6

AI Detection and Grammarly

Hi everyone, We have faculty using the AI report and are appreciating that feature. An interesting question came up recently: the report showed 64% AI-generated text, but the student claimed he had only used Grammarly, which also helps to improve phrasing. Has anyone else experienced an AI report picking up on Grammarly, and is there a way of distinguishing between a "grammar help" tool and other generative AI?

Second, will the AI report be visible to students soon? Faculty assumed that students could see it, but the report is only visible to faculty in our LMS.

Thanks,

Jennifer Douglas, American Public University System

119 replies

null
    • Mitra_Asmani
    • 1 mth ago
    • Reported - view

    Hello All, I facing the same problem. Turnitin is claiming 46% AI use (one whole question), but he is claiming that he only used Grammarly, and he is the only one with this high of AI use. No one else in the class has used it.  

      • Carol_Bailey.1
      • 1 mth ago
      • Reported - view

      Hi Karen, according to the Grammarly pricing info, even their free and premium versions (i.e. not GrammarlyGo) contain limited access to text generation https://www.grammarly.com/plans. I haven't seen how this looks in practice, but is it possible that students are using GenAI unwittingly, having been told by their institutions that Grammarly is OK to use?

      I was asked yesterday if my university should disable the (free) Grammarly add-on in MS 365 to avoid giving the message to students that we fully endorse this software. I wonder what other institutions are doing?  

      • Senior Teaching & Learning Innovations Specialist
      • Karen_Smith
      • 1 mth ago
      • Reported - view

      Carol Bailey Here is the response from our FAQs: Our detector is not tuned to target Grammarly-generated spelling, grammar, and punctuation modifications to content but rather, other AI content written by LLMs such as GPT-3.5. Based on initial tests we conducted on human-written documents with no AI-generated content in them, in most cases, changes made by Grammarly (free & premium) and/or other grammar-checking tools were not flagged as AI-written by our detector. Please note that this excludes GrammarlyGo, which is a generative AI writing tool and as such content produced using this tool will likely be flagged as AI-generated by our detector.

      • Carol_Bailey.1
      • 1 mth ago
      • Reported - view

      Thanks Karen Smith - maybe it's not evident from the thread layout that I was responding to your post 9 days ago where you quoted this exact same extract from your FAQs. My point is: on the basis of the Turnitin FAQs, institutions may be telling students 'Grammarly is OK but don't use GrammarlyGo'. In fact, ALL versions of Grammarly include a text-generating element.

      • Senior Teaching & Learning Innovations Specialist
      • Karen_Smith
      • 1 mth ago
      • Reported - view

      Carol Bailey Sorry about that - I'll make a note to myself to review here on the TEN site, rather than responding from email! 

      I'm curious about how other users are addressing this sort of problem?

    • Faculty
    • Shannon_Frizzell
    • 1 mth ago
    • Reported - view

    I had a student, who submitted their original assignment and the Turnitin score was 5%, AI score was 100%. As part of the investigative process, I had them submit their Word doc for statistical analysis and it was revealed to have 3 edits and total time in doc did not correlate with assignment length. I required a resubmission of the assignment and upon analysis it was 0% AI generated content. I reviewed the assignment, which had significant similarity to the original and I met with the student. They told me that all they did was "add some spelling mistakes, remove some punctuation and change a few words."  This is concerning if this is all that is needed to go from 100% to 0% on AI detection. 

      • Senior Teaching & Learning Innovations Specialist
      • Karen_Smith
      • 1 mth ago
      • Reported - view

      Shannon Frizzell Hi! Could you create a support ticket by emailing tiisupport@turnitin.com, please? Please include the paper ID of both submissions, for us to investigate further. Once you submit a ticket, if you share it with me, we can have the product team look into it.

      Thanks!

      • Faculty
      • Shannon_Frizzell
      • 1 mth ago
      • Reported - view

      Karen Smith Here is the support ticket ID:  Case #22307752

      • Senior Teaching & Learning Innovations Specialist
      • Karen_Smith
      • 1 mth ago
      • Reported - view

      Shannon Frizzell Thank you - I'm passing it on to someone who can possibly give us a response.

    • Ahmed_AlAlousi
    • 1 mth ago
    • Reported - view

    How something as critical as someone's future can be left to tools, with very questionable "detection" techniques, is an absolute disgrace. And before the moderators, or anyone else for that matter, shoot me down, I've been against applying automation tools in plagiarism checks since the dawn of their time. Crude at best, not deterministic at all, and subject to whims and feelings. And the reason: reliance on regular expressions? really ?! and I don't care what PhD research came up with it: it's nothing more or less than a regular expression scanner. Then there's the massive privacy question, which absolutely no student/tutor/researcher ...etc. has a hope in hell of addressing, let alone objecting to, when the lawyers in educational establishment have done a sterling job of presenting a simple, binary choice: submit, or fail. And that question, in case I need to state the obvious, is huge Big Brother DB (or is it a data lake now, or something even more whacky ?!) that feeds the automata, for which no one knows a size or extent (everything maybe a good starter for ten) ?

    In comes "AI detection". Yet another flop, and yet another RE scanner. Only this time, it's supposed steroids (Not !!).

    So here's one for you Turn it In: do you really think that by suggesting this is a "guidance tool", and that tutors have total discretion is really believable ?! have you seen the workloads university and college staff - indeed all those in the education sector in the UK, at least - have to deal with?! if you did, and I'm sure you did, then it becomes blatantly obvious that Turn it In is not just being used as a tool, but as a load bearing relief pillar, in the background, while no one's listening.

    And here's the cynically funny bit: I'm yet to actually see an "AI Detection" tool, turn it in included, which can't be fooled into detecting human writing as machine generated, and vice versa. Something which wouldn't fool a human (remember these ?!).

    The reality we all face is this: generative AI is here, for better or for worse. And the sensible thing to do is to teach students, indeed all users, of the tech how to utilise it to help process the ever increasing demands work and research is putting on people. To use the tools as effective search engines and tools, to deliver their own creative work, without fear from someone's paranoia actually ruining their present, and their future alike.

      • Turnitin Admin, University of Huddersfield, UK
      • s_d_bentley
      • 1 mth ago
      • Reported - view

      Ahmed Al-Alousi An interesting take and I agree with quite a lot of what you say. But I wonder on the point about being a load bearing relief pillar if actually staff under the strain of workloads are more likely to ignore potential misconduct rather than deal with it sloppily?

      • Ahmed_AlAlousi
      • 1 mth ago
      • Reported - view

      Steve Bentley It cuts both ways. The same staff are likely to just let it slide, and accept the AI usage report as gospel, rather than take it all in context. In reality, no AI is going to replace a person; if you're not up to the job, and/or your heart's not in it, the answer is not to turn to an automaton for help - it's more either skill up, shape up, or get up and change jobs. And this is coming from a computer scientist - yours truly.

    • Patrick_Kelly
    • 1 mth ago
    • Reported - view

    In a recent 500-word scientific writing assignment, 6 out of 20 students came back with Turnitin AI scores of 32-100%, with 3 of the 6 having the maximum (100%).  In individual meetings with my students (not just those with positive AI scores), only one student admitted to using Chat GPT, and apologized for doing so.  At least one other mentioned using Grammarly.  

      • Ahmed_AlAlousi
      • 1 mth ago
      • Reported - view

      Patrick Kelly I can cite tens of examples to the contrary, for each of your 6 students. If you, as an educator, can't work out who's who in your class, and who's capable of what, to have these conversations (something our own teachers managed beautifully, without the horrible RE/pattern matching/robots we have today, thank you very much); if you can't actually do that, then why are you an educator in the fist place? job too stressful? too little pay? walk away man !! there are other challenges in life, and other ways to make a living !!
      But for anyone to turn around and claim that what's alarmingly common practice in education of using this form of tool as a yardstick, by which such critical decisions as someone's pass/fail (i.e. life impacting) is actually somehow OK, is absolutely and utterly a whitewash.
      Let me give another example from my other engagement as a full-time enterprise architect, working in a really sensitive and people focused consulting position in the UK government; you get them green as grass out of university, and they're eager to please - so you uneducated them, educate and train them again, and you have your next gen practitioners taking on real world challenges - that's how it's done, and I think this is just preaching to the converted. Come tools like ChatGPT, Bard-aka-Gemini and Claude; they'll generate the so-called material, and one look at it tells you it's moronic and bland - just one look; you send them back packing, and they do the work from first principles; that, on its own, is a lesson in life, and a lesson in education. So if it was going to add to the workload to have to do that for a calls, which I also teach part time, then so be it; and if that workload became too much for the pay, then it's a conversation with your employer, and that conversation doesn't include using the likes of Turn-it-In's AI detection tools to relieve your workload, let alone the massive privacy infringement that the system's very existence represents - by what right are these databases carrying so much PII, comparing against and coming back with reports of such critical nature? by what right, for example, are tools like Turn-it-In, and others, are allowed to keep mine, yours and everyone else's IP, under the guise of plagiarism detection? simply because someone way back when discovered that RE can be used to recognise regular expressions and occurrences? well, guess what: the main goal and  reason for RE matching was created was to compile and process programming languages, by machines; and I don't care what "PhD thesis" is being used to justify the status quo: this is utter madness.

    • Rhonda_Snover
    • 1 mth ago
    • Reported - view

    This year, my students have uploaded approximately 400 different pieces to Turnitin.com and only three have indicated AI. Nearly every student uses Grammarly as that is part of our school offerings. That being said, each time, I ask the student to show me their version history on Google Docs to support the work they did on their essay/poem, etc., and on all three occasions, the student showed no history of work, but rather a doc started at one time and completed within minutes. In one case, I used AI and asked it to write the same piece with certain parameters and it wrote a piece eerily similar to the one turned in.

      • Chushka_chuchka
      • 2 wk ago
      • Reported - view

       Here is the thing: any detector will only analyze the final text and check for words that are used in a specific sequence unlikely for humans and that is all. It will not tell you if the student asked an AI to generate the whole text from scratch or if he used it to rephrase what he already wrote for more clarity. As rephrasing is done by a software and not a human it will most likely be always detected and you will not know if this was just rephrasing or entirely generated. Now the question: is rephrasing a plagiarism ? Also this practice of using Turnitin is ethically not right: back at the time when there was a big doubt that a student essay was written by someone else nobody was going to punish him based only on this assumption without proofs- it was always investigated. Here we are talking about the same thing: high probability but not a definitive proof.

    • Iqra
    • 2 wk ago
    • Reported - view

    The term “controller” has been used to identify different professionals who analyze data and produce information to support decision-making processes. The appearance of controllers in companies is usually related to accounting since accountants have transitioned from recording data for financial analysis to management accounting to support decision-making. Some studies name this role the financial controller while others name it the business unit controller( (Ariel LA PAZ, Daniela GRACIA, 2020). The roles and responsibilities of the top-level professionals and senior managers working as Controllers have changed and adapted, ranging from ideal-named bean counters, scorekeepers, and watchmen to consultants, advisors, and business partners. The emergence of controllers is facilitated by the availability of data and computer technologies, but it is mainly explained by the increasing need to control and maintain the consistency of decisions, actions, resources, and objectives in dynamic business environments. Along with the evolution of the field and the changing organizational dynamics, the skills, tools, and capabilities needed by a professional in charge of management control functions contain ambiguities, complexities, and challenges to be solved by companies and professionals.

    • Dictionary_Tech
    • 2 wk ago
    • Reported - view

    Grammarly offers an AI writing feature that can be flagged by Turnitin if students use it. Although Turnitin claims an AI content detection accuracy of 98%, there is still a 2% chance of false positives. You can check our article about 8 reasons why Turnitin flags students' essays as AI writing

    • GHAZAL_AB
    • 2 wk ago
    • Reported - view

    hi. how can I enable ai detector on Turnitin?

      • Senior Teaching & Learning Innovations Specialist
      • Karen_Smith
      • 2 wk ago
      • Reported - view

       Contact Turnitin to schedule a consultation.

Content aside

  • 6 Likes
  • 2 wk agoLast active
  • 119Replies
  • 39224Views
  • 52 Following