Fighting Fire with Fire: Scalable Oral Exams
behind-the-enemy-lines.com
behind-the-enemy-lines.com
[...]
> Take-home exams are dead. Reverting to pen-and-paper exams in the classroom feels like a regression.
Yeah, not sure the conclusion of the article really matches the data.
Students were invited to talk to an AI. They did so, and having done so they expressed a clear preference for written exams - which can be taken under exam conditions to prevent cheating, something universities have hundreds of years of experience doing.
I know some universities started using the square wheel of online assessment during covid and I can see how this octagonal wheel seems good if you've only ever seen a square wheel. But they'd be even better off with a circular wheel, which really doesn't need re-inventing.
And they didn't even bother to test the most important thing. Were the LLM evaluations even accurate! Have graders manually evaluate them and see if the LLMs were even close or were wildly off.
This is clearly someone who had a conclusion to promote regardless of what the data was going to show.
https://i.imgur.com/EshEhls.png
When someone at that level pretends to not understand it, there is no way to mince words.
This is malice.
https://elevenlabs.io/app/talk-to?agent_id=agent_8101k9d1pq4...
Also, given that there's so many ways for LLMs to go off the rails (it just gave me the student id I was supposed to say, for example), it feels a bit unprofessional to be using this to administer real exams.
That does not resemble any good professor I've ever heard. It's very aggressive and stern, which is not generally how oral exams are conducted. Feels much more like I'm being cross examined in court.
This is not true; the professor and the TAs graded every student submission. See this paragraph from the article:
(Just in case you are wondering, I graded all exams myself and I asked the TA to also grade the exams; we mostly agreed with the LLM grades, and I aligned mostly with the softie Gemini. However, when examining the cases when my grades disagreed with the council, I found that the council was more consistent across students and I often thought that the council graded more strictly but more fairly.)
Having said that, LLMs can be good tutors if used correctly.
At least in Germany, if there are only 36 students in a class, usually oral exams are used because in this case oral exams are typically more efficient. For written exams, more like 200-600 students in a class is the common situation.
If you are going to set an exam that can be graded in 5-10 min, you are not getting a lot of signal out of it.
I wanted to do oral exams, but they are much more exhausting for the prof. Nominally, each student is with you for 30 min, but (1) you need to think of slightly different question for each student (2) you need to squeeze all the exams in only a couple of days to avoid giving later students too much extra time to prepare.
That's entirely false; this is why we have multiple-choice tests.
Thinking deeper, though, multiple choice tests require SIGNIFICANTLY more preparation. I would go so far as to say almost all individual professors are completely unqualified to write valid multiple choice tests.
The time investment in multiple choice comes at the start - 12 hours writing it instead of 12 hours grading it - but it’s still a lot of time and frankly there is only very general feedback on student misunderstandings.
I don't believe that your argument is more than an ad-hoc value judgment lacking justification. And it's obvious that if you think so little of your colleagues, that they would also struggle to implement AI tests.
- They have a base marks of 20-25% (by random guessing) instead of 0.
- You never see the working. So you can't check if students are thinking correctly. Slightly wrong thinking can get you right answers.
- They don't even remotely reflect real life at all. Written worked through problems on the other hand - I still do those in my professional life as a scientist all the time. It's just that I am setting the questions for myself.
- The format doesn't allow for extended thought questions.
In my undergrad, I had some excellent profs who would set long work through exam question in such a way that you learned something even in the exams. Simply a joy taking those exams that gave a comprehensive walk through of the course. As a prof, I have always tried to replicate that.
Right on point. I find particularly striking how little is said about whether the best students achieve the best grades. Authors are even candid that different LLMs asses differently, but seem to conclude that LLMs converging after a few rounds of cross reviews indicate they are plausible so who cares. The apparences are safe.
The objections to SFH exist and are strikingly similar to objections to WFH, but the economics are different. Some universities already see value in offering that option, and they (of course) leave it to the faculty to deal with the consequences.
Even for distance education though, proctored testing centers have been around longer than the internet.
It is about a third of the students I teach, which amounts to several hundreds per term. It may be niche, but it is not insignificant, and definitely a problem for some of us.
> Even for distance education though, proctored testing centers have been around longer than the internet.
I don't know how much experience you have with those. Mine is extensive enough that I have a personal opinion that they are not scalable (which is the focus of the comment I was replying to). If you have hundreds of students disseminated around the world, organising a proctored exam is a logistical challenge.
It is not a problem at many universities yet, because they haven't jumped on the bandwagon. However domestic markets are becoming saturated, visas are harder to get for international students, and there is a demand for online education. I would be surprised that it doesn't develop more in the near future.
I think the end result though is that schools either limit their students to a smaller number of locations where they can have proctored exams, or they don’t and they effectively lose their credentialing value.
In 1789 there were 1,000 enrolled college students total, in a country of 2.8M. In 2025, it is 19M students in a country of 340M. https://educationalpolicy.org/wp-content/uploads/2025/11/251...
In 1950, 5.5% of adults ages 25-34 had completed a 4 year college degree. In 2018, it was 39%. https://www.highereddatastories.com/2019/08/changes-in-educa...
With attendance increasing at this rate (not to mention the exploding costs of tuition), it seems possible that the methods need to change as well.
Work study and TA jobs were abundant when I was in college. It wasn't a problem in the past and shouldn't be a problem now.
One of these is not like the others.
Some people dream that technology (preferably duly packaged by for-profit SV concerns) can and will eventually solve each and every problem in the world; unfortunately what education boils down to is good, old-fashioned teaching. By teachers. Nothing whatsoever replaces a good, talented, and attentive teacher, all the technologies in the world, from planetariums to manim, can only augment a good teacher.
Grading students with LLMs is already tone-deaf, but presenting this trainwreck of a result and framing it as any sort of success... Let's just say it reeks of 2025.
If a student is willing and desire to learn, an LLM is better than a bad teacher.
If a student doesn't want to learn, and is instead being forced to (either as a minor, or via certification required to obtain work & money), then they have every incentive to cheat. An LLM is insufficient in this case - a teacher is both the enforcer and the tutor in this case.
There's also nothing wrong with a teacher using an LLM to help with the grading imho.
> In our new "AI/ML Product Management" class, the "pre-case" submissions (short assignments meant to prepare students for class discussion) were looking suspiciously good. Not "strong student" good. More like "this reads like a McKinsey memo that went through three rounds of editing," good...Many students who had submitted thoughtful, well-structured work could not explain basic choices in their own submission after two follow-up questions. Some could not participate at all...Oral exams are a natural response. They force real-time reasoning, application to novel prompts, and defense of actual decisions. The problem? Oral exams are a logistical nightmare. You cannot run them for a large class without turning the final exam period into a month-long hostage situation.
Written exams do not do the same thing. You can't say 'just do a written exam'. So sure, the students may prefer them, but so what? That's apples and oranges.
Some universities and professors have tried to move to a take-home exam format, which allows for more comprehensive evaluation with easier logistics than a too-brief in-class exam or an hours-long outside-of-class sitting where unreasonable expectations for mental and sometimes physical stamina are factors. That "take-home exams are dead" is self-evident, not a result of the experiment in the article. There used to be only a limited number of ways to cheat at a take-home exam, and most of them involved finding a second person who also lacked a moral conscience. Now, it's trivial to cheat at a take-home exam all by yourself.
You also mentioned the hundreds of years of experience universities have at traditional written exams. But the type and manner of knowledge and skills that must be tested for vary dramatically by discipline, and the discipline in question (computer science / software engineering) is still new enough that we can't really say we've matured the art of examining for it.
Lastly, I'll just say that student preference is hardly the way to measure the quality of an exam, or much of anything about education.
Did I say "conclusion" ? Sorry, I should have said the section just before the acknowledgements, where the conclusion would normally be, entitled "The bigger point"
That is, the author concluded that AI tools provide viable alternatives to the other available options, and which solve many of their problems.
Have exams ever been about humanity and the optics of it ?
In addition to being non-deterministic LLMs can product vastly different output from very slightly different input.
That’s ignoring how vulnerable LLMs are to prompt injection, and if this becomes common enough that exams aren’t thoroughly vetted by humans, I expect prompt attacks to become common.
Also if this is about avoiding in person exams, what prevents students from just letting their AI talk to test AI.
They mention getting 100% agreement between the LLMs on some questions and lower rates on other, so if an exam was composed of only questions where there is near 100% convergence, we'd be pretty close to a stable state.
I agree it would be reassuring to have a human somewhere in the loop, or perhaps allow the students to appeal the evaluation (at cost?) if they is evidence of a disconnect between the exam and the other criteria. But depending on how the questions and format is tweaked we could IMHO end up with something reliable for very basic assessments.
PS:
> Also if this is about avoiding in person exams, what prevents students from just letting their AI talk to test AI.
Nothing indeed. The arms race hasn't started here, and will keep going IMO.
Humans are incredibly good at solving problems, but while one person is solving 'how do we prevent students from cheating' a student is thinking 'how I bypass this limitation preventing me from cheating'. And when these problems are digital and scalable, it only takes one student to solve that problem for every other student to have access to the solution.
So the whole thing is a complete waste of time then as an evaluation exercise.
>council of AIs
This only works if the errors and idiosyncrasies of different models are independent, which isn’t likely to be the case.
>100% agreement
When different models independently graded tests 0% of grades matched exactly and the average disagreement was huge.
They only reached convergence on some questions when they allowed the AIs to deliberate. This is essentially just context poisoning.
1 model incorrectly grading a question will make the other models more likely to incorrectly grade that question.
If you don’t let models see each other’s assessments, all it takes is one person writing an answer in a slightly different way that causes disagreement among models to vastly alter the overall scores by tossing out a question.
This is not even close to something you want to use to make consequential decisions.
Appeals aren't a solution either, because students won't appeal (or possibly even notice) a small bias given the variability of all the other factors involved, nor can it be properly adjucated in a dispute.
If the goal is to assess whether a student properly understood the work they submitted or more generally if they assimilated most concepts of a course, the evaluation can have a bar low enough for let's say 90% of the student to easily pass. That would give enough of margin of error to account for small biases or misunderstandings.
I was comparing to mark sheet tests as they're subject to similar issues, like students not properly understanding the wording (and usually the questions and answer have to be worded in pretty twisted ways to properly) or straight checking the wrong lines or boxes.
To me this method, and other largely scalable methods, shouldn't be used for precise evaluations, and the teachers proposing it also seem to be aware of these limitations.
Such classes do not have the luxury of pen-and-paper exams, and asking people to go to testing centers is a huge overkill.
Take home exams for such settings (or any other form of written exam) are becoming very prone to cheating, just because the bar to cheating is very low. Oral exams like that make it a bit harder to cheat. Not impossible, but harder.
When I was a student, I would have been quite vocal with my clear preferences for all exams being open-book and/or being able to amend my answers after grading for a revised score.
What I'm saying is, "the students would prefer..." isn't automatically case closed on what's best. Obviously the students would prefer a take-home because you can look up everything you can't recall / didn't show up to class to learn, and yes, because you can trivially cheat with AI (with a light rewrite step to mask the "LLM voice").
But in real life, people really will ask you to explain your decisions and to be able to reason about the problem you're supposedly working on. It seems clear from reading the revised prompts that the intent is to force the agent to be much fairer and easier to deal with than this first attempt was, so I don't think this is a bad idea.
Finally, (this part came from my reading of the student feedback quotes in the article) consider that the current cohort of undergrads is accustomed to communicating mainly via texting. To throw in a further complication, they were around 13-17 when COVID hit, decreasing human contact even more. They may be exceedingly nervous about speaking to anyone who isn't a very close friend. I'm sympathetic to them, but helping them overcome this anxiety with relatively low stakes is probably better than just giving up on them being able to communicate verbally.
How do you expect that to work? After the exam, you talk to your friends (and to ChatGPT) and know the correct answers even if you could have never produced them during the exam.
This was pre-LLM, but you could cheat back then too. LLMs make it a bit easier by showing you the work to "show" on your corrections.
I went to school long before LLMs were even a Google Engineer's brianfart for the transformer paper and the way I took exams was already AI proof.
Everything hand written in pen in a proctored gymnasium. No open books. No computers or smart phones, especially ones connected to the internet. Just a department sanctioned calculator for math classes.
I wrote assembly and C++ code by hand, and it was expected to compile. No, I never got a chance to try to compile it myself before submitting it for grading. I had three hours to do the exam. Full stop. If there was a whiff of cheating, you were expelled. Do not pass go. Do not collect $200.
Cohorts for programs with a thousand initial students had less than 10 graduates. This was the norm.
You were expected to learn the gd material. The university thanks you for your donation.
I feel like i'm taking crazy pills when I read things about trying to "adapt" to AI. We already had the solution.
It is a sad world we live in.
Also, IMO oral examinations are quite powerful for detecting who is prepared and who isn't. On the down side they also help the extroverts and the confident, and you have to be careful about preventing a bias towards those.
This is true, but it is also why it is important to get an actual expert to proctor the exam. Having confidence is good and should be a plus, but if you are confident about a point that the examiner knows is completely incorrect, you may possibly put yourself in an inescapable hole, as it will be very difficult to ascertain that you actually know the other parts you were confident (much less unconfident) in.
> I wrote assembly and C++ code by hand, and it was expected to compile. No, I never got a chance to try to compile it myself before submitting it for grading.
Do you, like, really think this is the best way to assess someone's ability? Can't we find a place between the two extremes?
Personally, I'd go with a school-provided computer with a development environment and access to documentation. No LLMs, except maybe (but probably not) for very high-level courses.
Lots of my tests involved writing pseudocode, or "Just write something that looks like C or Java". Don't miss the semicolon at the end of the line, but if you write "System.print()" rather than "System.out.printLn()" you might lose a single point. Maybe.
If there were specific functions you need to call, it would have a man page or similar on the test itself, or it would be the actual topic under test.
I hand wrote a bunch of SQL queries. Hand wrote code for my Systems Programming class that involved pointers. I'm not even good with pointers. I hand wrote Java for job interviews.
It's pretty rare that you need to actually test someone can memorize syntax, that's like the entire point of modern development environments.
But if you are completely unable to function without one, you might not know as much as you would hope.
The first algorithms came before the first programming languages.
Sure, it means you need to be able to run the code in your head and be able to mentally "debug" it, but that's a feature
If you could not manage these things, you washed out in the CS101 class that nearly every STEM student took. The remaining students were not brilliant, but most of them could write code to solve problems. Then you got classes that could actually teach and test that problem solving itself.
The one class where we built larger apps more akin to actual jobs, that could have been done entirely in the lab with locked down computers if need be, but the professor really didn't care if you wanted to fake the lab work, you still needed to pass the book learning for "Programming Patterns" which people really struggled with and you still needed to be able to give a "Demo" and presentation, and you still needed to demonstrate that you understood how to read some requests from a "Customer" and turn it into features and requirements and UX
Nobody cares about people sabotaging their own education except in programming because no matter how much MBAs insist that all workers are replaceable, they cannot figure out a way to actually evaluate the competency of a programmer without knowing programming. If an engineer doesn't actually understand how to evaluate static stresses on a structure, they are going to have a hard time keeping a job. Meanwhile in the world of programming, hopping around once a year is "normal" somehow, so you can make a lot of money while literally not knowing fizzbuzz. I don't think the problem is actually education.
Computer Science isn't actually about using a laptop.
This applies to prose as much as code. A computer completely changes the experience of writing, for the better.
Yes, obviously people made do with analog writing for hundreds of years, yadda yadda, I still think it's a stupid restriction.
And why is this a flex exactly? Almost sounds like fraud. Get sold on how you'll be taught well and become successful. Pay. Then be sent through an experience that filters so severely, only 1% of people pass. Receive 100% of the blame when you inevitably fail. Repeat for the other 990 students. The "university thanks you for your donation" slogan doesn't sound too hot all of a sudden.
It's like some malicious compliance take on both teaching and studying. Which shouldn't even be surprising, considering the circumstances of the professors e.g. where I studied, as well as the students'.
Mind you, I was (for some classes) tested the same way. People still cheated, and grading stringency varied. People still also forgot everything shortly after wrapping up their finals on the given subjects and moved on. People also memorized questions and compiled a solutions book, and then handed them down to next year's class. Because this method does jack against that on its own. You still need to keep crafting novel questions, vary them more than just by swapping key values, etc.
Do you think you're just purchasing a diploma? Or do you think you're purchasing the opportunity to gain an education and potential certification that you received said education?
It's entirely possible that the University stunk at teaching 99% of it's students (about as equally possible that 99% of the students stunk at learning), but "fraud" is absolute nonsense. You're not entitled to a diploma if you fail to learn the material well enough to earn it.
If teaching was so simple that you could just tell people to go RTFM then recite it from memory, I don't know why people are bothering with pedagogy at all. It'd seem that there's more to teaching and learning than the bare minimum, and that both parties are culpable. Doesn't sound like you disagree on that either.
> you're purchasing the opportunity to
We can swap out fraud for gambling if you like :) Sounds like an even closer analogy now that you mention!
Jokes aside though, isn't it a gamble? You gamble with yourself that you can [grow to] endure and succeed or drop out / something worse. The stake is the tuition, the prize is the diploma.
Now of course, tuition is per semester (here at least, dunno elsewhere), so it's reasonable to argue that the financial investment is not quite in such jeopardy as I painted it. Not sure about the emotional investment though.
Consider the Chinese Gaokao exam, especially in its infamous historical context between the 70s and 90s. The number of available seats was way lower than the number of applications [0]. The exams grueling. What do you reckon, was it the people's fault for not winning an essentially unspoken lottery? Who do you think received the blame? According to a cursory search, the individual and their families (wasn't there, cannot know) received the blame. And no, I don't think in such a tortured scheme it is the students' fault for not making the bar.
If there are fewer seats than what there is demand for, then that's overbooking, and you the test authoring / conducting authority are biased to artificially induce test failures. It is no longer a fair assessment, nor a fair dynamic. Conversely, passing is no longer an honest signal of qualification. Or rather, not passing is no longer an honest signal of unqualification. And this doesn't have to come from a single test, it can be implemented structurally too, so that you shed people along the way. Which is what I'm actually alluding to.
[0] ~4.8%, so ~95% of people failed it by design: https://en.wikipedia.org/wiki/Class_of_1977%E2%80%931978_%28...
I do not! A situation where roughly 1% of the class is passing suggests that some part of the student group is failing, and also that there is likely a class design issue or a failure to appropriately vet incoming students for preparedness (among, probably, numerous other things I'm not smart enough to come up with).
And I did take issue with the "fraud" framing; apologies for not catching your tone! I think there is a chronic issue of students thinking they deserve good grades, or deserve a diploma simply for showing up, in social media and I probably read that into your comment where I shouldn't have.
> Jokes aside though, isn't it a gamble?
Not at all. If you learn the material, you pass and get a diploma. This is no more a gamble than your paycheck. However, I think that also presumes that the university accepts only students it believes are capable of passing it's courses. If you believe universities are over-accepting students (and I think the evidence says they frequently are not, in an effort to look like luxury brands, though I don't have a cite at hand), then I can see thinking the gambling analogy is correct.
Yeah, that's fine, I can definitely appreciate that angle too.
As you can probably surmise, I've had quite some struggles during my college years specifically, hence my angle of concern. It used to be the other way around, I was doing very well prior to college, and would always find people's complaints to be just excuses. But then stuff happened, and I was never really the same. The rest followed.
My personal sob story aside, what I've come to find is that while yes, a lot of the things slackers say are cheap excuses or appeals to fringe edge-cases, some are surprisingly valid. For example, if this aforementioned 99% attrition rate is real, that is very very suspect. Worse still though, I'd find things that people weren't talking about, but were even more problematic. I'll have to unfortunately keep that to myself though for privacy reasons [0] [1].
Regarding grading, I find grade inflation very concerning, and I don't really see a way out. What affects me at this point though is certifications, and the same issue is kind of present there as well. I have a few colleagues who are AWS Certified xyz Engineers for example, but would stare at the AWS Management Console like a deer in the headlights, and would ask exceedingly stupid questions. The "fee extraction" practice wouldn't be too unfamiliar for the certification industry either - although that one doesn't bother me much, since I don't have to pay for these out of my own pocket, thankfully.
> If you learn the material, you pass and get a diploma. This is no more a gamble than your paycheck
I'd like to push back on this just a little bit. I'm sure it depends on where one lives, but here you either get your diploma or tough luck. There are no partial credentials. So while you can drop out (or just temporarily suspend your studies) at the end of semester, there's still stuff on the line. Not so much with a paycheck. I guess maybe a promotion is a closer analog, depending on how a given company does it (vibes vs something structured). This is further compounded by the social narrative, that if you don't get a degree then xyz, which is also not present for one's next monthly paycheck.
[0] What I guess I can mention is that I generally found the usual cycle of study season -> exam season to be very counter-productive. In general, all these "building up hype and then releasing it all at once" type situations were extremely taxing, and not for the right reasons. I think it's pretty agreeable at least that these do not result in good knowledge retention, do not inspire healthy student engagement, nor are actually necessary. Maybe this is not even a thing in better places, I don't know.
[1] I have absolutely no training in psychology or pedagogy, so take this with a mountain of salt, but I've found that people can be not just uninterested in learning, but grow downright hostile to it, often against their own self-recognized best interests. I've experienced it on myself, as well as seen it with others. It can be very difficult to snap someone out of such a state, and I have a lingering suspicion that it kind of forms a pipeline, with the lack of interest preceding it. I'm not sure that training and evaluating people in such a state results in a reasonable assessment, not for them, nor for the course they're taking.
Colleges exist to collect tuition, especially from international students who pay more. Teaching anything at all, or punishing cheating, just isn’t that important.
You could easily raise the bar without sacrificing quality of education (and likely you'd improve it just from the improvement in student:teacher ratio).
In another European country, schools get paid for students that passed.
Perhaps lifetimerubyist means "1000 students took the mandatory philosophy and ethics 101 class, but only 10 graduated as philosophy majors"
If it is, I'd be fascinated to learn more.
I mean, the logistics would be pretty wild - even a large university's largest lecture theatres might only have 500 seats. And they'd only have one or two that large. It'd be expensive as hell to build a university that could handle multiple subjects each admitting over a thousand students.
That's quite a high non-completion rate - but it's nowhere near 99%.
[1] https://nieuws.kuleuven.be/en/content/2023/42-6-of-new-stude...
My experience is the same except I think ~50% or so graduated[0].
[0]: Disclaimer that my programme was pretty competitive to get into, which is an earlier filter. Statistics looked worse for programmes at similar level with less people applying.
For comparison we had lengthy sessions in a jailed terminal, week after week, writing C programs covering specific algorithms, compiling and debugging them within these sessions and assistants would follow our progress and check we're getting it. Those not finishing in time get additional sessions.
Last exam was extremely simple and had very little weight in the overall evaluation.
That might not scale as much, but that's definitely what I'd long for, not the Chuck Norris style cram school exam you are drawing us.
The old ways do not scale well once you pass a certain number of students.
You have a very weird idea of education if a teaching method that results in a 99% failure rate is seen as good by yourself. Do you imagine a professional turning out work that was 99% suboptimal?
Don't tell me about GenZ. I had oral exams in calculus as undergrad, and our professor was intimidating. I barely passed each time when I got him as examiner, though I did reasonably well when dealing with his assistant. I could normally keep my emotions in check, but not with my professor. Though, maybe in that case the trigger was not just the tone of professor, but the sheer difference in the tone he used normally (very friendly) and at the exam time. It was absolutely unexpected at my first exam, and the repeated exposure to it didn't help. I'd say it was becoming worse with each time. Today I'd overcome such issues easily, I know some techniques today, but I didn't when I was green.
OTOH I wonder, if an AI could have such an effect on me. I can't treat AI as a human being, even if I wanted to, it is just a shitty program. I can curse a compiler refusing to accept a perfectly valid borrow of a value, so I can curse an AI making my life difficult. Mostly I have another emotional issue with AI: I tend to become impatient and even angry at AI for every small mistake it does, but this one I could overcome easily.
I wish that wasn't a thing.
Interviews are similar, but different: I'm presenting myself.
In my graduate studies in Germany, most of my courses used oral exams. It's fine, and it's battle-tested.
Just like vote-counting, testing students is perfectly scalable without anything but teachers. But: In Europe, I have witnessed oral exams at the Matura, and at the final Diploma test. In the US, I understand all PhDs need a oral defense session.
To me, this mindset of delegating to AI because of laziness is perfectly embodied in "Experimenta Felicitologica" (sp?) By Stanislaw Lem.
AI is great when performing somewhat routine tasks, but for anything inherently adversarial, I'm skeptical we'll soon see good solutions. Building defeating AIs is just too inexpensive.
I wonder what that means for AI warfare.
This is a summary of sorts:
"Trurl, having decided to make the entire Universe happy, first sat down and developed a General Theory of All-Possible Happiness... Eventually, however, Trurl grew weary of the work. To speed things up, he built a great computer and provided it with a programmatic duplicate of his own mind, that it might conduct the necessary research in his stead.
But the machine, instead of setting to work, began to expand. It grew new stories, wings, and outbuildings, and when Trurl finally lost his patience and commanded it to stop building and start thinking, the machine—or rather, the Trurl-within-the-machine—replied that it couldn't possibly think yet, for it still didn't have enough room. It claimed it was currently housing the Sub-Trurls—specialized programs for General Felicitology, Experimental Hedonistics, and Happiness-Machine-Building—who were currently occupied with their quarterly reports.
The 'Clone-Trurl' told him marvelous tales of the results these sub-Trurls had already achieved in their digital simulations. Trurl, however, soon discovered that these were all cut from the same cloth of lies; not a single sub-Trurl existed, no research had been done, and the machine had simply been using its processing power to enjoy itself and expand its own architecture. In a fit of rage, Trurl took a hammer to the machine and for a long time thereafter gave up all thought of universal happiness."
It's a great allegory. A real shame there is no english translation.
When I was doing a lot of hiring we offered the option (don’t roast me, it was an alternative they could choose if they wanted) of a take-home problem they could do on their own. It was reasonably short, like the kind of problem an experienced developer could do in 10-15 minutes and then add some polish, documentation, and submit it in under an hour.
Even though I told candidates that we’d discuss their submission as part of the next step, we would still get candidates submitting solutions that seemed entirely foreign to them a day later. This was on the cusp of LLMs being useful, so I think a lot of solutions were coming from people’s friends or copied from something on the internet without much thought.
Now that LLMs are both useful and well known, the temptation to cheat with them is huge. For various reasons I think students and applicants see using LLMs as not-cheating in the same situations where they wouldn’t feel comfortable copying answers from a friend. The idea is that the LLM is an available tool and therefore they should be able to use it. The obvious problem with that argument is that we’re not testing students or applicants on their abilities to use an LLM, we’re using synthetic problems to explore their own skills and communication.
Even some of the hiring managers I know who went all in on allowing LLMs during interviews are changing course now. The LLM-assisted interviewed were just turning into an exercise of how familiar the candidate was with the LLM being used.
I don’t really agree with some of the techniques they’re using in this article, but the problem they’re facing is very real.
You've piqued my interest!
Where do we go from there? At some point soon I think this is going to have to come firmly back to real people.
Next steps are bone conduction microphones, smart glasses, earrings...
And the weeding out of anyone both honest and with social anxiety.
As an aside, I'm surprised oral exams aren't possible at 36 students. I feel like I've taken plenty of courses with more participants and oral exams. But the break even point is probably very different from country to country.
I think the most I experienced at the physics department in Aarhus was 70ish students. 200 sounds like a big undertaking.
this is also known as 'logistical nightmare', but yeah it's the only reasonable way if you want to avoid being questioned by robots.
It depends on how frequent and how in-depth you want the exams to be. How much knowledge can you test in an oral exam that would be similar to a two-hour written exam? (Especially when I remember my own experience where I would have to sketch ideas for 3/4th of the time alloted before spending the last 1/4th writing frenetically the answer I found _in extremis_).
If I were a teacher, my experience would be to sample the students. Maybe bias the sample towards students who give wrong answers, but then it could start either a good feedback loop ("I'll study because I don't want to be interrogated again in front of the class") or a bad feedback loop ("I am being picked on, it is getting worse than I can improve, I hate this and I give up")
> And here is the delicious part: you can give the whole setup to the students and let them prepare for the exam by practicing it multiple times. Unlike traditional exams, where leaked questions are a disaster, here the questions are generated fresh each time. The more you practice, the better you get. That is... actually how learning is supposed to work.
They're even more possible if you do an oral exam only on the highest grades. That's the purpose, isn't it? To see if a good, very good, or excellent student actually knows what they're talking about. You can't spare 10 minutes to talk to each student scoring over 80% or something? Please
If you're looking for suggestions, I'd love for you to start with a problem that isn't trivially fixable.
If this is the only way to keep the existing approach working, it feels like the only real solution for education is something radically different, perhaps without assessment at all
I did, however, pepper my answers with statements like "it is widely accepted that the industry standard for this concept is X". I would feel bad lying to a human, but I feel no such remorse with an AI.
No, we do not want to eliminate the pen and paper exam. It works well. We use it.
The oral exam is yet another tool. Not a solution for everything.
In our case, we wanted to ensure that the students who worked on the team project: (a) contributed enough to understand the project, (b) actually understood their own project and did not rely solely on an LLM. (We do allow them to use LLMs, it would be stupid not to.)
The students who did badly in the oral exam were exactly the students who we expected to do badly in the exam, even though they aced their (team) project presentations.
Could we do it in person? Sure, we could schedule personalized interviews for all the 36 students. With two instructors, it would have taken us a couple of days to go through. Not a huge deal. At 100 students and one instructor, we would have a problem doing that.
But the key reason was the following: research has shown that human interviewers are actually worse when they get tired, and that AI is actually better for conducting more standardized and more fair interviews. That result was a major reason for us to trust a final exam on a voice agent.
I'm not sure why you're saying this so confidently. Using LLMs on school work is like using a forklift at the gym. You'll technically finish the task you set out to do, and it will be much easier. So why not use a forklift at the gym?
>But the key reason was the following: research has shown that human interviewers are actually worse when they get tired, and that AI is actually better for conducting more standardized and more fair interviews. That result was a major reason for us to trust a final exam on a voice agent.
I think that in an "AI class" for MBA students, the material is probably not complex enough to require much more than a Zork interpreter, but if you tried this on something in which nuance is required, that comparison would change dramatically. For something like this, which is likely going to be little more than knowledge spot checks to catch the most blatant cheaters, why not just have students do multiple choice questions at a kiosk?
For the use of LLM in classes: I understand the reasoning, but I found LLMs to be extremely educational for parsing through dense material (eg parsing an NTSB report for an Uber self-driving crash). Prohibiting students from using LLMs would be counterproductive.
But I still want students to use LLMs responsibly, hence the oral exam.
In my BSc and MSc we were all basically locals who are in all aspects about the same except from the aptitude to study. In the university where I did my PhD there were much more divisions (aka diversity) in which every oral examiner would need to navigate so one group does not feel to be made preferential over another.
CFR 46.104 (Exempt Research):
46.104.d.1 "Research, conducted in established or commonly accepted educational settings, that specifically involves normal educational practices that are not likely to adversely impact students' opportunity to learn required educational content or the assessment of educators who provide instruction. This includes most research on regular and special education instructional strategies, and research on the effectiveness of or the comparison among instructional techniques, curricula, or classroom management methods."
https://www.ecfr.gov/current/title-45/subtitle-A/subchapter-...
So while this may have been a dick move by the instructors, it was probably legal.
> Exempt human subjects research is a specific sub-set of “research involving human subjects” that does not require ongoing IRB oversight. Research can qualify for an exemption if it is no more than minimal risk and all of the research procedures fit within one or more of the exemption categories in the federal IRB regulations. *Studies that qualify for exemption must be submitted to the IRB for review before starting the research. Pursuant to NU policy, investigators do not make their own determination as to whether a research study qualifies for an exemption — the IRB issues exemption determinations.* There is not a separate IRB application form for studies that could qualify for exemption – the appropriate protocol template for human subjects research should be filled out and submitted to the IRB in the eIRB+ system.
Most of my research is in CS Education, and I have often been able to get my studies under the Exempt status. This makes my life easier, but it's still a long arduous paperwork process. Often there are a few rounds to get the protocol right. I usually have to plan studies a whole semester in advance. The IRB does NOT like it when you decide, "Hey I just realized I collected a bunch of data, I wonder what I can do with it?" They want you to have a plan going in.
[1] https://irb.northwestern.edu/submitting-to-the-irb/types-of-...
Imagine otherwise: a teacher who wants change their final exam from a 50 item Scantron using A-D choices, to a 50 item Scantron using A-E choices, because they think having 5 choices per item is better than 4, would need to ask for IRB approval. That's not feasible, and is not what happens in the real world of academia.
It is true that local IRBs may try to add additional rules, but the NU policy you quote talks about "studies". Most IRBs would disagree that "professor playing around with grading procedures and policies" constitutes a "study".
It would be presumed exempted.
Are you a teacher or a student? If you are a teacher, you have wide latitude that a student researcher does not.
Also, if you are a teacher, doing "research about your teaching style", that's exempted.
By contrast, if you are a student, or a teacher "doing research" that's probably not exempt and must go through IRB.
Reminder: This professor's school costs $90k a year, with over $200k total cost to get an MBA. If that tuition isn't going down because the professor cut corners to do an oral exam of ~35 students for literally less than a dollar each, then this is nothing more than a professor valuing getting to slack off higher than they value your education.
>And here is the delicious part: you can give the whole setup to the students and let them prepare for the exam by practicing it multiple times. Unlike traditional exams, where leaked questions are a disaster, here the questions are generated fresh each time. The more you practice, the better you get. That is... actually how learning is supposed to work.
No, students are supposed to learn the material and have an exam that fairly evaluates this. Anyone who has spent time on those old terrible online physics coursework sites like Mastering Physics understands that grinding away practicing exams doesn't improve your understanding of the material; it just improves your ability to pass the arbitrary evaluation criteria. It's the same with practicing leetcode before interviews. Doing yet another dynamic programming practice problem doesn't really make you a better SWE.
Minmaxing grades and other external rewards is how we got to the place we're at now. Please stop enshittifying education further.
Also, with all the progress in video gen, what does recording the webcam really do?
Ask about any teacher, scalability is a serious issue. Students being in classes above and under their level is a serious issue. non-interactive learning, leading to rote memorization, as a result of having to choose scaling methods of learning is a serious issue. All these can be adjusted to a personal level through AI, it's trivial to do so, even.
I'm definitely not sold on the idea of oral exams through AI though. I don't even see the point, exams themselves are specifically an analysis of knowledge at one point in time. Far from ideal, we just never got anything better, how else can you measure a student's worth?
Well, now you could just run all of that student's activity in class through that AI. In the real world you don't know if someone is competent because you run an exam, you know if he is competent because he consistently shows competency. Exams are a proxy for that, you can't have a teacher looking at a student 24/7 to see they know their stuff, except now you can gather the data and parse it, what do I care if a student performs 10 exercises poorly in a specific day at a specific time if they have shown they can do perfectly well, as can be ascertained by their performance the past week?
But isn’t the whole point of a class to move from incompetent to competent?
Isn’t the poor performance on those exercises also part of their overall performance? Do you mean just that their positive work outweighs the bad work?
This sounds as though it was written by an LLM too.
On the other hand, can an AI exam really simulate the conditions necessary for improving at this skill? I think this is unlikely. The students' responses indicate not a general lack of expertise in oral communication but also a discomfort with this particular environment. While the author is making steps to improve the environment, I think it is fundamentally too different from actual human-to-human discussion to test a student's ability in oral communication. Even if a student could learn to succeed in this environment, it won't produce much improvement in their real world ability.
But maybe that's not the goal, and it's simply to test understanding. Well, as other commenters have stated, this seems trivially cheatable. So it neither succeeds at improving one's ability in oral communication nor at testing understanding. Other solutions have to be thought of.
LLM oral exams can provide assessment in a student's native language. This can be very important in some scenarios!
Unlimited attempts won't work in the presented model. No matter how many cases you have, all will eventually find their way to the various cheating sites.
There is no silver bullet. There's no solution that works for all schools. Strategies that work well for M.I.T. with competitive enrollment and large budgets won't work for a small community college in an agricultural state, with large teaching loads per professor, no TAs, and about 15-25 hours of committee or other non-teaching work. That was my situation.
Teaching five courses and eight sections, 20-30 students per section, 10-20 office hours every week (and often more if the professor cared about the students), leaves little time for grading. In desperation I turned to weekly homework assignments, 4-6 programming projects, and multiple choice exams (containing code and questions about it). Not ideal by any means, just the best I could do.
So I smile now (I'm retired) when I hear about professors with several TAs each, explaining how they do assessment of 36 students at a school with competitive enrollment.
I could accept this for a 300 students class, but 36? When I got my degree, ALL exams had an oral component, usually more than 30 minutes long. The prof and one or two TAs would take a couple days and just do it. For 36 students it’s more than doable. If I was a student being examined by an LLM I would feel like the professor didn’t care enough to do the work.
This is to be expected. The big commercial LLMs generally respond with text that agrees with the user.
> But here's what's interesting: the disagreement wasn't random. Problem Framing and Metrics had 100% agreement within 1 point. Experimentation? Only 57%.
> Why? When students give clear, specific answers, graders agree. When students give vague hand-wavy answers, graders (human or AI) disagree on how much partial credit to give. The low agreement on experimentation reflects genuine ambiguity in student responses, not grader noise.
The disagreement between the LLMs is interesting. I would hesitate to conclude that "low agreement on experimentation reflects genuine ambiguity in student responses." It could be that it reflects genuine ambiguity on the part of the graders/LLMs as to how a response should be graded.
Part 2 is that when you are ready, an examiner sits with you, looks over your stuff and asks questions about it, like clarifications, errors to see if you can fix them, fake errors to see if you can defend your solution, sometimes even variations or unrelated questions if they are on the fence as to the grade. Typically that takes 3-10 minutes per person.
Works great to catch cheating between students, textbook copying and such.
Given that people finish asynchronously you don't need that many examiners.
As to being more stressful for students I never understood this argument. So is real life.. being free from challenge based stress is for kindergarteners
In a presentation, you are in control. You decide how you will present the information and what is relevant to the theme. Even if you get questions, they will be related to the matter at hand that you need to dominate in order to present.
In oral exams, the pressure is just too great. I doubt it translates to a proper job. When I'm doing my job, I don't need to come up with answers right there on the spot. If I don't remember something, I have time to think it through, or to go and check it out. I think most jobs are like this.
I don't mind the pressure when something goes wrong in the job and needs a quick fix. But being right there, in an oral exam, in front of an antagonistic judge (even if they have good intentions) is not really the way to show knowledge, I think.
(I invented some kind of metric based on a centered gaussian around a country ahaha)
One big issue that I had is that the system asked for a number in dollars, but if I answer $2000,2000,2000 per agent per month, the answer was always the same, I cannot accept a number, give it in words, after many tries I stopped playing, it wasn't clear what it wanted.
I could see myself using the system. With another voice as it was kind of agressive. More guidelines would be needed to know exactly how to pass a question or specify numbers.
I don't know my grade, so I don't know how much we can bullshit the system and pass
'This next thing is the best idea ever and you will agree! Recruiters want to sell bananas '
'OK, good, what is the... '
I hope this is catched by the grading system afterward.
The student is supposed to submit a whole conversation with an LLMs.
The LLM is prompted to answer a question or resolve a problem, and the LLM is there to assist. The LLM is instructed to never reveal the answer.
More interesting is the concept that the whole conversation is available to the instructor for grading. So if the LLMs makes mistake, or give away the solution, or if the student prompt engineer around it. It is all there and the instructor can take the necessary corrective measures.
87% of the students quite liked it, and we are looking forward to doubling the students that will be using it next quarter.
Overall, we are looking for more instructor to use it. So if you are interested in it please get in touch.
More info on: https://llteacher.blogspot.com/
I'm still somewhat concerned about exposing kids to this level of sycophancy, but I guess it will be done with or without using it in education directly.
Students are very simply NOT doing the work that is require to learn.
Before LLMs, homeworks were a great way to force students to approach the material. Students did not have any other way to get an answer, so they were forced to study and come up with an answer to the homeworks. They could always copy from classmates, but that was considered quite negatively.
LLMs change this completely. Any kind of homework you could assign undergraduates classes are now completed in less than 1 second, for free, by LLMs.
We start to see PERFECT homeworks submitted by students who could not get a 50% grade in classes. Overall grades went down.
This is a common pattern with all the educators I have been talking with. Not a single one has a different experience.
And, I do understand students. They are busy, they may not feel engaged by all the classes, and LLMs are a way too fast solution for getting homeworks done and free up some time.
But it is not helping them.
Solutions like this are to force students to put the correct amount of work in their education.
And I would love if all of this would not be necessary. But it is.
I come from an engineering school in Europe - we simply did not have homework. We had frontal classes and one big final exams. Courses in which only 10% of the class would pass were not uncommon.
But today education, especially in the US, is different.
This is not forcing student to use LLMs. We are trying to force student to think and do the right thing for them.
And I know it sounds very paternalistic - but if you have better ideas, I am open.
- The stuff being covered in high school is indeed pretty useless for most people. Not all, but most, and it is not that irrational for many to actually ignore it.
- The reduction in social mobility decreasing the motivation for people to work hard for anything in general, as they get disillusioned.
- The assessment mechanisms being easily gamed through cheating doesn't help.
It's probably time to re-evaluate what's taught in school, and what really matters. I'm not that anti-school but a lot of the homework I've experienced simply did not have to be done in the first place, and LLM is exposing that reality. Switching to in-person oral/written exams and only viewing written works as supplementary, I think, is a fair solution for the time being.
The key implementation detail to me is that the whole class is sitting in on your exam (not super scalable, sure) so you are literally proving to your friends you aren’t full of shit when doing an exam.
I wonder: with a structure like this, it seems feasible to make the LLM exam itself available ahead of time, in its full authentic form.
They say the topic randomization is happening in code, and that this whole thing costs 42¢ per student. Would there be drawbacks to offering more-or-less unlimited practice runs until the student decides they’re ready for the round that counts?
I guess the extra opportunities might allow an enterprising student to find a way to game the exam, but vulnerabilities are something you’d want to fix anyway…
To the extent of wondering what value the human instructors add.
Absolutely the easiest solution would have been to have a written exam on the cases and concepts that we discussed in class. It would take a few hours to create and grade the exam.
But at a university you should experiment and learn. What better class to experiment and learn than the “AI Product Management”. Students were actually intrigued by the idea themselves.
The key goal: we wanted to ensure that the projects that students submitted was actually their own work, not “outsourced” (in a general sense) to teammates or to an LLM.
Gemini 3 and NotebookLM with slide generation were released in the middle of the class, and we realized that it is feasible for a student to have a flaweless presentation in front of the class, without understanding deeply what they are presenting.
We could schedule oral exams during the finals week, which would be a major disruption for the students, or schedule exams during the break, violating university rules and ruining students vacation.
But as I said, we learned that AI-driven interviews are more structured and better than human-driven ones, because humans do get tired, and they do have biases based on who is the person they are interviewing. That’s why we decided to experiment with voice AI for running the oral exam.
https://sibylline.dev/articles/2025-12-31-how-agent-evals-ca...
if we want to educate people 'how people work', companies should be hiring interns and teaching them how people work. university education should be about education (duh) and deep diving into a few specialized topics, not job preparedness. AI makes this disconnect that much more obvious.
This is why we need to continue to educate humans for now and assess their knowledge without use of AI tools.
Here, I'll identify another: There is much pain and suffering in this world.
Coming up with a solution is left as an excercise for the reader.
Perhaps we as humans should stop making choices which cause pain.
Why do you make choices that cause pain in yourself and others?
In reality, they cheat when a culture of cheating makes it no longer humiliating to admit you do it, and when the punishments are so lax that it becomes a risk assessment rather than an ethical judgment. Same reason companies decide to break the law when the expected cost of any law enforcement is low enough to be worth it. When I was in college, overt cheating would be expulsion with 2 (and sometimes even 1 if it was bad enough) offenses. Absolutely not worth even giving the impression of any misconduct. Now there are colleges that let student tribunals decide how to punish their classmates who cheat (with the absolutely predictable outcome)
The real problem is students and universities have collectively bought into a "customer mindset". When they do poorly, it's always the school's fault. They're "paying customers" after-all, they're (in their mind) entitled to the degree as if it is a seamless transaction. Getting in was the hardest part for most students, so now they believe they have already proven themselves and should as a matter of routine after 3-4 years be handed their degree because they exchanged some funds. Most students would gladly accept no grades if it was possible.
Unfortunately, rather than having spines, most schools have also adopted a "the customer is always right" approach, and endlessly chase graduation numbers as a goal in and of itself and are terrified of "bad reviews."
There has been lots of handwringing around AI and cheating and what solutions are possible. Mine is actually relatively simple. University and college should get really hard again (I'm aware it was a finishing school a century ago, but the grade inflation compared to just 50 years ago is insane). Across all disciplines. Students aren't "paying for a degree", they're paying to prove that they can learn, and the only way to really prove that is to make it hard as hell and to make them care about learning in order to get to the degree - to earn it. Otherwise, as we've seen, the value of the degree becomes suspect leading to the university to become suspect as a whole.
Schools are terrified of this, but they have to start failing students and committing to it.
I graduated from a SUNY school in 2012. At the time, you could still actually go to school and work part time and get through it. Not saying it was easy by any stretch but it was possible. Tuition + living expenses were about $17/year on campus , less expensive housing was available off campus.
Now, even state schools have tuition which is only affordable through family wealth or loans. Going to university is no longer a low stakes choice - if you flunk you’re stuck with that debt forever. Not to say students aren’t responsible for understanding that when signing up, but the stakes are just a lot higher than what it used to be.
I suppose there are other fields where the degree might be used mostly as a filtering mechanism, where cheating through graduation might get you a job doing work different than your classes anyway. However, even in those cases it's hard to break the habit of cheating your way around every difficult problem that comes your way.
The two solutions to this are (1) as some commenters here are suggesting, give up entirely and focus only on quality of output, or (2) teach students to care about being more than appearance. Make students want to write essays. It is for their personal edification and intellectual flourishing. The benefits of this far surpass output.
Obviously this is an enormously difficult task, but let us not suppose it an unworthy one.
Would you mind expanding on what exactly that entails?
I would prefer to write responses to textual questions rather than respond verbally to spoken questions in most cases.
Wouldn't a written exam--or even a digital one, taken in class on school-provided machines--be almost as good?
As long as it's not a hundred person class or something, you can also have an oral component taken in small groups.
Students had and still have the option to collectively choose not to use AI to cheat. We can go back to written work at any time. And yet they continue to use it. Curious.
Wouldn't that be a fine outcome?
Individuals can't "collectively" choose anything.
This test is given to the entire class, including people who never touched AI.
Students could absolutely organize a consensus decision to not use AI. People do this all the time. How do you think human organizations continue to exist?
I know we've had historical record of people saying this for 2000 years and counting, but I suspect the future is well and truly bleak. Not because of the next generation of students, but because of the current generation of educators unable to successfully adapt to new challenges in a way that is actually beneficial to the student that it is supposed to be their duty to teach.
Ask the student to come to the exam and write something new, which is similar to what they've been working on at home but not the same. You can even let them bring what they've done at home for reference, which will help if they actually understand what they've produced to date.
Let me ask, how do you generally feel when you contact customer service about something and you get an AI chatbot? Now imagine the chatbot is responsible for whether you pass the course.
If the class cost me $50? Then sure, use Dr. Slop to examine my knowledge. But this professor's school charges them $90,000 a year and over $200k to get an MBA? Hell no!
At that point what’s the value add over using YouTube videos and ChatGPT on your own?
Adding this as an additional optional tool, though, is an excellent idea.
And universities wonder why enrollment is dropping.
> 36 students examined over 9 days > 25 minutes average (range: 9–64)
It appears that they examined only 4hrs each day, one student at a time. This is incredibly inefficient.
In my experience, the greatest benefit of doing something like this would be to be able to run these exams in parallel, while retaining a somewhat impartial grading system.
> The grading was stricter than my own default. That's not a bug. Students will be evaluated outside the university, and the world is not known for grade inflation.
Good!
> 83% of students found the oral exam framework more stressful than a written exam.
That's alright -- that's how life goes. This reminds me of a history teacher I had in middle school who told us how oral exams were done at the university he had studied in: in class, each student would come up to the front, pick three topics at random from a lottery-ball-picker type setup, and then they'd have a few minutes in which to explain how all three are related. I would think that would be stressful except to those who enjoy the topic (in this case: history) and mastered the material.
> Accessibility defaults. Offer practice runs, allow extra time, and provide alternatives when voice interaction creates unnecessary barriers.
Yes, obviously this won't work for deaf students. But why must it be an oral examination anyways? In the real world (see above example) you can't cheat at an oral examination because you're physically present, with no cheat sheets, just you, and you have to answer in real time. But these are "take-at-home" oral exams, so they had to add a requirement of audio/video recording to restore the value of the "physically present" part of old-school oral exams -- if you could do something like that for written exams, surely you would?
Clearly a take-home written exam would be prone to cheating even with a real-time AI examiner, but the real-time requirement might be good enough in many cases, and probably always for in-class exams.
Oh, that brings me to: TFA does not explicitly say it, but it strongly implies that these oral exams were take-at-home exams! This is a very important detail. Obviously the students couldn't do concurrent oral exams in class, not unless they were all wearing high quality headsets (and even then). The exams could have been in school facilities with one student present at a time, but that would have taken a lot of time and would not have required that the student provide webcam+audio recordings -- the school would have performed those recordings themselves.
My bottom-line take: you can have a per-student AI examiner, and this is more important than the exam being oral, as long as you can prevent cheating where the exam is not oral.
PS: A sample of FakeFoster would have been nice. I found videos online of Foster Provost speaking, but it's hard to tell from those how intimidating FakeFoster might have been.
---
> Only 13% preferred the AI oral format. 57% wanted traditional written exams. 83% found it more stressful.
> Here is an email from a student: "Just got done with my oral exam. [...] I honestly didn't feel comfortable with it at all. The voice you picked was so condescending that it actually dropped my confidence. [...] I don't know why but the agent was shouting at me."
> Student: "Can you repeat the question?" Agent: paraphrases the question in a subtly different way.
> Students would pause to think, and the agent would jump in with follow-up probes or worse: interpret the silence as confusion and move on.
---
Based on these highlights, you'd think the experiment was a wash. The author disagrees!
> But here's the thing: 70% agreed it tested their actual understanding: the highest-rated item.
Man, you could shoot me with a gun, then make me write an essay, & I'd be forced to agree that you had tested my "actual understanding." That doesn't mean my performance wouldn't suffer. Also, 70% is not very high. That's barely two thirds.
Even the grading was done by LLMs (rather than having a TA grade a transcript, and the results were lower. The author defends this by saying, "Students will be evaluated outside the university, and the world is not known for grade inflation," but the world isn't "known for grade inflation" because it doesn't grade you at all. That's not even an excuse, it's just nonsense. It'll toughen you up, or whatever. Was this post written by an LLM too?
> Take-home exams are dead. Reverting to pen-and-paper exams in the classroom feels like a regression.
"Regression"? I mostly wrote pen & paper exams, and I only graduated a few years ago. If students want more flexibility, team up with other courses to supervise multiple exam sessions. Leaked questions aren't going to be any more of a problem than it was for take-home exams, especially since they can't take the booklets with them when they go.
It sounds like these students had a terrible time, and for what? Written exams work fine. These guys just wanted to play with LLMs.
...but OTOH if cheating is so easy it's impossible to resist and when everyone cheats honest students are the ones getting all the bad grades, what else can you do?
Thank god I had a chance to study in pre-AI times.
First, the business school administration and faculty firmly commits, that plagiarism, including with AI, means prompt dismissal.
Then, the first time you have a suspicion of plagiarism, you investigate.
After the first student of a class year is found guilty, and smacked to curb, all the other students will know, and I bet your problem is mostly solved for that class year.
Then, one coked-up nepo baby sociopath will think they are too smart or meritorious to "fail" by getting caught. Bam! Smacked to the curb.
Then one of those two will try sue, and the university PR professionals will laugh at them, for putting their name in the news as someone who got kicked out of business school for cheating. The business school will take this opportunity to bolster their reputation for excellence.
At this point, it will become standard advice for the subsequent class years, that cheating at this school is something only an idiot loser does, not a winner MBA.
Some people dream that technology (preferably duly packaged by for-profit SV concerns) can and will eventually solve each and every problem in the world; unfortunately what education boils down to is good, old-fashioned teaching. By teachers. Nothing whatsoever replaces a good, talented, and attentive teacher, all the technologies in the world, from planetariums to manim, can only augment a good teacher.
Grading students with LLMs is already tone-deaf, but presenting this trainwreck of a result and framing it as any sort of success... Let's just say it reeks of 2025.