School Did Nothing Wrong When It Punished Student For Using AI, Court Rules 68
An anonymous reader quotes a report from Ars Technica: A federal court yesterday ruled against parents who sued a Massachusetts school district for punishing their son who used an artificial intelligence tool to complete an assignment. Dale and Jennifer Harris sued Hingham High School officials and the School Committee and sought a preliminary injunction requiring the school to change their son's grade and expunge the incident from his disciplinary record before he needs to submit college applications. The parents argued that there was no rule against using AI in the student handbook, but school officials said the student violated multiple policies.
The Harris' motion for an injunction was rejected in an order (PDF) issued yesterday from US District Court for the District of Massachusetts. US Magistrate Judge Paul Levenson found that school officials "have the better of the argument on both the facts and the law."
"On the facts, there is nothing in the preliminary factual record to suggest that HHS officials were hasty in concluding that RNH [the Harris' son, referred to by his initials] had cheated," Levenson wrote. "Nor were the consequences Defendants imposed so heavy-handed as to exceed Defendants' considerable discretion in such matters." "On the evidence currently before the Court, I detect no wrongdoing by Defendants," Levenson also wrote. "The manner in which RNH used Grammarly -- wholesale copying and pasting of language directly into the draft script that he submitted -- powerfully supports Defendants' conclusion that RNH knew that he was using AI in an impermissible fashion," Levenson wrote. While "the emergence of generative AI may present some nuanced challenges for educators, the issue here is not particularly nuanced, as there is no discernible pedagogical purpose in prompting Grammarly (or any other AI tool) to generate a script, regurgitating the output without citation, and claiming it as one's own work," the order said.
Levenson concluded with a quote from a 1988 Supreme Court ruling that said the education of youth "is primarily the responsibility of parents, teachers, and state and local school officials, and not of federal judges." According to Levenson, "This case well illustrates the good sense in that division of labor. The public interest here weighs in favor of Defendants."
The Harris' motion for an injunction was rejected in an order (PDF) issued yesterday from US District Court for the District of Massachusetts. US Magistrate Judge Paul Levenson found that school officials "have the better of the argument on both the facts and the law."
"On the facts, there is nothing in the preliminary factual record to suggest that HHS officials were hasty in concluding that RNH [the Harris' son, referred to by his initials] had cheated," Levenson wrote. "Nor were the consequences Defendants imposed so heavy-handed as to exceed Defendants' considerable discretion in such matters." "On the evidence currently before the Court, I detect no wrongdoing by Defendants," Levenson also wrote. "The manner in which RNH used Grammarly -- wholesale copying and pasting of language directly into the draft script that he submitted -- powerfully supports Defendants' conclusion that RNH knew that he was using AI in an impermissible fashion," Levenson wrote. While "the emergence of generative AI may present some nuanced challenges for educators, the issue here is not particularly nuanced, as there is no discernible pedagogical purpose in prompting Grammarly (or any other AI tool) to generate a script, regurgitating the output without citation, and claiming it as one's own work," the order said.
Levenson concluded with a quote from a 1988 Supreme Court ruling that said the education of youth "is primarily the responsibility of parents, teachers, and state and local school officials, and not of federal judges." According to Levenson, "This case well illustrates the good sense in that division of labor. The public interest here weighs in favor of Defendants."
How stupid? (Score:5, Insightful)
"The manner in which RNH used Grammarly -- wholesale copying and pasting of language directly into the draft script that he submitted -- powerfully supports Defendants' conclusion that RNH knew that he was using AI in an impermissible fashion
It's pretty incredible that a High School student is so stupid as to not at least re-word what the LLM provided. Never mind the cheating aspect of what he did, although that's bad enough. But that degree of witlessness, at that age, does not bode well for him being able to function in any job requiring intellectual sophistication, analytical ability, or even just basic common sense.
Additionally, those lawsuit-filing parents need to take a good long look in the mirror, get a fucking clue, and ask themselves if their precious bundle of joy came out of the womb with such intellectual deficits or if their parenting was simply that bad.
This kid may be salvageable, but Mom and Dad need to stop enabling their son and get him some help that doesn't come from a lawyer or from an LLM being used as a substitute for thinking and learning.
Re: (Score:3)
In which way "cheating" means "best possible education"?
Re: (Score:3)
In which way "cheating" means "best possible education"?
Cheating in high school can help him get into a better college.
Re:How stupid? (Score:4, Insightful)
Which wont do him any good if he's just going to cheat there too.
Cheating in college can help him get into a good law school.
Then he can go into politics.
Re: (Score:1)
Cheating in college can help him get into a good law school.
Then he can go into politics.
And if he keeps cheating and lying there, he might eventually become president, even twice, as supported by empirical observation and evidence.
Yes, yes, I know. TDS and all. I couldn't care less. It had to be said.
Re: (Score:2)
You were modded Insightful, which I agree with - but my first reaction was a belly laugh. Great way to start the day - thanks!
Re: (Score:2)
Re: How stupid? (Score:1)
Well they're setting him up for an excellent career on Wall Street
Re: (Score:2)
Re: (Score:3)
Re: (Score:3, Insightful)
I use that all the time, because my grammar isn't great.
If that concerns you,
1. Get a good book on English grammar, study it, and apply it.
2. Read the classics of English literature, from Dr Johnson and Gibbon to Kingsley Amis and Gore Vidal. You will gradually soak up the habits of good, clear writing.
Re: How stupid? (Score:2)
Get a book on processing disabilities, esp dyslexia. For many, study wonâ(TM)t fix it; the brain is literally using wrong sections to process language. Itâ(TM)s like saying âread a book, run a 4 minute mile.â(TM)
Re: (Score:2)
Hang on, slow down, its Grammarly we're refering to, not ChatGPT. Its just a grammar checker isn't it? Really, just a step up from a spelling checker.
I use that all the time, because my grammar isn't great. Thats not 'cheating' its just cleaning up text. If it was a primary school english exam, sure, but we're throwing the baby out with the bathwater if we ban grammar and spelling checkers.
I think the parents where right to protest this.
No, Grammarly has AI content generation now as well. They advertise it on their homepage now. It is not just to check grammar anymore.
Re: (Score:2)
Isn't Grammarly just a better spellcheck?
Re: (Score:2)
It does have some "advanced" capabilities, like rewording and offering feedback, but yeah. It's not ChatGPT.
Re: (Score:2)
It does have some "advanced" capabilities, like rewording and offering feedback, but yeah. It's not ChatGPT.
I get the point that you and some previous posters have made that Grammarly isn't a ChatGPT clone. Thanks for the correction - what the student did wasn't as egregious as I thought, and my criticism was too harsh.
That said, Grammarly's own site says "Get personalized suggestions based on what you’re writing and who will read it" and "Be perfectly professional, clear, and convincing in a few clicks, not a few hours". So yeah, I still say that the student cheated, and that he should have known better.
Wh
Paper (Score:2)
My son is the editor of his school paper, and he rejected a submission by a writer that was clearly AI generated. Beyond the janky prose, the smoking gun was the use of an em dash, only available on school computers by hitting ALT-0151. The newspaper class is pass/fail, and you only have one assignment per semester, so that student failed the class.
The maddening thing, he said, was that it was a puffball piece the student could have wrote in one class period. It's one of the easiest classes, if you can writ
Re: (Score:2)
If you are using Word, it is a lot easier than that to enter an em dash. Sometimes, it will autocorrect a regular dash to an em dash, or you can do Ctrl+Alt+[Numeric keypad -] to get it.
Re: (Score:2)
the smoking gun was the use of an em dash, only available on school computers by hitting ALT-0151.
Microsoft Word replaces inline "--" with em dash. So does Google Docs, as does Pages (at least on ios). Janky prose aside, if your son actually called that out as the smoking gun he has no business being an editor.
Re: (Score:2)
the smoking gun was the use of an em dash, only available on school computers by hitting ALT-0151.
Microsoft Word replaces inline "--" with em dash. So does Google Docs, as does Pages (at least on ios). Janky prose aside, if your son actually called that out as the smoking gun he has no business being an editor.
Add Libre Office to that list as well.
Re: (Score:2)
...and definitely don't get the kid a BMW for his birthday. It would only seal his fate.
Kudos to school for standing up! (Score:5, Insightful)
All the administrators I've worked with cave in at the slightest threat of legal action. That this school had the integrity to fight the lawsuit indicates its students should be proud to study there.
Re: (Score:2)
And....is there a prohibition of teachers using AI to reduce their workload?
No. Why should there be? The purpose of the teacher is to impart knowledge to the student, if AI is making them better at that task, then that's a win. On the other hand, if it's a shitty teacher leaning on AI and actually getting worse at their job, that's a problem--but the problem isn't the AI.
Isn't it unethical to ask your students to do your job as a teacher and then to profit by repackaging and reselling their work?
Yes, absolutely. That is plagiarism and likely copyright infringement. That is a separate issue from the above.
Imagine: Only using the top 5 student's homework, repackaging it into a document to sell on Teachers pay Teachers, using it in the next year's class instead of doing what is a teacher's job.
Oh FFS, this is just stupid. Issues relating to the ethics of stealing the work of your students
Re: (Score:2)
What's the alternative? If they give cheats the highest grade or even, a passing grade, there's no incentive for any student to be honest. They may as well, close the school.
Re: (Score:2)
What's the alternative?
I think quite often it's the worst thing possible. To ignore it but then to find ways to manipulate down the suspected cheats marks. I was very early using printed out papers on a computer. One year I got a pretty bad mark for a school paper that I thought was okay. The next year, I had an actual external examiner for a science paper (not normal in the UK). After talking to me he gave me a great mark. I heard from my teacher, though, that he'd looked and to begin with thought I'd copied it from a journal an
Re: (Score:3)
Please come up with A SINGLE EXAMPLE of a High School level HOMEWORK problem that could not be done by an LLM.
I will proceed to show you how to use an LLM to help solve it.
Re:LLMs are here to stay, use them (Score:5, Funny)
Easy, ask how many Rs there are in "Strawberry"
Re: (Score:2)
write me a python script which counts the occurrence of an input character in an input word and outputs the number of times the input character occurs.
That should get you started.
Re: (Score:3)
Not sure if this is whoosh or not but that is exactly what Deepseek LLM (or at least what is being used in Kaggle and math related functionality) does. It writes and executes a Python program. Then it vacillates back and forth because it "feels" like it should only have 2 r's based on non-Python LLM foolishness like phonetics and broken syllable splitting, but then decides yeah it really has 3. :u
Re: (Score:2)
Zero.
Or is that ZeRo?
Re: (Score:2)
I'm 99% certain that question never came up in my high school syllabus, it certainly would have stuck out between the questions on rudimentary calculus and the "calculate the supply/demand equilibrium" type questions I had in high school economics.
If the kid thinks about girls all day, its probably not gonna be in the exam. If instead he thinks about Cooklie Monster and Blues Clues, it probably will be.
This was a high school kid. In trouble for using a gramma
Re: (Score:3)
Here:
What are letters in the word strawberry? Omit letters that are not R. How many letters remain?
ChatGPT: To solve this:
The word strawberry is spelled: s-t-r-a-w-b-e-r-r-y.
Removing all letters except R, you're left with: r-r-r.
There are 3 "R" letters in the word strawberry.
Re: (Score:2)
Seems to prove the GP's point that you can't just ask it "how many letter Rs are in the word strawberry?" because you had to phrase it in a way that instructed ChatGPT on how to reach the correct answer.
Re: (Score:2)
My prompt: How many Rs there are in "Strawberry"
ChatGPT Response: In the word **"Strawberry"**, there are **three** R's.
What are you talking about?
Re: (Score:2)
They've fixed it now, but it used to be a problem.
Another one they fixed, but used to be a problem was asking it which is further north - London England or London Ontario. It quoted correct GPS coordinates for both, but then drew the wrong conclusions from that information.
Re: (Score:2)
Which is incorrect. There are no "R" characters in "strawberry". There are however 3 "r" characters. Any fule kno that "R" and "r" are different characters.
Re: (Score:2)
LRMs like OpenAI's o1, or now the open source Deepseek R1 [deepseek.com] can do these problems just fine (they're still not perfect, but they're much better). Don't even bother with "strawberry", since everyone uses that example, pick something more obscure.
Note that LLMs are inherently disadvantaged in these, as they don't see "letters", but rather, tokens. And there's way more tokens than letters, and there's not going to be anything in the training dataset for most of them to explain how many of what letters are in t
Re: LLMs are here to stay, use them (Score:3)
Re: (Score:2)
You use Wolfram Alpha for that! Duh.
Re: (Score:2)
Its bad enough (Score:1)
Re: (Score:2)
When a student cheats you arnt learning a skill go progress to the next step and whether you get caught or not, it only ends up hurting them. When the parents are reinforcing the behavior, its never likely they will ever correct that and do anything but look for a shortcut which will ultimately hurt them and everyone working around them.
In other words, he's a prime candidate for a high-powered Wall Street job!
Actually it is not cheating (Score:2)
it is plagiarizing a copypasta from an AI and not citing a source.
Re: (Score:1)
it is plagiarizing a copypasta from an AI and not citing a source.
You technically can't plagiarize a LLM's output because it isn't copyrightable in the first place. So, it really does fall under the category of cheating rather than failing to give credit to the original author.
Of course, this situation will probably only last until some company like Disney creates something of actual value using AI, then all bets are off.
Re: (Score:2)
You technically can't plagiarize a LLM's output because it isn't copyrightable in the first place.
Plagiarism is not copyright infringement; it's an academic moral offense not a legal one.
Plagiarism covers "passing off ideas as your own" which are not even copyrightable in the first place. You cannot copyright an idea.
It is plagiarism if you copy a work or ideas from a work without citing: Schools have even considered it plagiarism if it's our own work that you copied!
Using the same essay you created for
Re: (Score:2)
The parent's need to own up to the fact they failed as a parent and then reinforced that parenting failure for everyone to see by suing the school. Such a family is obviously a mess with no morals.
Re:Rules? I don't need no stinking Rule. (Score:4, Insightful)
The only one harmed by the student using AI was the student.
If the student is admitted to a prestigious college based on unearned grades, a more deserving student is denied that slot.
The "public" has no interest at all at stake here.
We fund public schools with tax dollars precisely because there is a public interest and benefit in an educated workforce.
Can you imagine a judge saying that about a decision involving the competing interests of two companies?
Judges do say that. Anti-trust law is a good example. A competitor can't just claim they were harmed. They must show that the public was harmed.
"The law directs itself not against conduct which is competitive, even severely so, but against conduct which unfairly tends to destroy competition itself." -- United States Supreme Court
Re: (Score:2)
The public has in interest in being able to trust school grades as a reliable measure of skill in a particular subject.
Re: (Score:2)
It also protects the school by setting a precedent that cheating can not be covered up by being sued by parents. Once again this is also in the public's best interest.
Colleges will also see the true nature of the student giving other applicants a fair chance. How can you not see a public interest?
Well this is interesting (Score:1, Interesting)
Now that Microsoft 365 ships with Copilot by default enabled without any option to disable it, this includes Family, Personal, and Student editions I am not sure schools can easily mandate that a student must not use AI at any level of education.
Especially when many educational facilities require Microsoft compatible formats which OpenOffice and LibreOffice do not quite achieve.
Honour system (Score:2)
Schools & other educational institutions need find another way to ensure that their students are assess
Re: (Score:3)
And the reason so many people fail...
BE EXPLICIT BUT ALSO GENERIC.
Don't say "iPads are not allowed". Say "Tablet computers, or other types of computers, are not allowed".
Don't say "AI submissions are not allowed." Say "Your submission must be entirely your own work."
People who draft policies go too specific, too narrow, and don't add "or other similar devices" etc. They state specific things, rather than stating the intention and then listing a few pertinent contemporary examples. ("The use of electroni