Forum menu
They claimed they used an online thesaurus.
I was suspicious...
...skeptical, disbelieving, unconvinced, distrustful...
^^I'd penalise them for sure for using a US thesaurus.
The reference to Loftus looks appropriate, except that the student has mistyped it as 'Lotfus'. An AI wouldn't do that, so unless the student is cunning enough to have backfitted a typo, they probably typed it themself. And if they did backfit a typo, they're good at criminology so fair enough.
in their respected fields
'respective' fields surely?
I dont think this is AI written (or if it is, it has then been well re-written. If you are suspicious that it is not the student's work, then it may be that they know an adult who has provided some input and 'tidied up' their work.
Either way, not much you can do.
Have you spoken to other teachers? Particularly any that have taught them English over a period of years?
I have her work from last year in a filing cabinet. She got an A on something similar last year. I wasn't suspicious about her writing then. I'll be checking it now.
Stuff written in class is nothing like that.
The reference to Loftus looks appropriate, except that the student has mistyped it as ‘Lotfus’. An
It is. It looks to me like she has added a correct reference where the Ai hasn't.
tidied up’ their work.
In 17ish years of teaching a level I've never had a piece of work this good!
It's not a students tidied up. Re written perhaps by an extremely competent writer.
Stuff written in class is nothing like that.
That's an indicator of some sort of outside help. Good writers are good writers wherever they do it.
Not conclsive though as some pupils do perform better given more time to write.
It’s unlikey to be plagiarised given the errors a peer reviewer would pick up on first read
First trick in the book, blindly copy then edit in some goofs to make it more plausible.
I could answer this myself, but have handed it over to ChatGPT:
Looking at the student's work, there are several indicators that suggest it may not have been generated by GPT or a similar AI model. These include:
1. Misspelling: The name "lotfus et al" appears to be a misspelling of "Loftus et al," a well-known reference in psychology. GPT is typically good at spelling and would likely not make this mistake.
2. Punctuation: The use of punctuation is inconsistent. For instance, the comma after "et al" is misplaced and there's a missing apostrophe in "witness’ memory". GPT generally uses punctuation correctly and consistently.
3. Capitalisation: The beginning of the sentence starts with a lowercase "p" in "psychological". GPT would typically start a sentence with a capital letter.
4. Sentence structure: The sentence structure is a bit awkward in places, such as "In complex technical cases, the outcome frequently depends on the testimony provided by an expert, be it a medical specialist or a forensic scientist." GPT tends to produce more fluid and coherent sentences.
5. Terminology: The phrase "respected fields" seems like a misuse of the word "respected" where "respective" would be more appropriate. GPT is generally good at using the correct terminology.
Now, I'll write a paragraph about the reliability of eye-witness evidence so you can compare the two:
"Eye-witness testimony in criminal cases holds a significant weight in courtrooms, yet its reliability has been questioned by numerous psychological studies, including the seminal work of Elizabeth Loftus. Factors such as the time elapsed since the event, interactions with others, and the interrogation methods used can all influence the accuracy of a witness's recollection. Particularly in cases where a significant amount of time has passed or where the witness's attention was drawn to specific details, such as the presence of a weapon, their testimony may be less reliable. In complex cases that require specialized knowledge, expert witnesses, such as medical professionals or forensic scientists, are often relied upon. These experts are expected to possess a level of expertise in their respective fields that surpasses that of the average juror or legal professional."
If you're really sceptical then my guess is that this was written by a family member. Might be he's just good at writing essays, though.
If you’re really sceptical then my guess is that this was written by a family member. Might be he’s just good at writing essays, though.
I wasn't taught how to really write an essay until AS levels. We'd had the generic "plan, introduction, body, conclusions" type stuff, but not how to write a paragraph in 6/7 sentences and a sketch that would reliably get you 7/14/21/28/35 marks on each question (the exam board criteria was a maximum of 7 marks per point/answer, so a 35 mark essay was implicitly telling you you needed to give 5 correct answers, 7 marks only required one. Went from C's, B's and running out of time to not dropping a single mark through the entire Geography A-level and carried that style on for the next 20 years if I'm writing long technical assessments, studies etc.
Statement
Explain
eXpand
Example
Diagram
sum UP
Confronted student
Me- your works has been written with ai.
S- it hasn't
M-I've checked it with AI checkers
S- those checkers flag up loads of work as ai
Interesting they knew this correct fact
M- if you hand that in you will fail for plagiarism
S- it's not plagiarised.
S- my work from last year is the same
M- yes I'm off to check that now.
So that went well. I was hoping they would back down and re write it in their own words- which Is fine.
My head of faculty is fine with them being failed if they continue to hand in work that's plagiarised.
Next up is the head of 6th form. They were extremely clear that I need to repeat the plagiarism policy to all students and make sure the particular student is very clear that they will be failed if they hand in plagiarised work.
I checked 6 other bits of work through AI checkers. One came up as 1% written by ai, the other 5 came back as 100% human.
The student in questions work came back as 51% chance of being written by ai.
Their work from last year is nothing like this. It's been in a filing cabinet since last christmas.
Terrible spelling-its Typed id have expected a spell checker.
Terrible grammar for the most part.
But, bits of good grammar with odd things missing. An excellent sentence missing a comma, that changes the sentence and takes it from excellent to terrible.
So
So that went well. I was hoping they would back down and re write it in their own words- which Is fine.
I'm not sure it was the most constructive challenge you could have come up with. In your work in criminology how often does a genuinely guilty party who initially denied it just put their hands up and say "oops". Do they do this to their original accuser, a short while after being caught - or to someone less confrontational at a later date? In your work as a teacher how often do kids in trouble deny is no matter what? Now if she didn't use AI and she hasn't plagiarised it, she is quite right to be digging her heels in. I think you've possibly come OTT here.
My head of faculty is fine with them being failed if they continue to hand in work that’s plagiarised.
Of course, thats what we would expect - but how will he back you up if you can't show who's work was copied? The premise for a claim of plagiarism is surely - this isn't your work because it very closely resembles this work which existed before yours. AI makes that near impossible, but is asking AI for help plagiarism?
What right of appeal does the student have if they don't agree with your opinion it is plagiarised? What support to they get to explain themselves? How independent is the "tribunal" to which they can challenge this? There's a lovely "criminology" case study here for the student on what it might be like to be wrongly accused of a crime, or what it might be like to be accused but backed into a corner where denial is easier than honesty.
Next up is the head of 6th form. They were extremely clear that I need to repeat the plagiarism policy to all students and make sure the particular student is very clear that they will be failed if they hand in plagiarised work.
Does the plagiarism policy reflect the realities of 2023 and the existence of AI? What if the student writes it and AI "polishes" it? What if AI writes it and the student modifies it?
I checked 6 other bits of work through AI checkers. One came up as 1% written by ai, the other 5 came back as 100% human.
Scientifically validated tools of course?
The student in questions work came back as 51% chance of being written by ai.
So a toss of coin... I'd be careful, you could end up looking stupid here.
Their work from last year is nothing like this. It’s been in a filing cabinet since last christmas.
Do none of your other students work improve during the course of a year?
Terrible spelling-its Typed id have expected a spell checker.<br />Terrible grammar for the most part.<br />But, bits of good grammar with odd things missing. An excellent sentence missing a comma, that changes the sentence and takes it from excellent to terrible.
So are you certain that AI did the "work" not just the "fixing" - "ChatGPT make my essay sound better" or even just extensive use of Grammarly or similar tools which are presumably not considered cheating in this context? Whilst my gut feel, like yours, is that was not written by an 18 year old, especially one who has never previously shown linguistic promise I think your approach is probably not the most constructive to achieve what presumably is the ultimate aim - to get her to write lots more pieces which are both factually correct and as well written as this? A less "I've already decided you must have cheated" approach might have been to sit down and discuss how good it is, and why it might be questioned if it was theirs and suggest that previous drafts, notes, and an explanation of any external proof readers or automated tools might be useful to support the authenticity. If they use Microsoft 365 or Google Docs for creating the document it will have a history in the document.
I checked 6 other bits of work through AI checkers. One came up as 1% written by ai, the other 5 came back as 100% human.
You’re looking for evidence that supports your hypothesis. You should really be looking for something that contradicts it. Given that ChatGPT itself is sceptical that it wrote it, I’d suggest that you need further proof.
The easiest solution, surely, is to set your class a single question that they answer in front of you.
FWIW if I was your student and you accused me of plagiarism, and could not support your claim other than suspicion, I would fight it to the bitter end.
Terrible spelling-its Typed id have expected a spell checker.
Terrible grammar for the most part.
But, bits of good grammar with odd things missing. An excellent sentence missing a comma, that changes the sentence and takes it from excellent to terrible.
^^So are you writing that badly deliberately to prove you're human?
That was clearly not the voice of a teenager. Unless they are exceptionally gifted, in which case you would already be aware that they are gifted. Has a style of confidence one would expect at postgrad level or after years of familiarity with the subject matter. Fail it, set them another assignment followed up by a verbal questioning of the subject matter and therein will be your answer.
Genuine question .... why do you need to prove it ?
I'd imagine you guys have more than enough on your plate without spending time worrying about the likes of this?
He's only cheating himself, right ??
He’s only cheating himself, right ??
Unless it's assessed coursework counting towards a final grade.
This is a lot like the pro-cycling/drugs thing. We want to believe in progress but too much draws suspicion.
My 12 year old would be able to use words in a sentence that he couldn’t ably explain in meaning. He knows the context of their use, but might sometimes struggle to articulate the meaning in descriptive non synonymous words.
He’s set 1 for English language and literature.
@Flaperon
I know that the AI checkers can give a false positive. So I thought Id check what it came up with on other pieces of students work.
The point of the thread is to see if I can prove my suspicion that it is not theirs.
Can you suggest any other ways to check?
My experience of students writing says that this writing is far too good for any 17 year old I've ever taught. Sample size about 30 a year for 17 years.
Unless it’s assessed coursework counting towards a final grade.
It is work that they can use towards their final grade. Hence spending time to check if it is theirs.
As someone said earlier- if I can't show use of ai, what is stopping the whole course using it?
If I don't stop it at before it goes into the exam then when it gets moderated, there is a strong possibility all students on the course will have their grade affected. If our making of the work is not accurate, then the moderator might/ will/ should change the grades.
^^So are you writing that badly deliberately to prove you’re human
Unfortunately- that is the unmoderated quality of my writing on a phone. I'm reasonably sure I couldn't produce writing of the quality the student handed in. I was always just a wee bit above the test scores to diagnose dyslexia when I was at school.
You can't prove it, you can only let the student know that you know, and that you will be watching their work closely in future.
She seems to be pretty confident and brazen even when challenged though. What's the betting that the parents come in and start moaning as well?
Can you suggest any other ways to check?
skimming through this thread you’ve been offered a few ways to check that make sense to me - basically a verbal review around the subject - do they actually understand it and can talk around it once they ate no longer able to rely on their polished, prepared text. This is basically what happens at a job interview for technical subjects. It very quickly shows who’s got a CV that reflects reality and who’s just making it up.
It is work that they can use towards their final grade. Hence spending time to check if it is theirs.
As someone said earlier- if I can’t show use of ai, what is stopping the whole course using it?
After years of moving away from exams as a method of assessment I wonder if we'll see a return to them? It's an easy way of dealing with the AI generated coursework issue.
Isn't it a little ironic to crowdsource how to do your job on here to detect student use of external tools? 🙂
Can you suggest any other ways to check?
There have been a number of occasions where I have managed to delete or somehow loose a report or piece I've been writing. Once I'd got over my utter stupidity and annoyance, I simply sat down and wrote it again. No drama, just redid it, and often it was imo better than the lost version.
That's the test for me. Can they repeat it?
Poly the peer reviewer edukating Edukator the peer reviewer - who always commented when the grammer made things ambiguous or difficult to follow. 🙂
WTF happened to a presumption of innocence, tall_martin? You're teaching a legal course yet launch into a character assassination with, by you own admission and analysis, no proof. Everything you've shown us so far says it's original text even if the ideas expressed are old hat. There is nothing new or original in year 13, it's just rehashing what students have read; a synthesis, an essay.
Exam time shows what students are really capable of, some are very impressive with just a pen, paper and what's in their head.
Yeah but no. Not if , in the modern day and age, the paper is a percentage of the final Grade. As the snow flake of today get the a chance to shine outside of the exam hall by handing in work that they might,or might not have done themselves
Nothing new in arguments against continuous assessment that isn't under exam conditions. Kids have parents, brothers, sisters, friends, private tutoring... . Every teacher knows that when you're marking homework you're marking the socio-economic background of the child, so you don't, you just correct it and comment on it, but don't use it towards a final mark.
We were amused when junior gave us the rundown of the stereotypes that made it into Science Po, Paris. The rich, the privately educated, the quota from the ZEPs, and teachers' kids. 🙂
If you copy/paste text into chatGPT it should be able to tell you if it was generated by chatGPT in the first place. Bing on the other hand keeps no trace of what it has generated - so the smart way to cheat is to generate using chatGPT, then get Bing to rephrase the text (and if, after that, you put the text back into chatGPT it will not recognise that it created it in the first place). This is something we’re grappling with at the moment…
@educator
WTF happened to a presumption of innocence, tall_martin?
In 17ish years of teaching a level I’ve never had a piece of work this good!
Which is suspicious.
From a student who has previously handed in work that was correct, but poorly written.
Who when questioned couldn't explain what some of the words meant and couldn't explain what some of the phrases meant.
I've had plenty of people use massive words they couldn't quite explain. Mostly it was copied from the internet. Hence the verbal questions.
There is nothing new or original in year 13, it’s just rehashing what students have read; a synthesis, an essay
I'm sure this student could verbally answer almost all of the concepts in the writing. It's the sudden increase in quality of writing from poor for a year 13 to better than a textbook that's causing my suspicions.
I've been reading this whole thread thinking the student was a 13 year old, only just realising after the last post that it's a year 13 A level student! I did think criminology was an odd subject at that age 😀
I need time to recalibrate
Hang on a minute. OP teaches criminology and cannot figure this one out ?
Title: Unveiling Plagiarism and AI Utilization in Criminology Essays
Introduction:
The academic landscape is constantly evolving, and with the rise of technology, the challenges associated with maintaining academic integrity have become more complex. In the realm of criminology essays, the issues of plagiarism and the use of artificial intelligence (AI) present growing concerns. This essay aims to explore the methods of detecting plagiarism and AI use in students' criminology essays and the implications of such practices.
Body:
1. **Plagiarism Detection:**
Plagiarism, the act of presenting someone else's work as one's own, can be detected through various means. Educational institutions commonly employ plagiarism detection tools like Turnitin and Grammarly. These tools compare a student's work against a vast database of academic and online content, highlighting potential instances of plagiarism. Additionally, manual examination by educators can uncover inconsistencies in writing style, shifts in language proficiency, or abrupt changes in the level of sophistication in the essay.
2. **AI Assistance:**
The integration of artificial intelligence in academic settings poses a dual challenge. Students may resort to AI tools for generating content, leading to essays that exhibit a level of coherence and complexity beyond their typical capabilities. Identifying such cases requires educators to be vigilant in recognizing anomalies in writing styles or sudden shifts in the depth of analysis.
3. **Cross-Referencing Sources:**
Another effective method is cross-referencing the essay's content with reputable sources. Plagiarized material often stems from existing publications, and a thorough examination can reveal instances where the student fails to attribute ideas or quotes correctly. By utilizing academic databases and scrutinizing cited sources, educators can pinpoint discrepancies and assess the originality of the work.
4. **Educational Initiatives:**
To combat plagiarism and AI use, educational institutions must prioritize awareness and prevention. Educating students about the consequences of academic dishonesty and fostering a culture of integrity can act as deterrents. Clear guidelines on proper citation and source attribution, along with regular reminders about the importance of independent research, contribute to a more ethical academic environment.
Conclusion:
In the dynamic landscape of criminology studies, the specters of plagiarism and AI use cast shadows on the authenticity of students' work. Vigilant educators armed with advanced plagiarism detection tools, a keen eye for inconsistencies, and a commitment to fostering academic integrity play a crucial role in maintaining the sanctity of education. By addressing these challenges through proactive educational initiatives, institutions can empower students to engage in honest and independent scholarly pursuits, ensuring the credibility and value of their academic achievements.
Hang on a minute. OP teaches criminology and cannot figure this one out ?
Maybe this entire thread is a criminology experiment set up to be analysed by the A level class. I wonder how we're doing...
😃 that's a great idea!
Maybe I'll embed it into the course for next year as a way of explaining the perils of plagiarism
Hang on a minute. OP teaches criminology and cannot figure this one out ?
No crimes have been committed, just a bit of moral deviance
Speaking with my wide this morning (prof at Man U) she thinks its a piece from an essay mill
@nickc
Full Member
Speaking with my wide this morning (prof at Man U) she thinks its a piece from an essay mill
I wouldn't have thought it would be worth writing stock essays for our specification.
I would want quite a bit of cash to write the bit of work that was handed in.
Maybe my student has bought the work and not realised whoever sold it used ai to write it
Perhaps that's why they are so adamant it's not been written by ai.
Previously I have had some plagiarism from a flashcard site that you needed to sign into. That student admitted to plagiarism.
Perhaps that’s why they are so adamant it’s not been written by ai.
Kids can get quite literal about this stuff. I’ve had a number of younger pupils being quite indignant about me accusing them of 'copy and paste' because they hadn't actually copied and pasted. They had manually typed the text word for word into their own piece!
All part of the learning process.
What are his previous writings like ?
Can you copy and paste another of his more recent assignments for us to critique?
Then we can do a compare and contrast with the two pieces of work side by side.
I know people develop and learn skills, but not in a few months though.
The original handed in work reads to me like the writing of a teacher or lecturer though , not a kid.
Ai wouldn't make this error: witness’
Maybe this entire thread is a criminology experiment set up to be analysed by the A level class. I wonder how we’re doing…
Or get your class to actually use ChatGPT to write an essay, but their job is to mark it themselves and identify the problems / errors / inconsistencies...
@whatgoesup deserves a shout out but the American spelling is a dead give away. A-
Someone pointed out it could have just been processed by Grammarly or similar, anyone got any thoughts on that?
Would be a good fit, otherwise bright enough person that can't write for shit seems to be the target market.
Speaking with my wide this morning
Well she won't be speaking to you for a while now.
I know that the AI checkers can give a false positive. So I thought Id check what it came up with on other pieces of students work.
6 pieces is hardly scientific though - you presumably had some unconscious bias in their selection, because they had not already raised suspicion. Nor does it seem like you are treating this student fairly - it comes across as you have decided she is guilty and will now find the evidence to prove it. Some involvement of outside help does seem likely, whether AI, a parent etc. I’m genuinely interested though at what point in the “running a spell checker in word” to “parent proof reading and suggesting revising unclear sections” to “tutor coaching around the specific question” to “AI drafting the words based on your bullet points” cross the line? There’s a really interesting ethical point about when as a teacher do you basically accuse a student of lying to you potentially putting them off you and the subject, v’s encourage students to put in some extra effort (like spell check, grammarly, asking a parent etc) and risk being accused of being too good!
I think there are some genuinely interesting questions asked in this thread and if your time allowed some really interesting experiments too, which would probably teach your kids more about human behaviour, the AI issues the world face etc than any plagiarism policy they haven’t read.
I wonder if you asked everyone in the class to USE AI to answer a (carefully worded) question if you could still spot the students who understand the course, because they would give the AI better input to seed the response?