The Philosophers' Cocoon

A safe and supportive forum for philosophers.

Owned & Moderated by Marcus Arvan (University of Tampa)

Contact: marvan@ut.edu

Blog mission & moderation policy

IMPORTANT NOTE on how to ensure anonymity in commenting
Job-Market Reporting Thread (2025-26)
Job-Market Discussion Thread (2026-27)

recent posts

about

Assessing students with an at-home essay without oral exams/software monitoring?

In our new “how can we help you?” thread, a reader writes:

In the paragon of arrested development that is UK higher education, I am required to assess my MA students this autumn like this: 100% of their grade will be an at-home essay of 3,000 words. Essay prompts must be submitted by me in advance. What the hell do I do given the state of LLMs, which are better than the vast majority of my students. I also don’t think there is much room for us to accuse students based on pangram or similar, there has to be hard evidence and the process is an ordeal. I am beyond being outraged by this and now just wondering:
– is there anything i can do to mitigate this more than a sham assessment?
– if it’s not, how should i approach it as a grader? Just mark the best essays with the highest grades regardless of how obviously they are AI-generated (assuming they aren’t catchable).
– how should i explain the assignment to my students? should i just say ‘no AI use’ and we all pretend they won’t? Say ‘we both know you’re going to do it so let’s just focus on the non-assessed parts of the course and learn some philosophy’?
-how would you approach this? Give a spiel about how they’re wasting a great opportunity to expand their minds if they outsource their essay to AI, etc.?

Just to add: I can’t do anything interview the students / do oral exam type audits, and I can’t do anything like ask them to use software that monitors their workflow.

This sounds like a really tough situation given the constraints the OP has to work within (given that we are all in a hard-enough situation with assessing at-home work already).

Do readers have any helpful tips to share?

Posted in ,

14 responses to “Assessing students with an at-home essay without oral exams/software monitoring?”

  1. Anonymous

    I share a lot of those same worries regarding AI. But for the part “just mark the best essays with the highest grades regardless of how obviously they are AI-generated”, I honestly don’t see how an obviously AI-written essay could actually be the best one. I’d love to know what made the OP say that. In my experience, if an essay is totally or mostly written by AI, it’s usually at best just mediocre (assuming we don’t dock points for those obvious AI giveaways).

    Reply
  2. Anonymous

    I don’t think the majority of students who rely on AI are bad intentioned. Instead, I think that there’s just not enough friction, and they wrongly believe that AI will make the results clearly better. Much of AI-written essays are not good, but it just takes a bit more brain power on our parts to figure it out. They are frequently softly inconsistent — not logically inconsistent, but vibes inconsistent. Additionally, I think many students might do the work themselves first, then run it through ChatGPT to correct any mistakes. If they know that the AI isn’t good at philosophy, they might be less inclined to do this. So telling them explicitly that AI sucks and frequently gets worse grades than human-written papers might help.

    As for strategies, I have started to rely on a mixture of friction and self-motivation. First, I say that AI use on the assignments is strictly prohibited. Then, I walk my students through the value of an education and then try to get them to see how it’s embarrassing — shameful — to use AI. It’s something they’ll regret in the future. They had the opportunity to actually learn something and they squandered it for what? For more time scrolling Instagram Reels? Why are they even there at that point?

    For adding friction, I’ve been requiring my students to print out things and do them by hand, then upload the results (no “my printer doesn’t work” excuses). This makes the shame even stronger, to go through the effort of printing something out, and then to go back to your computer and hand-write what the robot tells you. Yuck.

    For your particular class, you may consider requiring students to meet with you about their topic once they have an outline. You’ll be able to suss out AI use fairly easy after that.

    Reply
  3. Aleksei

    I like this suggestion from Anonymous “For your particular class, you may consider requiring students to meet with you about their topic once they have an outline. You’ll be able to suss out AI use fairly easy after that.” and I would do something the lines of that. You mention you are not allowed to change how you grade them, the components of the grade, etc. That is evaluative assessment, and it’s fixed in the books. However, you are allowed to put as many hurdles along the way and introduce friction, what is considered formative assessment. Anything that does not translate into a separate grade is fair game. So add friction as a way of formative assessment. Students know that, by contract, they have to hand in an essay and it’s 100% of their grade. They do not know how that essay is to be produced and where. Are you sure you are not allowed to ask them to write it in class? You can introduce some gates in the essay development: topic has to be approved by you, they need to hand in an outline, they need to present it to the class (or pitch it), they need to justify why this topic and why this angle. You can add as many of these steps along the way and say clearly that, if they fail to do the process-steps, the final grade of the essay starts from a minimum. As long as you discuss with them along the way to check progress, and encourage them to have a voice in the essay, maybe even including peer review from other students and their response to it, the final essay will be more likely their voice. As teacher, you have a lot of agency in how you set the process-rules to achieve that final result, and you can give bonus points or deductions for the process.

    Reply
  4. Anonymous

    I think professors assume too much of the responsibility/burden here. My institution is not as extreme as the example here. But our admin gives zero support for rooting out AI use. In fact, they generally encourage students to use AI.

    Assuring academic integrity is an institutional effort. Why should one professor think that they have to pick up the slack because dozens of others are failing at their jobs?

    My recommendation: Give the high-minded spiel about why they should not use AI. And then just grade everything straight. If students are cheating, it is not your fault. It is the institution’s fault.

    Reply
    1. Anonymous

      I understand the sentiment behind this, but it just genuinely feels bad to be grading essays where most are AI generated (or I believe that a significant portion are AI generated). Regardless of whose fault it is, it is we who are grading the papers.

      Reply
    2. Anonymous

      I also ban AI, give a spiel about why they should not use AI (I mention the dishonesty, but focus on the fact that not doing the work is permanently diminishing their own capacities) and grade more or less straight. (If something is obvious AI, I return it ungraded to be redone. I’ve never gotten the new version back, so the problem takes care of itself.)

      Until the university (or the government) does something to prevent AI cheating, there’s not more that can be reasonably done. Not by me, anyway; I’m not the plagiarism police, in part because I haven’t been given the tools to be detective and judge. At least the essays I am reading are a little better than I used to get! FWIW, my essays are due in three stages: a thesis statement and outline, a draft, and a final version. Stages 2 and 3 MUST show responsiveness to the comments I give on stages 1 and 2. That helps a bit.

      BTW governments could do something. California just required AI companies to attach difficult-to-remove “this is AI” labels on images and video, to combat abusive deepfakes and political propaganda. Something similar could be arranged for essays, with sufficient political will.

      Reply
  5. dazesensationallye39c0dc2cd

    I share your aversion to grading such papers. I wouldn’t do it unless I had a way of enforcing my AI ban.

    What I’ve done in my classes:

    1. Make a self-enforcing essay prompt. For example, assign and teach only parts of an author, excluding the rest, but be a little more general in the essay prompt. Like teaching Kant’s CI but not his FOH, then asking “What would Kant’s argument that we discussed say about X and why…” Or teach Frank Jackson’s (of late) materialism, without teaching “Epiphenomenal Qualia.” The LLMs (and Google) always cast a wider net, including the stuff I know I didn’t teach.

    2. Use terms you coin only in the course, with special meanings you make up, but which have other meanings to AI (and to the world). Anyone whose essay draws the more conventional meaning will simply fail that part of the prompt.

    This has worked rather well for me, albeit with some kinks now and then, which I am working to overcome. Onward and upward with the arms race…

    .

    Reply
    1. Anonymous

      Could you give an example for 2.? I have a hard time coming up with one. Thanks!

      Reply
      1. dazesensationallye39c0dc2cd

        Like if I’m teaching Heidegger, I might describe a certain point – just in this class – as Heidegger’s Being in General argument (as opposed to stuff he says about this being, or that being, say). That is, Heidegger’s B.I.G. argument. When I then ask in the prompt for stuff on “Heidegger’s BIG argument,” only those who paid attention get it right. AI and Google in general (GIG) have no idea, and they’ll make up an answer that’s way off.

        Also, for those tricky Gemini trasncribers in the audience, I don’t say “Big argument” while lecturing. Instead I always say “Being in General,” saving the acronym for the prompt.

        (Just to be clear, this is an example I fashioned for your question; I’ve never taught anything even near the neighborhood (or continent) of Heidegger. But hopefully you get the idea.)

      2. Anonymous

        This is a great suggestion, but OP will need to implement it carefully. I work in the UK, and was told by higher authorities that nothing that puts remote participants at a disadvantage is allowed to be in the exam/final assignment.

  6. Anonymous

    I have much sympathy for OP, as I also work for a UK institution. Some thoughts

    – Reflecting something already said, if you can, formulate assignments, something like “1000 word” essay outline 1 month before the official deadline.

    – I only briefly mention the problem of outsourcing one’s thinking, and I don’t think it has any purchase whatsoever on students who just want a certificate. Unfortunately, I think UK higher-ed is full of such students. But for the students who care, saying something very brief should suffice.

    – DON’T waste your time fighting admins. They probably don’t know they are destroying your institution’s reputation in the long run, but even if they do, they won’t care. When your institution’s is eventually destroyed, they have already retired or moved on to a higher paid job, and will be on track to destroy another institution. Instead, try to find a better institution, and apply.

    – Word will get out, and reputation will eventually catch up, but not any time soon. So you have plenty of time to apply before admission committees and employers find out that the degree is worth nothing. And it will take some further delay before prospective students find out.

    – It is not your responsibility to save everyone, especially those who don’t want to be saved. Save yourself instead.

    Reply
    1. Anonymous

      Totally agree with this take. Asychonrous online education has become a joke. The entire model needs to be overhauled. It would be one thing if we could require students taking these courses to go to in-person testing centers to take secure proctored exams. But we can’t (at least not at my institution). File under: Not My Problem.

      Reply
  7. Anonymous

    Just to add, if you dislike your employer, spread the word. I think it will lead to the degree being disvalues in the long run. (Though it is also possible that in the short term, it will boost enrolment from those who just want a degree.)

    Reply
  8. Anonymous

    Just a quick question here, and it might be naive, but in my experience, it seem that the MA students are not the ones using AI to write their essays, it generally tends to be undergraduates who are using it in fairly egregious ways. MA students, who are choosing to do an MA in philosophy at my institution are doing certain subjects or courses by choice and interest. They tend to use AI to help figure out how to phrase certain points, but very rarely (if at all) to actually write their work or do the thinking for them.

    Are others finding that MA students are using AI in problematic ways and pervasively so? If so, this is a massive surprise to me and doesn’t align with my experience at all.

    Reply

Leave a Reply

Discover more from The Philosophers' Cocoon

Subscribe now to keep reading and get access to the full archive.

Continue reading