

He gave the details of the assignment and said he would run their paper through “ChatGTP” again, and if there was any inkling of AI writing their paper, they would fail.

If that happened, he said, they would be brought up before the office of academic dishonesty and their ability to attend the college would come into question.

Um. Sir?

It’s ChatGPT, not ChatGTP.

PT. Not TP. TP is the stuff people hoarded in the early days of Covid.

One student replied immediately to say he did not use ChatGPT and can prove it by the timestamps on the Google doc he used to write his paper.

The teacher replied to say “I don’t grade AI bulls**t.”

Another student started to cry and called her fiance. She didn’t really even know what ChatGPT was or how it worked. She’d never used it. But her diploma was being withheld because of the teacher’s accusation.

Her fiance posted the entire mess on Reddit last week.

Didn’t take long to hit the media. First it hit The Washington Post and Rolling Stone. PC Magazine wasn’t far behind.

ChatGPT has already corrected the problem that started the whole mess…?

When Forbes picked up the story, the writer did his own test. He pasted in text from a report that pre-dated the existence of ChatGPT.

Much like the professor’s experience, ChatGPT took credit.

The Reddit thread is full of people doing pasting stuff into ChatGPT.

It took credit for everything.

Apparently, it even took credit for text from Crime and Punishment, written by Dostoyevsky and published in 1866.

Ironically, someone found the professor’s dissertation from years ago and pasted it into ChatGPT. Yup. It took credit for that, too.

ChatGPT seems to have addressed the problem.

This morning, I put the exact same prompt into ChatGPT. It now says the statement was not generated by ChatGPT.

The problem isn’t solved… not even close.

This morning, I asked ChatCPT to write a summary of Barack Obama’s address to the National Academy of Sciences.

Then I closed the ChatGPT session and the browser tab.

ChatGPT has the ability to “remember” within a session. So I can have a conversation (of sorts) and it will remember what it said. Within the session. So I ended the session and closed the tab.

Then I opened a new tab and a new session.

I asked if it wrote that text.

It said no.

So now ChatGPT is not taking credit for everything.

But it’s not taking credit for the words it generated, either. I fed it the exact same text it wrote for me, not five minutes later. It didn’t know it wrote it.

AI detection is hard.

That’s why 5/6 programs got it wrong when I tested AI detection last week.

The professor is now facing extreme harassment

Early reports named the teacher and the college. Within 48 hours, news outlets were electing not to use his name. Because of harassment.

The instructor did not respond to a request for comment. (We’ve withheld his name, since the instructor is apparently now facing harassment.) source: PC Magazine

The college, however, still has his full profile on their site, complete with his photo, email address and cellphone number.

He’s apparently up the creek with the college now. Not for using ChatGPT wrongly, but for the foul language he used to students.

Some people think he’s going to get fired.

If he does, I’m sure some people will crow about it on Reddit.

All the college is saying about the entire incident is that the professor is working with students on a one-to-one basis to re-evaluate their papers and determine authorship and their final grade. Plus, they said no one is being denied their diploma.

To make everything worse, two students came forward and confessed that they had used ChatGPT on assignments. Two out of twenty. 10% of the class. It’s a pretty low number, really.

Apparently, 68% of undergrads and 43% of grad students cheat.

According to research and surveys over the past 12 years by retired Rutgers University professor Donald McCabe and the International Center for Academic Integrity, 68 percent of undergraduates and 43 percent of graduate students admit to cheating on tests or written assignments. (source)

Maybe we’re missing the boat on the real problem?

In the original Reddit thread, a retired teacher offered a thought. She said the same thing happened when calculators were invented. Schools had to learn to work “with” new technology, not fight it.

I’m just not sure that’s the problem.

ChatGPT has only been around for a few months, but according to the study by the retired Rutger’s professor, students have been cheating on their assignments for over 12 years. Probably longer than that.

Students aren’t cheating “because of” ChatGPT. They’ve been cheating on assignments all along. ChatGPT is just a new way to cheat.

Maybe the assignments are the problem?

For example, the class affected by the Chat CPT fiasco was an animal science class. These are the kids that will become vet-techs, farm managers or animal health inspectors. Maybe some will move on to higher studies in the field and become veterinarians or work in animal science research.

Do we really need animal science students to have mad essay writing skills?

Because that’s what the longitudinal studies show. They aren’t cheating on tests. They are cheating on their essay writing assignments.

Anyone who has been to school on any level beyond fifth grade knows that what one teacher likes is not necessarily what another teacher likes. Teacher A loves the way you write. Teacher B hates it.

Strangely, this case kind of highlights that problem.

Because 15 students out of 20 were accused of using ChatGPT to write their essays. But we know know ChatGPT was “claiming” anything fed into it, including Crime and Punishment and the professor’s own paper.

So why were 5 of the 20 students “not” targeted?

Are those the kids that live in the teacher’s favor? Are those the students whose work the teacher favors, so he didn’t run them through ChatGPT? I don’t know the answer to that, but it does make me wonder.

Maybe it’s time we asked whether essays are the most efficient way to test whether or not students have absorbed the knowledge they were taught in class? Like, why are we so fixated on essays?

If a student is going to become a teacher, sure.

But otherwise?

Because seriously, if I have a sick animal whether it’s a cow or a dog, I don’t care if the vet-tech can write a wicked essay. I care if they know their stuff.

Maybe that’s worth thinking about.