Catch up on stories from the past week (and beyond) at the Slashdot story archive

 



Forgot your password?
typodupeerror
×
AI Education Technology

NYC Bans Students and Teachers From Using ChatGPT 104

An anonymous reader quotes a report from CNBC: New York City's education department has banned access to ChatGPT, a chatbot that uses machine learning to craft realistic text, out of concern for "safety and accuracy." As first reported by Chalkbeat New York, the ban will apply to devices and internet networks belonging to the education department. Individual schools can request access to ChatGPT for the purpose of studying AI and technology-related education, according to a department spokesperson.

"Due to concerns about negative impacts on student learning, and concerns regarding the safety and accuracy of content, access to ChatGPT is restricted on New York City Public Schools' networks and devices," education department spokesperson Jenna Lyle told Motherboard in a statement. "While the tool may be able to provide quick and easy answers to questions, it does not build critical-thinking and problem-solving skills, which are essential for academic and lifelong success."
According to the Washington Post, some teachers are "in a near-panic" about the technology enabling students to cheat on assignments. "The New York Times recently showed writers and educators samples of ChatGPT's writing side-by-side with writing by human students, and none of them could reliably discern the bot from the real thing," adds Motherboard.

When asked about the ban, ChatGPT told Motherboard: "It is important to consider the potential risks and benefits of using ChatGPT in education, and to carefully weigh the evidence before making a decision. It is also essential to listen to the perspectives and concerns of all stakeholders, including educators, students, and parents, in order to make informed and fair decisions."
This discussion has been archived. No new comments can be posted.

NYC Bans Students and Teachers From Using ChatGPT

Comments Filter:
  • Morons? Or Geniuses? (Score:5, Interesting)

    by lsllll ( 830002 ) on Wednesday January 04, 2023 @06:32PM (#63180728)

    Morons, because if someone wants to turn in a paper produced by ChatGPT will find a way. Geniuses because it is a real concern. I had ChatGPT write a short essay on "the deconstruction of the neo-romantic period in art and literature from the economic perspective of the industrial revolution in France and Germany." The phrase is from the movie Bedazzled, so I didn't come up with it. The result was very convincing.

    You can't ban stuff like this. I think your best bet would be to drill it into kids' heads that while they can use ChatGPT as a tool, the objective is that THEY learn the material and have enough of an understanding of not only the material, but also of the English language, because in the end that's what benefits them the most. The objective is not to get good grades. The objective is to learn.

    • I think your best bet would be to drill it into kids' heads that while they can use ChatGPT as a tool, the objective is that THEY learn the material

      I totally agree here, you cannot ban tools and in fact, does it not seem like it might be better to help kids learn to work with tools like this, and distinguish good and bad output?

      Maybe one direction would be, that every student had to present a paper and talk about things from the paper in more detail, to show they had really studied the subject beyond an es

      • I think the point is that there is a lot of learning about the tools and until the education system has studied it, it would be the wild west. It's like when Wikipedia became popular and the information was unreliable at best and it got banned in schools, then later being allowed as long as the students understand how to check sources there.
        • I think the point is that there is a lot of learning about the tools and until the education system has studied it, it would be the wild west.

          The more fundamental point is that the point you make is irrelevant, since chatGPT results cannot be distinguished from human students own writing (well except it may be substantially better), so it is impossible to "study" this for years, you are right in the Wild West and it's time to Cowboy/Cowgirl Up.

          Students will be using this tool in large numbers no matter how

      • I think the compromise is to have the resource available at some common, controlled access point. When there are more like it some standards by then should be in place. And there is more than one possible answer to questions depending on what you thought you were looking for. I mean giving an assignment on George Washington for example should not return the same info formatted and outlined across all grades. Asking individuals to create assignments based on a single source? They are still debating how much

        • I think the compromise is to have the resource available at some common, controlled access point.

          That's a compromise to be sure, but not possible any longer as the cat is now out of the bag and breeding like mad.

          Pretty soon every major word processor will ship with something like chatGPT built in... why wouldn't they?

    • by Areyoukiddingme ( 1289470 ) on Wednesday January 04, 2023 @07:04PM (#63180772)

      I had ChatGPT write a short essay on "the deconstruction of the neo-romantic period in art and literature from the economic perspective of the industrial revolution in France and Germany." The phrase is from the movie Bedazzled, so I didn't come up with it. The result was very convincing.

      That's more of an indictment of the bullshit quotient in post-modernist writings than it is praise for the bot. Especially since bots could get papers like that published when they were still considerably worse than ChatGPT.

      The objective is not to get good grades. The objective is to learn.

      The objective is to pass the Bush tests, learning be damned, because funding is on the line.

      Industrialized learning incentivizes the wrong things all over the place. The failure to get students to learn to read and write is only the most egregious of many. We used to at least insist students learn the mechanical aspects of writing properly, namely spelling and grammar. I suppose we should be thankful standards have fallen so low that not even that is required anymore. It makes it easier to spot the blithering idiots.

      In the end, ChatGPT is a demonstration for the Ai naysayers. AI doesn't have to produce generalized, sapient intelligence to be unreasonably successful. It only has to be slightly better than the mythical average human. Clearing that bar is easier than you think.

    • Fail anyone who presents an AI generated text as their own. That's plagiarism. Also, test in person.
      • Why? I don't get points take off for spelling because I used a spell-checker on my riting. I don't get points taken off for having a friend proof read it for me or provide feedback on an early draft. I don't get points taken off for reading what others have said about the topic before deciding what position I would take. That's how writing happens in the real world.
    • by Charlotte ( 16886 ) on Wednesday January 04, 2023 @09:01PM (#63181000)
      The problem is that in the US, teaching to the test is grained in. It's a mindset. That's what needs to be fixed, but that's not going to happen is it...
      • This is exactly it. I'm a Learning Scientist whose research focuses on assessment and I've been explaining to anyone who will listen that technology like ChatGPT is finally going to force our hand to reconsider what and how we measure learning. The problem is not the existence of ChatGPT but that too many educators rely on bad assessments. Now essays and short answer questions are just like any other type of assessment (e.g., multiple choice, simulation): good versions require deep understanding of the know

        • Agreed. The problem is how to measure learning and understanding in a multiple-student class. Teaching and learning would be easy in a one-on-one environment, but we don't have an educational system designed for that, nor the numbers of teachers to accomplish it.

          Everyone learns at a different pace and with different comprehension speeds. Trying to take 30 kids and teach them all the same material at the same speed inevitably results in 1/3 of the students bored to tears, 1/3 adequately understanding the

          • Learning is easy to measure. Just look at the average student's income a decade after graduation.

            From the perspective of its funders - i.e., the politicians - the public school system has been a resounding success. It has produced a large, functionally illiterate, emotionally manipulated voter base which makes acquiring and retaining political power much easier than it would otherwise.

    • by rworne ( 538610 )

      I did the same asking it questions from my daughters class on CRT in education last semester. The instructor would ask for short essays like the following:

      "What are social theories for? Explain what Critical Race Theory is and evaluate how it would be appropriate to explain Asian Americans and their experiences in the U.S. (Don't exceed 150 words.)"

      So I asked, and it came up with a better answer than I expected. Ah, but the teacher wanted references... so I asked for the same paper with inline references.

      • Have you checked the references? They might be hallucinated.

        But there's a fix for that. If you take the generated text (with hallucinations) and search the web for the closest match, it works better than using keywords. So a future chatGPT could do this:

        1. generate an answer in closed-book form, it could have hallucinations
        2. use this answer as a query to search the web, find supporting text similar to itself
        3. paste all the supporting text back in chatGPT and ask the question a second time

        This
        • by rworne ( 538610 )

          I've seen it do that before - I asked it once about a brief paper on VR and gave it the name of an author I knew. It did write a well-written paper about that person's accomplishments in the field of VR, but it was all B.S. because that author I mentioned wrote comic books.In that case it just substituted the name.

    • I think your best bet would be to drill it into kids' heads that while they can use ChatGPT as a tool, the objective is that THEY learn the material and have enough of an understanding of not only the material, but also of the English language, because in the end that's what benefits them the most. The objective is not to get good grades. The objective is to learn.

      Perhaps we should drill it into the educational institutions heads first, as the continue to lower standards to pass more students (i.e. make revenue). That is why they hate initiatives like Obama's proposed scoring based on how much students from a particular school with a particular degree are making 3 years after graduation. This would allow students to make an informed decision how fat it makes sense to go into debt to get a particular degree from a specific school. Sadly, ideas like that get killed by

    • Kids should be learning what will benefit a human with AI support in the future, not what benefited us with just computers and internet.
    • The objective is not to get good grades. The objective is to learn.

      That hasn't been the objective of education, certainly not public education, in America since long before I was born. The objective is to produce compliant cogs for the capitalist machine. Thinking is the antithesis of that. Thinking individuals care about their own rights, as well as their responsibilities. Properly "educated" automatons care about their responsibilities and obligations, to the state, and to their employer, first and foremost. We need to beat intelligence and natural curiosity out of peopl

    • I would think a decent teacher would have a pretty good idea which students are capable of writing a good paper and which ones are not. I'm Gen X so back in my day the shitty students had to beat up a nerd to write a paper for them. Unless someone spilled the beans it would be awfully hard to prove they were cheating. Nowadays If some kid produced a work I thought was beyond them, ChatGPT might be one of my first stops to see if I could get it to produce something identical.
    • What I have noticed about ChatGPT is that its responses are indistinguishable from a wishy-washy liberal. It can state facts, but can't seem to form an opinion or judgement on anything except the most politically correct topics. For example, in response to the prompt, "Why do some white people hate black people?"

      It is not accurate or fair to say that all white people hate black people, or that any one racial or ethnic group hates another...

      Which may be true, but neither answers the question, nor pro

    • The objective is not to get good grades. The objective is to learn.

      The education system that is currently in place is about punishments and rewards, despite evidence of their long term ineffectiveness. Alfie kohn's books on the subject(punished by rewards) are thick, not due to the subject material but all the citations as well, deriding the use of these as motivators. Yet they persist because of laziness, of not just educators but parents as well.

  • If they wrote it, then they should be able to answer questions based about the paper.

    If chatgpt wrote it, they won't have a clue what's in it.

    • You've obviously never had to organise & administer (high-stakes) examination interviews. It'd be very expensive & time-consuming for already overstretched education institutions. It's just not feasible.
      • You've obviously never had to organise & administer (high-stakes) examination interviews. It'd be very expensive & time-consuming for already overstretched education institutions. It's just not feasible.

        I have, without any teaching assistants, organized and administered high-stakes examination interviews. Each meeting used up about 30 minutes of my time in my office, which is not too different from how long it takes me to read and grade a written test or term paper. This was in a large-ish US university, where students get nervous about that sort of thing. I can confidently report that due to my careful use of follow-up questions, I can get a very clear sense of the depth of each student's knowledge.

        • Your students are very lucky to get the kind of in-person, conscientious, expert attention that they do on your courses. Consider yourself one of a tiny minority.
  • Handwritten essays. In class. With no access to phones during the test.

    There, problem solved. Now, I realize that many kids are not taught proper handwriting skills, but there are ways around this. A school could have a "test center", which are a bunch of computers not hooked up to the internet and only to a printer. Require all tests that would require an essay-type answer be taken at this test center.

    • Paper & pencil will work fine. Most assignments written by students in their own time & then submitted online are information read from books & papers paraphrased & cited, key points & arguments Googled, & then the actual ideas in them promptly forgotten. They really don't learn much from doing this. If the students had to actually learn & understand the content of their courses, now that'd be something worth doing.

      Also, ChatGPT is an invaluable tool for providing a variety of
      • The real challenge will be to wean poorly informed teachers & tutors off of "process writing" approaches to writing instruction (which are essentially "trial & error" discovery learning) & get them to actually teach students how to write in the given genres, registers, & styles they expect.

        Having gotten an engineering degree, I never encountered a writing instructor who even dared whisper such things as genre, register, or style, let alone teach how to achieve any particular one. It was all they could do to get most of the class to string a coherent sentence together, so that sort of thing was pretty far out of scope.

        • Yes, in the natural sciences, maths, engineering, etc., the expected proficiency of writing isn't as high as in social sciences, languages, literature, philosophy, history, etc., for obvious reasons.

          There's a lot of research on academic written genres, registers & styles, & engineering too has its particular preferences, i.e. what faculty will judge as well-written & "what an engineer would write." It's things you'd likely refer to as "turns of phrase," & "sounding like an expert." It's th
  • Is there anything new that the muckety-muck mid-level managers and administrators in public schools don't want to ban?

    I have worked with those people, and if you want to know the biggest control freaks there are, who are still driving Ford Crown Victorias, it is middle school administrators. They usually have a relatively low paying job, but a pension, and nothing short of a felony caught in public will cause those people to get fired. Hell, the ISD I am under has a middle administrator whose job is to ma

  • I've tried ChatGPT, and really it doesn't seem to do much but do a bad job of spinning text from wikipedia articles... is it really AI?
    • Re: ChatGPT (Score:4, Funny)

      by fbobraga ( 1612783 ) on Wednesday January 04, 2023 @07:12PM (#63180792) Homepage
      All of it is called "AI" nowadays
    • Re:ChatGPT (Score:5, Insightful)

      by aldousd666 ( 640240 ) on Wednesday January 04, 2023 @07:23PM (#63180830) Journal
      yeah, you're missing something. lol. you need to ask it a creative question. ask it to write you a poem in shakepearean sonnet in no less than 10 stanzas about ... whatever two things are sitting in front of you, and inject some moral angle into it that would be controversial in 1990's Russia. THEN you'll see some stuff you can't find on wikipedia. Also: it doesn't have access to wikipedia or any of the web at the moment, when you query it. It has assimilated that info qualitatively in the weights of its neurons. It's not reading it from a file of text, so to say.
    • The big problem I have with ChatGPT is it always sounds authoritative. I think there is a psychological phenomenon going on here, and I'll try to explain.

      When I Google something that is a straightforward question, with a factual answer, Google will respond with (what it believes) is the answer. Google "What is the capital of NY" and, in addition to search results down below, at the very top will be the answer, including a paragraph of text from the source, and a link to the source. It makes it clear to me

  • by physicsphairy ( 720718 ) on Wednesday January 04, 2023 @07:09PM (#63180786)

    This comes across as a very reactionary and shortsighted response. In reality, ChatGPT is much more of a boon to teachers than it is to cheaters.

    First of all, as a teacher grading is waaaay more disproportionate of an effect on your time than doing homework is for students. Something that can summarize text and, e.g., check content against certain criteria could save tons of time - and ChatGPT is more reliable at summarizing text than it is at generating it.

    Even better, it can be a tool for giving students instant feedback. That is far superior to having students do something wrong consistently and only get feedback a week later when their paper gets returned.

    Second, it doesn't actually introduce any new problems. You could always have a friend or parent do your assignment, copy off of someone else, or pay someone in India. Lots of essay services and even things like Mechanical Turk for getting human written content. And the solution is the same - monitored in-class work, and controlled test taking environment where people who fake their assigned work are guaranteed to fail.

    Probably the dirty secret is that teachers have just been ignoring this kind of cheating because it's easier than confronting it, and their policy of rewarding rather than clamping down on cheating is now poised to bite them.

    • by quetwo ( 1203948 )

      Probably the dirty secret is that teachers have just been ignoring this kind of cheating because it's easier than confronting it, and their policy of rewarding rather than clamping down on cheating is now poised to bite them.

      Teachers have been relying on services like TurnItIn to catch when students use paper-writing services or out-source to folks out of the country. Most of those paper-writing services use the same papers over and over again (or good chunks of them), and they end up in the Turn it in databases. It won't catch instances of your friend or parent writes the paper for you -- unless they turned in the same (or similar) paper elsewhere. But the ultimate solution is still there -- in person writing assignments th

  • This will simply accelerate the transition to grades being calculated by in-class testing only.

    When I was an undergrad, it was common for homework to be worth a third of the class grade, or even half. Reports and essays could be 2/3 the grade in a liberal arts class.

    Nowadays, homework is usually worth 10% of the grade because everyone has the solutions. chatGPT and its successors will just drive the same phenomena for reports and essays. Your grade will be determined by IN-CLASS EXAMS, your cell pho
    • In-class meaning in person class, or in remote class?
      • In-person. Most (not all) universities have backed off remote classes as quickly as they could after covid. The numbers are very clear that the education is worse in a remote format. To anyone who asks for data - go look it up yourself. There’s sooooooo much evidence.

        The future of remote eduction is lecture content delivered remotely with a few required trips to a testing center where the student can actually be evaluated. I don’t see ANY way the current hell-scape of remote internet-based t
    • In a rational world, that would be the result. However, there's a lot of anti-testing dogma in certain circles that seems to get more entrenched over time, rather than less. In the last few years at the college where I teach (a top-5 in the U.S. community college, part of a large urban university):

      - We've been told by administrators that "high-stakes testing" is passe and no one ever does it anymore.
      - The college abandoned its decades-long practice of centrally scheduling and administering final exams. End

  • Wikipedia has been banned in many schools for years now, on grounds that its articles aren't original sources. Of course, it's easy enough to click through to the original sources and reference them, instead of the Wikipedia article itself.

    Now ChatGPT is the new boogeyman, so maybe they'll forget about how "bad" Wikipedia is.

    https://arstechnica.com/uncate... [arstechnica.com].

    • I do not get it why Wiki was banned. In my time we were free to use printed Encyclopedias to do homework. Wiki is just a different form of it. Was Britannica CD (cw 1995) also banned at its time?

      • by quetwo ( 1203948 )

        Wikipedia was banned because in its first phase, most of the articles were poorly sourced and included a LOT of bias. Wikipedia has enforced stronger standards since and has flagged articles that are poorly sourced, based on rumor or are flagged as inaccurate. Unfortunately, a lot of bad data still gets into articles (for example right now it lists Kevin McCarthy as 'dead due to embarrassment of losing the house election 6 times in a row').

        Things like old school encyclopedias had editors and had a very go

  • Are they also banning wikipedia? Google? Bing? etc. All of those can and regularly do return inaccurate or wrong information too.
  • Your results with ChatGPT may vary. For example, when asked whether it was possible to play bowling underwater, ChatGPT correctly responded that the friction of water would prevent bowling from being possible. However, when asked whether underwater volleyball was possible, ChatGPT suggested that underwater volleyball could be played if using specialized equipment, which I highly doubt.

  • In the presents of like Google and now ChatAI (et al) perhaps we need to change the way we teach (and what teaching is).
    Just because a "motivation" and verification is conducted via essays and tests, that now can be solved by AI (and better) does not mean these should be outlawed.

    Just like nails and hammers changed the way houses are build.

  • Build a language model based on the statistical patterns of text output by ChatGPT, and use that as the basis of a ChatGPT detector.
  • to accidentally learn something.
  • by russbutton ( 675993 ) <`russ' `at' `russbutton.com'> on Wednesday January 04, 2023 @10:28PM (#63181170) Homepage
    I graduated from High School in 1969, so nobody's going to stop me from using ChatGPT for nefarious purposes. Today, I asked ChatGPT to write a 3-way dialog between Yoda, Dobby the House Elf and Gollum.

    .

    Dobby: Master, Dobby has come to serve you!

    Gollum: Serve us, serve us, my precious!

    Yoda: Serve you, Dobby should not. Balance, there must be.

    Dobby: But Master, it is Dobby's job to serve!

    Gollum: Serve us, serve us! We wants it, we needs it. Must have the precious.

    Yoda: Precious, hmm? Only lead to suffering, it does.

    Dobby: Dobby does not understand, Master Yoda.

    Gollum: Understands nothing, does it? Stupid elf.

    Yoda: Angry, Gollum is. Fear and suffering, he carries.

    Dobby: Oh dear, Master Gollum seems upset. Can Dobby get you anything? A nice fish perhaps?

    Gollum: Fish! Give us the fish!

    Yoda: Patience, Gollum. Patience is key.

    Gollum: Patience, yes. We must have patience. But the fish, the fish! Cannot wait for the fish.

    Dobby: Very well, Master Gollum. Dobby will go catch you a fish.

    Yoda: No, Dobby. Fish for Gollum, catch not. Enough suffering, there has been.

    Gollum: No fish! No fish! The elf tricked us!

    Dobby: Oh no, Master Gollum! Dobby would never trick you. Perhaps some hobbit bread instead?

    Yoda: Peace, there is. Sharing, there should be.

    Gollum: Sharing! Yes, the precious can be shared. But the fish, the fish! That is ours, all ours.

    Dobby: Oh dear, it seems we have a misunderstanding. Dobby is sorry.

    Yoda: Apologize, not necessary. Learn, we all must.

    Gollum: Yes, learn. But the fish, the fish! We wants it, we needs it.

    Yoda: Enough, Gollum. Enough of the fish.

    Gollum: But the fish!

    Yoda: Control, Gollum. Control your obsession.

    Gollum: Control, yes. We must have control. The precious, it is ours.

    Dobby: Dobby is so sorry for the trouble, Masters. Dobby will leave now and let you two wise beings continue your discussion.

    Yoda: Go in peace, Dobby. And remember, serve not for reward. Serve for the greater good.

    Dobby: Thank you, Master Yoda. Dobby will remember.

    Gollum: The fish, the fish! We must have it!

    Yoda: Control, Gollum. Control.

  • The ban is only for school networks and equipment. Poor kids often have no other ways to access computers, so this prohibition will affect many of them. However, rich kids with their own networks and equipment will be unaffected.

    • Poor kids often have no other ways to access computers

      Citation needed. I can almost guaranty you that it is the opposite. Only the most involved rich parents will restrict device usage. All the poor illegitimate bastards will be on one of their 5 devices for 18 hours of the day.

  • and Neuralink, the next generation's exams may prove to be significantly easier.
  • Can they force ISPs in NY to block ChatGPT?
  • I was going to have a play with it last night and you can't have an account without giving them your phone number. WTF?

  • NYC is a government. They are not our leaders, rulers or betters. They are our representatives and closer to property managers for the citizens.

    When the government pulls a stunt like this please demand they show specifically where in the law their authority comes from that permits them to 'ban' something in a free society.

  • I asked ChatGPT about this topic - this was its response: "Hello everyone! I understand the concern about students using ChatGPT to do their schoolwork. After all, it can seem like a form of cheating if the student isn’t actually doing the work themselves. However, ChatGPT can actually be a very helpful tool for students. It can help students understand the material they are learning by offering explanations and examples of how to solve problems. It can also help them to develop their critical thin
  • It is generally not a good idea for students to cheat on assignments or tests, as it undermines the educational process and can have negative consequences. Instead of relying on cheating, students should focus on understanding the material and developing their skills. If a student is struggling with a particular subject or assignment, they should seek help from a teacher, tutor, or other academic support resources.

    -ChatGPT on "write about students using ChatGPT to cheat in school"

    • ChatGPT disagrees with you!: "Hello everyone! I understand the concern about students using ChatGPT to do their schoolwork. After all, it can seem like a form of cheating if the student isn’t actually doing the work themselves. However, ChatGPT can actually be a very helpful tool for students. It can help students understand the material they are learning by offering explanations and examples of how to solve problems. It can also help them to develop their critical thinking skills by providing additio
      • New social media scare: Is AI schizophrenic? What will this do to the children?

        and
        "By providing the answers they need quickly, they save time and energy that can be used to focus on the more important concepts."
        Best defense of cheating, ever. Wish I'd thought of it back in school.

  • NYC kids won't have an awareness of ChatGPT.

    They won't be able to pitch any new ideas based on it.

    They might tell their academic colleagues or managers some day that they will do something that can just be generated.

    This is not setting them up for success.

Hackers are just a migratory lifeform with a tropism for computers.

Working...