It is. I think the professor here was being naive, but I appreciate his optimism. When I was in college (in the 90s), take home exams allowed a knowledgeable student to really shine. I’m not saying that they weren’t eminently cheatable back then—they were—but they also had the odd side-effect that, if it was a class you cared about, the test itself could be a learning experience.
For context, I am also a faculty member at a highly selective college. I had a similar shocking realization last year that it was likely that there was widespread cheating on homework assignments, which I used to favor heavily toward their grades. To verify my suspicions, I generated custom tests for every student in the class: the exam included code from students’ own programming assignment submissions. All I asked them to do was explain what they wrote.
The class performed badly on this exam, and the results were strongly bimodal. Roughly half the class aced the exam. The other half could make neither heads nor tails out of the code. For the students who wrote things like “lol,
i have no idea” (real response) I opened honor cases.
I think many faculty right now are going through the stages of grief. We all knew that even at selective institutions, cheating existed, that many students were in it for the credentials. But as long as the numbers of known cases was low, we could convince ourselves that the few doing it were outliers. When a class does it en masse, it’s more than a slap in the face; it makes you feel like a chump. Have we been fooling ourselves this entire time? Was all the time I spent becoming a subject-matter expert a waste? Are the students just rolling their eyes when I turn my back? Those thoughts hurt. I personally chose to become a faculty member because it seemed like research and teaching were the best ways to maximize my impact.
I still have some hope. After all, I still spend my days working and socializing with like-minded thinkers, some of whom are truly brilliant. And every year, a handful of students come out of the woodwork and surprise me. But it’s hard not to think that the group of people who find joy in learning and creating is shrinking.
But it’s hard not to think that the group of people who find joy in learning and creating is shrinking.
I'm not sure you should think it is shrinking. There are a lot of people in this world that hate to learn, and literally are incredibly apathetic about any topic. To such, learning anything is work, never a joy.
Before AI they had to learn to succeed. Now they see a shortcut. You said half showed they were learning, that's not so bad. I think you should be glad it's that high. I am.
College is a box to check off because we have lost the ability to support a workforce without college degrees (at least in the US, UK, AU, where I have some experience).
Trades are critical but looked down on. Manufacturing is gone (which isn't in itself a bad thing). The service industry doesn't pay a living wage (thankfully it's reasonable here in AU). Apprenticeships don't pay enough. And pretty much all knowledge work jobs expect a degree from the beginning at a junior level.
We should be planning for a system where <20% of people go to university, instead of expecting >60% to go. Robust minimum wages, good trade schools, apprenticeships that pay enough for a wide range of roles, and changing the culture to not look down on folks that take these paths.
The majority of people with this mindset should not go to college and should just get a trade job or manufacturing job like they did 100 years ago. People were like this a few decades ago as well when I got my degree. I didn't have consistently great classmates until honors and accelerated grad courses in my 3rd and 4th year, along with various domain specific student orgs like groups for hardware hacking, computer security, etc.
Obviously this can't happen without some structural change (virtually impossible in the US due to its political ossification and indefinite deadlock) because a degree is now just a way of gatekeeping the middle class, but dull and incurious minds made ideal manual laborers in the past, and, at some point, we lost sight of that and started rotting our corporate world out with them.
> The majority of people with this mindset should not go to college
Seems obvious now that this is the solution. Unfortunately, universities would never agree to downsize, not due to politics or any US-specific problems.
> The majority of people with this mindset should not go to college and should just get a trade job or manufacturing job like they did 100 years ago
As long as this allows them to provide for themselves and their families, buying homes, and retiring the same way as someone who goes to college.
Going to college should not be something people do to escape abject poverty, but something you want to do to enrich your life and make yourself a better person.
My experience as a student has always been that most of the class will cheat given the chance. I remember in high school being one of maybe a dozen in a grade of over 200 who actually read the assigned novels. Everyone else used cliff/spark notes not just to familiarise themselves but also to plagiarise essays.
> It is. I think the professor here was being naive, but I appreciate his optimism. When I was in college (in the 90s), take home exams allowed a knowledgeable student to really shine. I’m not saying that they weren’t eminently cheatable back then—they were—but they also had the odd side-effect that, if it was a class you cared about, the test itself could be a learning experience.
I've never realized it before, but this is so true. Take home exams (or graded homeworks) allowed me to pour so much time in subjects I enjoyed, that I wouldn't have done otherwise.
Yeah, I’m enrolled in an online degree program and use of AI in exams would be quite difficult.
The proctoring service my school uses requires a special browser with admin permissions and an external webcam with your entire workspace, screen, and face clearly visible (wide angle webcam preferred). Prior to the exam you have to photograph the entirety of your room, and if there’s even an open door that can disqualify you. Only one screen is allowed and smartphones and smart watches are banned. A proctor is watching you and you’re being recorded the whole time. I have no idea how one could slip something like AI use past all this.
Sounds like smart glasses might do it, if you normally wear glasses. Don't know how visible the display is from the outside, but I imagine if it isn't already they'd eventually be made invisible for privacy.
For now smart glasses all still look strange (e.g. chunky in places normal glasses never are) which acts as a tell and displays remain at least somewhat visible. Hiding screens entirely would likely entail also blocking the light bouncing back from the user’s face in the screen areas which would look odd.
How is this even a debate? Before Covid most tests were in person right? Sure some classes had final projects that were take home, but in person tests were very norma. So what’s all the hand wringing about? Just do in person tests and move on?
I didn't attend an Ivy League, but I think I went to a good school. I was very nervous before I left for school - a little intimidated, so I talked to an academic mentor. He told me something I'll never forget: "You're gonna be around a lot of really smart kids. No doubt about it. But, mostly, what you're gonna find is you're surrounded by a lot of rich kids." He was 100% correct. Lots of smart kids, and lots of kids from well-to-do families. I think I met, maybe, 2 other kids that were as broke as my family.
The first time I attended a selective school was graduate school. Like you, I was extremely nervous. “They’re all going to be smarter than me. I’m going to feel like an idiot.”
And it turned out to be true. Many of the students I went to school with had far better preparation than I did. And not only did I feel like an idiot, another person called me an idiot in front of everyone. Suspicion confirmed.
The thing is, once I accepted that, yes, maybe my preparation was worse, and that it was possible that I was admitted by mistake, I found a way forward. After all, if literally everyone is smarter than you, then in a way, you’re the luckiest person there: you’re surrounded by smart people, and almost any conversation you have with your peers will benefit YOU more than it benefits THEM.
Over time, I realized that the thing that mattered most was “time on task.” Unlike my peers, who had better instruction, because they went to better schools, had private tutors, etc. I had to work for everything. And I started graduate school late: I turned 30 the year I enrolled. So I was not distracted by social events, finding a romantic partner, or deep questions like “what do I want to do with my life?” I was all-in. I may have started a bit behind, but I finished well ahead of most of my peers.
I think it’s easy for students from my kind of background to wither under the pressure of an elite environment. As a faculty member, I’ve seen it happen many times, sadly. But there IS a way through it, and largely, the way forward is to value oneself, do develop one’s internal compass for good work, and to not let the social pressures overwhelm. I don’t mean to make this sound easy, but it IS possible.
I’m not sure why that’s controversial - I have met many Ivy League students and grads; they are all intelligent, at least in an academic way. The only other common characteristic is that they almost always had some form of privilege. Either rich parents, or adults around them who worked very hard to get them to that level.
I agree that they are intelligent, just don't know about the "definition" part. A typical Ivy Leaguer isn't a dumbass. What's wrong with calling one intelligent?
Try visiting a Walmart and interacting with literally anyone. That's the average. Let's not allow our egos to gatekeep who we consider intelligent, fellow HNians.
Why jump on the opportunity to prune reading by rejecting the lot as soon an unrelated premise you disagree with is presented?
Perhaps at that point if you stop reading after the first sentence, you could churn the entire article through AI to summarize it into a single sentence, and see if the invalid premise is core to the message?
> Why jump on the opportunity to prune reading by rejecting the lot as soon an unrelated premise you disagree with is presented?
Ars Technica has already trashed its reputation with the infamous controversy over publishing AI hallucinated quotations.
I couldn't decide whether the opening sentence of this article was so dumb that it had to be written by AI or so dumb that it had to be written by a human, and coming to a conclusion on that issue didn't seem worth it.
> you could churn the entire article through AI to summarize it
Seems like an application of Goodhart's law; measuring worth by degree or grades stopped measuring learning or ability.
This was a lot harder to cheat before AI, but now the floodgates are open and grades and degrees earned post-AI are showing that they mean little.
Cheating on college tests should be a jailable criminal offense (similar to computer fraud) so that there is dignity in the degree again. Considering the money involved, I don't see why not.
But this probably won't happen, because many rich people are very happy to buy their degrees. See also [1]
In France there's this law that cheating on any official exam gets you a 5 year ban on any other official exams. That includes finals, but also the exam to get the driver's license, among others.
This won't happen for the same reason very few professors stand up to this shit and why everyone gets straight A's at Harvard now - Goodhart's Law again. The money involved means getting accepted into the school is the whole task, grades are irrelevant and learning is opt-in at best. Stamp the kids, extract the money, funnel it up to the administration.
'
“56 percent of undergraduate respondents [at Brown] and 67 percent of graduate and medical student respondents reported intentionally using GenAI tools daily or weekly,”
'
and the rest are lying.
(With apologies to the original example of anomalous self-reporting)
This is already a clusterfuck, but it's going to be so much worse in 10 years. We're going to have an entire generation trapped in the gig economy because their education is going to be considered worthless, and even if it wasn't worthless, there won't be enough entry level jobs for anyone to get into. Senior people will age out and our entire society is just going to be hollowed out.
The thing you're angry at, the thing that you're upset about? It's capitalism, it's the coupling of education with jobs, it's credentialism being overturned by new technology, etc.
I think education is incredibly important, but I understand that I'm going to have to retrain myself a little bit. A college degree can no-longer be assumed to be a proxy for having put in the effort to deeply study something.
Now what's the solution for this? I don't know, but we have made the mistake of conflating pieces of paper for expertise. And I say that as someone with 3 degrees.
Thinking back to my time as a professional pilot before I medicaled out and pivoted into tech, the FAA really (for all it's problems) has a pretty good system to train and test new pilots.
You have to have some hours with a certified instructor and some hours on your own. The tests to become a certified instructor are considered challenging, and many people fail. Then you take a written test, then you take a practical test. It's one on one. You and the examiner. And if you do not meet the standard, you fail. That's "ok." It's just fine to fail people who do poorly during a checkride. They go back, they get retrained, and they do it again.
If you have a lot of failures during training, you'll have to answer for them in interviews later on, but often times there's a sort of holistic treatment to it. If you busted a checkride 15 years ago, and have since been fine, you'll be ok. If it's a recurring theme, you'll have a hard time finding a job (and that's the right thing, IMO). But the format of "Written, Oral Exam, and Practical Exam" is the "right" model for making sure people know wtf they are doing.
How do we do that in tech? Hell if I know, maybe a proctored written exam, followed by an oral exam, then a project? But who knows.
While I think there are a lot of problems created by capitalism right now, I would argue that nothing about the current AI industry is actually capitalist at all. These are companies that are surviving despite absolutely bleeding cash because the ultra-rich have decided that things would be so much more convenient for them if they didn't need to deal with employees anymore. This isn't "well, people are doing this because it's profitable". They're doing it DESPITE it being WILDLY unprofitable, mostly based on a weirdly religious faith that it's just going to work out because they really really want it to.
Recently I saw Alex Karp (Palantir CEO) on one of those market shows (CNBC I think? Something like that) and he was upset at Dario's doom trolling. Why was he upset? Because he was worried that the populist rage towards AI was going to result in a wealth tax! And the thing is he's probably right! The thing these people fear the most, socialism, is the thing they're ironically probably going to make popular in the united states.
Honestly, and this is my "hot take of the decade" but I think we're going to get some sort of AI modulated/managed form of capitalism that will approach socialism in some regards but still allow for people to try to "get rich" or whatever.
What that's going to look like I couldn't begin to imagine, but I think you might onto something about the AI industry not being all that capitalistic presently. Not an economist, but we're definitely in oligopoly territory here? Because there are only like 4-5 serious companies here.
I too may be accused of having the faith that it's "all going to work out" though. I think it will. I think when things start getting weird enough a bunch of creative people will figure out logical solutions that work in the moment and we'll run with those. Barring revolution, that's kind of how things have always been. Things get weird, then we figure out how to fix it or go to blows over it. I think going to blows domestically is pretty unlikely as of yet though despite the "dooooom" narrative that a lot of people tout.
Cynical? The press release for the latest Grok version is sitting at the top of the front page. And half the people on this site don't see the problem with that. They'll welcome their roman-saluting overlords with open arms.
> Honestly, and this is my "hot take of the decade" but I think we're going to get some sort of AI modulated/managed form of capitalism that will approach socialism in some regards but still allow for people to try to "get rich" or whatever.
That's optimistic. IMHO, it's more likely the libertarian tech-lords make sure we get some kind of hyper-capitalism, where you're either rich or bone-crushingly poor.
I get that doom sense, but... I can't really see it happening because... well, people that have never truly been hungry are already frantically clacking at their keyboards with the slogan of "eat the rich!"
I don't know what we get? But if this is the rhetoric now, how many Luigis would we expect to see as things got really bad, and the vision you paint is really bad.
The thing with AI and AI-enabled automation is that it may lead to a world where the rich don't need anyone else anymore. I could totally imagine things shaking out where the rich and everything important is so-well defended that even an army of Luigis is impotent.
Also, after a couple assassinations of business leaders I think we'll see gun control laws like the US has never seen.
Took me a while to get rid of the mental image of an army of green moustached Luigis crowding a cartoon palace. But maybe it's not even such a bad image after all. Now where are the Marios when you need them?
There are so many firearms in America even if there was bipartisan support for a gun ban it wouldn’t happen. You literally couldn’t get rid of them if you tried.
Maybe? But I cannot see that as a stable end state. I think we'll get some sort of weird managed capitalism for exactly that reason.
I actually don't really want that sort of outcome, I'd like to see some sort of libertarian socialism arise out of post-scarcity? But what that looks like? I don't know. 20 year old me would have had lots of ideas about how we "ought to be" but now, pushing 40? I am less certain of the things I "know."
Regardless, I don't see the really bad guys winning this one...
So like, remember in 2008 when we had to bail out the major banks because they had failed so hard -- but for the common person it was "oh, well, you made a bad decision so tough luck, we're taking your house." In that case it was socialism for the big companies (where do you think the bailout money came from?), harsh capitalism for everyone else.
It's the same thing with AI right now. Oh, you guys need another trillion to burn and you have no plan for profitability? Here's a blank check! Oh, your job is going to be eliminated and we have no coherent answer for what you should do next, because we want to take that job too? Not our problem.
Capitalism:
> an economic system characterized by private or corporate ownership of capital goods, by investments that are determined by private decision, and by prices, production, and the distribution of goods that are determined mainly by competition in a free market
Any semblance of capitalism we have is a shambling corpse barely getting around. Do you know who pulls on all the strings of our economy, gives out student aid loans to kids who just barely graduates high school? Who controls market interest rates, and gives massive multi billion dollar contracts to Companies like OpenAI, Anthropic, Facebook, Google, Microsoft?
Who approved of the printing of the 50% increase in the dollar supply in the last 5 years?
Take a guess.
I'll give you a hint, if you don't pay them on a yearly basis 30-50% of your paycheck, you go to prison.
> Any semblance of capitalism we have is a shambling corpse barely getting around. Do you know who pulls on all the strings of our economy, gives out student aid loans to kids who just barely graduates high school? Who controls market interest rates, and gives massive multi billion dollar contracts to Companies like OpenAI, Anthropic, Facebook, Google, Microsoft? Who approved of the printing of the 50% increase in the dollar supply in the last 5 years?
Nothing in your provided definition of capitalism precludes government control of money supply, taxation, regulation, or even government contracts.
Private ownership still owns capital goods and private decision is the primary driver of investment, prices, production and distribution in most industries. This is why we have a hard time with things that are obviously good for society as a whole, like renewable energy, efficient transportation, and low-cost housing, that don't have a competitive short-term return on investment.
> How do we do that in tech? Hell if I know, maybe a proctored written exam, followed by an oral exam, then a project? But who knows.
Ever taken a Redhat exam - such as Red Hat Certified Engineer? That's the right model. It's a hand-on exam with arbitrary exam objectives you have to meet (by any means possible, there is more than one right way to achieve desired state). There is no internet, devices, or anything else to help you. People that pass allow you to make assumptions about them - for example... without stackoverflow, github, or the internet at all - this person can configure virtual hosts with nginx, firewall rules, host VMs, and administer network concepts(virtual switches, routing) between them. To do that without any GUI or internet available - it means you understand those concepts very well, and can implement them blindfolded.
you have access to all of the redhat documentation but no internet. if there is a right way to do it as written in the docs then you can often find it.
no internet-style indexing tho, so it's a lot of ctrl-f and hoping you pick the correct chapter -- which can be hard with a timelimit. on stuff I was strong in it was trivial, I'd already seen the docs and knew what to look for but on the few items I wasn't it was a lot of flailing.
Man, I'm not a redhat guy, but that's pretty cool? I know I couldn't do that without the docs and a lot of time, but that's a good model. A practical demo is legit.
Maybe I'm a weirdo, but part of me wishes I could be like a sysadmin in a powerplant or like something similar where we used tools like that and I'd have to get really good at using them.
Flying was like that, you had to know what you were doing, and aspects of tech touch that, but certainly not in the domain I'm currently in. Seriously, cool, I didn't even know this existed.
But education is almost worthless now already. Nowadays pretty much everyone has a degree and then couldn't find a job and lucking real skills and experience. Ai is not a cause, it will simply speed up the degradation of the system.
> Nowadays pretty much everyone has a degree and then couldn't find a job
The unemployment rate in the US is 3.7%. 38% of adults in the US have a college degree. 38% is hardly everyone, and if nobody can find a job, then the unemployment rate should be far higher than 3.7%, which is basically full employment.
I don't trust the unemployment numbers to be useful here. (Also google says 4.2%, but I don't trust google's hallucination prone search results). Lets say you graduated with a STEM degree and now you're working at starbucks or walmart. Congrats you're employed! Is that sustainable or good? It's just not a useful metric. It's like the super rich pointing at the stock market and going "we're not in a recession, the number is going up!" and yet real buying power and wages are declining for everyone except the massively wealthy.
Yup. You can either A) ban AI (good luck) and teach students to understand CS fundamentals deeply - which will of course make them totally unsuitable for an industry that is demanding tokenmaxxing from every engineer; or B) embrace AI at the university and produce students that are nothing more than prompt "engineers".
There is no win condition. Despite the usefulness claims of generative AI, it is a net negative on society and on education. Humanity will somehow make do and get used to it; but I doubt the upheaval we will have to go through the next X decades will have any silver lining other than “we made billionaires richer and exponentially increased wealth disparity”
I honestly pity watching my fellow software engineers rush to adapt to this new dystopia, just to stay afloat a little while longer before they are truly obsolete. It is quite tragic.
> there won't be enough entry level jobs for anyone to get into.
You know, I was reading an article about falling labor force participation rates, and one thing was pointed out was there seems to be a supply mismatch. Plenty of jobs are available.
I suspect the only retirement I’ll ever see will have to be via a shotgun.
The article, the teacher, and the general academic community skips the hard question when it comes to AI and that's whether these exams are testing knowledge that is still worth internalizing in the same way?
Academia has a long history of lagging behind acceptance of new cognitive tools where they claim to want to defend the students, but instead defend the assignments of the past at the expense of the students. Calculators were treated as threats to learning, even though they ultimately freed students to focus on higher-level math and provably improved their abilities across many different studies. Internet sources were dismissed as less legitimate than books, as if “published in an outdated book from the 70s” magically made it more trustworthy than the most scrutinized reference sources online.
It is not clear from the article exactly how much of this course falls into that category, but if the answers can be produced trivially with a prompt and chatgpt, then maybe memorizing that material is no longer the right educational target. Academia desperately needs to redesign itself around AI as a cognitive tool students should be trained to leverage. If a question is trivially answered by a prompt with it, then you need harder questions that actually require students to push beyond that. Simply removing AI from the equation, calling it cheating, and pretending that it isn't an ever-present asset people are expected to leverage in real life is naive and just repeats the mistakes of the past.
I don't think there's anything wrong with asking for things that can be prompted. You still need to understand the why behind things, being able to reason about them and choose between options. How will you teach this level of understanding or certify them without exams?
Of course, not all testing is good, but the written exam has survived and proven useful despite the internet age, I'm not sure an even better search engine really changes that.
You need to know what to ask and what the answer means. Also, using proper terminology greatly increases LLM accuracy to situate the response in the correct domain.
Ask an LLM to "clean up this code applying SOLID principles" is much more prescriptive than "clean up this code".
AI can just give you the details like an encyclopedia article. You still have to know at a high level how something works and which article to look for.
AI is just a more forgiving way to ask something, compared to the searches of the past (e.g. searching a scientific journal database). And (for now at least), AI is a good way to avoid the ad-ridden web.
> whether these exams are testing knowledge that is still worth internalizing ... It is not clear from the article exactly how much of this course falls into that category
It's very clear from the (excellent) article linked by dang [1] what the exams required:
> This year, the economist decided that both the midterm and the final exams for his course would be of the take-home, closed-book type (there is a certain tradition of this at Ivy League schools). “It’s a very nice kind of exam, because as you’re giving students practically unlimited time to complete it, it lets you make it harder than normal, to see how far they can go.” In this case, Serrano changed some of the model assumptions they had seen in class, and asked students to demonstrate whether certain statements were true or false under the new assumptions.
I have a few thoughts related to this, and maybe I can get them out and ties them back together at the end.
1a. Yes, college isn’t right for everyone, and testing the traditional way with paper and pencil certainly disadvantages some students who would be star performers in a real world setting but are not a good fit for college classes. Think “Good Will Hunting” type people. They do exist.
1b. However, there are certainly more people who only imagine themselves as Will Hunting type people and that they are just too smart for college, but the reality is that they are dumb, or didn’t learn the material. For every person who fails a test because they’re a genius who is a bad fit in the system, there are at least 10 idiots who imagine themselves geniuses, and they would have passed the class if only that PhD professor with all his book smarts had actually written the right kind of exam. Traditional schooling and testing doesn’t work for the extreme upper tail of intelligence, but it exists because it did quite well at educating and sorting the masses to support the Industrial Revolution.
1c. If college were only about the knowledge, you could learn most of it with internet access and a library card for much cheaper. The vast majority of people are in college for the credential, and the institution has to protect the signal of the credential.
2a. Most college assignments and exams are not a good reflection of full time employment. College credentials serve mainly as a networking aid and a signal to employers that you are compliant and competent enough to follow a professor’s instructions, and will likely be a compliant and competent employee, although the instructions might be different. If it is the AI who is the one who followed the professor’s instructions you water the signal down, and employers don’t want that. It is irrelevant that a student can copy and paste things into ChatGPT, and that ChatGPT can get the answer right on this test. That isn’t what college is supposed to signal.
2b. A problem is that I literally can’t write a test in most subjects now that I would expect a student to complete that can’t be completed by ChatGPT better and faster. I teach undergraduate math and a while ago we thought that since GPT-4o could get a C in calculus that we’d just raise the standard. Now Fable and GPT-5.5 can cruise to an A in literally every math course in our catalog, and they can also catch every tiny issue in an exam written by a human. But I have to teach these undergraduate subjects so that some students can go on to PhD studies so they can contribute to the field. If we just stop teaching undergraduate subjects then PhD production and novel research grinds to a halt and only a few fields will progress where an AI is capable of self improvement.
2c. I’ve seen that my best students know how to do do something by hand and use a computer to complement/increase their capabilities, not to cover over entire gaps. When you have literally zero skill in an area, you can’t spot when you got a totally bad output from the AI (these days usually because of a lack of context or bad prompting because the student didn’t understand the material, not because the AI wasn’t capable). Somehow I have to incentivize students to learn the material on their own, so that they can be a better user of AI in the future. And a really effective way to do that is a graded test in which AI is not available to help them.
So to try to tie it together, is that there is still some value to a college degree, at least until something better comes along. But that college degree is only useful when it is a signal about the person and not some other tool. And although AI is getting very capable, somehow we have to teach the lower stuff to build up to the higher stuff, so we do need to restrict the use of AI in some educational settings so that we can build a good foundation for future learning.
As to the point about old paper sources being considered more reliable than an internet site, I agree...
It's worth noting that even though you say people could just go online and start reading, college may be the last time most people ever read a textbook ever again. There's something tough about knowing that you want to do something but then somehow your daily instincts fight you.
It's also why you can't just throw kids in front of Khan Academy.
> Somehow I have to incentivize students to learn the material on their own... And a really effective way to do that is a graded test ...
I know. The desire to get a degree or the "threat" of wasted tuition and disappointed family members is really what gets most students to crack a book. Most people aren't self-motivated to learn. If people were self-motivated to learn the material, as opposed to getting a credential or some other reason for being in college, I wouldn't need to worry about cheaters.
I think the reality is that you should focus on and serve the 2c group (best students who use AI to fill gaps or speed up learning). We have 10s of thousands (or more) students taking advanced math every year. Only a slice of those pursue PhDs and even fewer contribute anything meaningful.
This is how it has been for years and AI doesn't really stunt that as long as students actually care about learning the material. Those kids will be fine. The kids that are fulfilling a prereq will be fine, for the most part, but will walk on crutches their whole lives (use AI) which might be a reasonable choice.
What we lose is the signal of which kids are interested in learning and which kids aren't. Grades and effort used to be the signal, now I think it has to be more about engagement" who asks questions, who comes in with strange ideas, who actually is struggling at times (actually grappling with the material) or learning things outside of the assignments.
Big picture, I think teachers should focus on teaching the learners and trying to motivate more students to join them. That has been the real job all along, AI just lays it bare.
> Calculators were treated as threats to learning, even though they ultimately freed students to focus on higher-level math and provably improved their abilities across many different studies.
Oh fuck that bullshit. I was a dumbass to believed shit like that in school, and it didn't fucking free me. What happened was I was always calculator-dependent, which made higher-level math harder, because I was always distracting myself by operating the goddamn thing, and I never developed a very good intuition for arithmetic.
It's laughable to be surprised by cheating on take-home exams. Does the professor have a comparative assessment before AI? Do you think the cheating didn't happen on them on before AI? It always did happen. It's structural, in the same ways that speeding on the highway is structural. And the structural fix is in-class exams.
Disallowing the use of AI is basically the modern equivalent of testing based on memorization. It is a laziness to find something more useful to evaluate on or an inability to distinguish real understanding in a field from first level understanding.
Professors need to step up and teach value beyond what LLMs know (very possible with or without LLMS). Or get out of the way from those building on the field with LLMs.
If you’re teaching students something that LLMs can score 100 on you are not adequately teaching them something useful for them in the future.
If your score falls 50% from not using AI, that's not testing memorization. That's testing learning anything on the topic being tested. The quantity of information on college exams is not that much to memorize. At least, personally, failing to remember something was rarely a reason why I failed to answer a question. Almost always it was because I did not fully understand the topic, so I tried to fall back on remembering if I answered the same question before.
The article clearly lays out that the take-home test was not a test of memorization, but a test of students' understanding of the class material and ability to reason about how different assumptions affect its conclusions.
95 comments
[ 3.3 ms ] story [ 60.4 ms ] threadhttps://www.baneproctoring.com/
... I'm sorry I couldn't help it
(sorry, old Internet forum reflexes activated)
For context, I am also a faculty member at a highly selective college. I had a similar shocking realization last year that it was likely that there was widespread cheating on homework assignments, which I used to favor heavily toward their grades. To verify my suspicions, I generated custom tests for every student in the class: the exam included code from students’ own programming assignment submissions. All I asked them to do was explain what they wrote.
The class performed badly on this exam, and the results were strongly bimodal. Roughly half the class aced the exam. The other half could make neither heads nor tails out of the code. For the students who wrote things like “lol, i have no idea” (real response) I opened honor cases.
I think many faculty right now are going through the stages of grief. We all knew that even at selective institutions, cheating existed, that many students were in it for the credentials. But as long as the numbers of known cases was low, we could convince ourselves that the few doing it were outliers. When a class does it en masse, it’s more than a slap in the face; it makes you feel like a chump. Have we been fooling ourselves this entire time? Was all the time I spent becoming a subject-matter expert a waste? Are the students just rolling their eyes when I turn my back? Those thoughts hurt. I personally chose to become a faculty member because it seemed like research and teaching were the best ways to maximize my impact.
I still have some hope. After all, I still spend my days working and socializing with like-minded thinkers, some of whom are truly brilliant. And every year, a handful of students come out of the woodwork and surprise me. But it’s hard not to think that the group of people who find joy in learning and creating is shrinking.
I'm not sure you should think it is shrinking. There are a lot of people in this world that hate to learn, and literally are incredibly apathetic about any topic. To such, learning anything is work, never a joy.
Before AI they had to learn to succeed. Now they see a shortcut. You said half showed they were learning, that's not so bad. I think you should be glad it's that high. I am.
Trades are critical but looked down on. Manufacturing is gone (which isn't in itself a bad thing). The service industry doesn't pay a living wage (thankfully it's reasonable here in AU). Apprenticeships don't pay enough. And pretty much all knowledge work jobs expect a degree from the beginning at a junior level.
We should be planning for a system where <20% of people go to university, instead of expecting >60% to go. Robust minimum wages, good trade schools, apprenticeships that pay enough for a wide range of roles, and changing the culture to not look down on folks that take these paths.
Obviously this can't happen without some structural change (virtually impossible in the US due to its political ossification and indefinite deadlock) because a degree is now just a way of gatekeeping the middle class, but dull and incurious minds made ideal manual laborers in the past, and, at some point, we lost sight of that and started rotting our corporate world out with them.
Seems obvious now that this is the solution. Unfortunately, universities would never agree to downsize, not due to politics or any US-specific problems.
As long as this allows them to provide for themselves and their families, buying homes, and retiring the same way as someone who goes to college.
Going to college should not be something people do to escape abject poverty, but something you want to do to enrich your life and make yourself a better person.
I've never realized it before, but this is so true. Take home exams (or graded homeworks) allowed me to pour so much time in subjects I enjoyed, that I wouldn't have done otherwise.
You cant even sneak paper on to your desk, where do you plan to hide the LLM?
(True or not, a story is a story.)
The proctoring service my school uses requires a special browser with admin permissions and an external webcam with your entire workspace, screen, and face clearly visible (wide angle webcam preferred). Prior to the exam you have to photograph the entirety of your room, and if there’s even an open door that can disqualify you. Only one screen is allowed and smartphones and smart watches are banned. A proctor is watching you and you’re being recorded the whole time. I have no idea how one could slip something like AI use past all this.
I stopped reading after the first sentence.
And it turned out to be true. Many of the students I went to school with had far better preparation than I did. And not only did I feel like an idiot, another person called me an idiot in front of everyone. Suspicion confirmed.
The thing is, once I accepted that, yes, maybe my preparation was worse, and that it was possible that I was admitted by mistake, I found a way forward. After all, if literally everyone is smarter than you, then in a way, you’re the luckiest person there: you’re surrounded by smart people, and almost any conversation you have with your peers will benefit YOU more than it benefits THEM.
Over time, I realized that the thing that mattered most was “time on task.” Unlike my peers, who had better instruction, because they went to better schools, had private tutors, etc. I had to work for everything. And I started graduate school late: I turned 30 the year I enrolled. So I was not distracted by social events, finding a romantic partner, or deep questions like “what do I want to do with my life?” I was all-in. I may have started a bit behind, but I finished well ahead of most of my peers.
I think it’s easy for students from my kind of background to wither under the pressure of an elite environment. As a faculty member, I’ve seen it happen many times, sadly. But there IS a way through it, and largely, the way forward is to value oneself, do develop one’s internal compass for good work, and to not let the social pressures overwhelm. I don’t mean to make this sound easy, but it IS possible.
Do you know what "by definition" means?
> I have met many Ivy League students and grads; they are all intelligent, at least in an academic way.
You probably wouldn't meet the dumb ones, because they're probaly not in your social class:
> rich parents
Try visiting a Walmart and interacting with literally anyone. That's the average. Let's not allow our egos to gatekeep who we consider intelligent, fellow HNians.
Yes, that's the point.
> A typical Ivy Leaguer isn't a dumbass.
But that's not what the quoted sentence said.
> Try visiting a Walmart and interacting with literally anyone. That's the average.
I've been to Walmart. Does that make me average? (You say literally anyone.) Do you think that Ivy Leaguers never go to Walmart?
> Let's not allow our egos to gatekeep who we consider intelligent, fellow HNians.
You say this in the same paragraph where you rip on Walmart customers.
Perhaps at that point if you stop reading after the first sentence, you could churn the entire article through AI to summarize it into a single sentence, and see if the invalid premise is core to the message?
Ars Technica has already trashed its reputation with the infamous controversy over publishing AI hallucinated quotations.
I couldn't decide whether the opening sentence of this article was so dumb that it had to be written by AI or so dumb that it had to be written by a human, and coming to a conclusion on that issue didn't seem worth it.
> you could churn the entire article through AI to summarize it
I don't do AI summaries, ever.
This was a lot harder to cheat before AI, but now the floodgates are open and grades and degrees earned post-AI are showing that they mean little.
Cheating on college tests should be a jailable criminal offense (similar to computer fraud) so that there is dignity in the degree again. Considering the money involved, I don't see why not.
But this probably won't happen, because many rich people are very happy to buy their degrees. See also [1]
https://stanforddaily.com/2026/04/09/the-real-reason-student...
How far back do you need to go to get to a time when degrees mattered?
and the rest are lying.
(With apologies to the original example of anomalous self-reporting)
And people wonder why I'm an AI hater.
I think education is incredibly important, but I understand that I'm going to have to retrain myself a little bit. A college degree can no-longer be assumed to be a proxy for having put in the effort to deeply study something.
Now what's the solution for this? I don't know, but we have made the mistake of conflating pieces of paper for expertise. And I say that as someone with 3 degrees.
Thinking back to my time as a professional pilot before I medicaled out and pivoted into tech, the FAA really (for all it's problems) has a pretty good system to train and test new pilots.
You have to have some hours with a certified instructor and some hours on your own. The tests to become a certified instructor are considered challenging, and many people fail. Then you take a written test, then you take a practical test. It's one on one. You and the examiner. And if you do not meet the standard, you fail. That's "ok." It's just fine to fail people who do poorly during a checkride. They go back, they get retrained, and they do it again.
If you have a lot of failures during training, you'll have to answer for them in interviews later on, but often times there's a sort of holistic treatment to it. If you busted a checkride 15 years ago, and have since been fine, you'll be ok. If it's a recurring theme, you'll have a hard time finding a job (and that's the right thing, IMO). But the format of "Written, Oral Exam, and Practical Exam" is the "right" model for making sure people know wtf they are doing.
How do we do that in tech? Hell if I know, maybe a proctored written exam, followed by an oral exam, then a project? But who knows.
Recently I saw Alex Karp (Palantir CEO) on one of those market shows (CNBC I think? Something like that) and he was upset at Dario's doom trolling. Why was he upset? Because he was worried that the populist rage towards AI was going to result in a wealth tax! And the thing is he's probably right! The thing these people fear the most, socialism, is the thing they're ironically probably going to make popular in the united states.
What that's going to look like I couldn't begin to imagine, but I think you might onto something about the AI industry not being all that capitalistic presently. Not an economist, but we're definitely in oligopoly territory here? Because there are only like 4-5 serious companies here.
I too may be accused of having the faith that it's "all going to work out" though. I think it will. I think when things start getting weird enough a bunch of creative people will figure out logical solutions that work in the moment and we'll run with those. Barring revolution, that's kind of how things have always been. Things get weird, then we figure out how to fix it or go to blows over it. I think going to blows domestically is pretty unlikely as of yet though despite the "dooooom" narrative that a lot of people tout.
That's optimistic. IMHO, it's more likely the libertarian tech-lords make sure we get some kind of hyper-capitalism, where you're either rich or bone-crushingly poor.
I don't know what we get? But if this is the rhetoric now, how many Luigis would we expect to see as things got really bad, and the vision you paint is really bad.
Also, after a couple assassinations of business leaders I think we'll see gun control laws like the US has never seen.
There are so many firearms in America even if there was bipartisan support for a gun ban it wouldn’t happen. You literally couldn’t get rid of them if you tried.
I actually don't really want that sort of outcome, I'd like to see some sort of libertarian socialism arise out of post-scarcity? But what that looks like? I don't know. 20 year old me would have had lots of ideas about how we "ought to be" but now, pushing 40? I am less certain of the things I "know."
Regardless, I don't see the really bad guys winning this one...
What? Where do you think all of the capital they are burning to stay afloat comes from? I'll give you a hint, it's from the capitalists.
It's the same thing with AI right now. Oh, you guys need another trillion to burn and you have no plan for profitability? Here's a blank check! Oh, your job is going to be eliminated and we have no coherent answer for what you should do next, because we want to take that job too? Not our problem.
Any semblance of capitalism we have is a shambling corpse barely getting around. Do you know who pulls on all the strings of our economy, gives out student aid loans to kids who just barely graduates high school? Who controls market interest rates, and gives massive multi billion dollar contracts to Companies like OpenAI, Anthropic, Facebook, Google, Microsoft? Who approved of the printing of the 50% increase in the dollar supply in the last 5 years?
Take a guess. I'll give you a hint, if you don't pay them on a yearly basis 30-50% of your paycheck, you go to prison.
Nothing in your provided definition of capitalism precludes government control of money supply, taxation, regulation, or even government contracts.
Private ownership still owns capital goods and private decision is the primary driver of investment, prices, production and distribution in most industries. This is why we have a hard time with things that are obviously good for society as a whole, like renewable energy, efficient transportation, and low-cost housing, that don't have a competitive short-term return on investment.
Ever taken a Redhat exam - such as Red Hat Certified Engineer? That's the right model. It's a hand-on exam with arbitrary exam objectives you have to meet (by any means possible, there is more than one right way to achieve desired state). There is no internet, devices, or anything else to help you. People that pass allow you to make assumptions about them - for example... without stackoverflow, github, or the internet at all - this person can configure virtual hosts with nginx, firewall rules, host VMs, and administer network concepts(virtual switches, routing) between them. To do that without any GUI or internet available - it means you understand those concepts very well, and can implement them blindfolded.
Any RHCEs want to chime in?
no internet-style indexing tho, so it's a lot of ctrl-f and hoping you pick the correct chapter -- which can be hard with a timelimit. on stuff I was strong in it was trivial, I'd already seen the docs and knew what to look for but on the few items I wasn't it was a lot of flailing.
Maybe I'm a weirdo, but part of me wishes I could be like a sysadmin in a powerplant or like something similar where we used tools like that and I'd have to get really good at using them.
Flying was like that, you had to know what you were doing, and aspects of tech touch that, but certainly not in the domain I'm currently in. Seriously, cool, I didn't even know this existed.
During my university time, all main exams were oral examinations: One professor, one protocol keeper, and the student. Duration: 1-2 hours.
The unemployment rate in the US is 3.7%. 38% of adults in the US have a college degree. 38% is hardly everyone, and if nobody can find a job, then the unemployment rate should be far higher than 3.7%, which is basically full employment.
For sure 20-y.o. adults have no degree and students are not count as unemployed.
I honestly pity watching my fellow software engineers rush to adapt to this new dystopia, just to stay afloat a little while longer before they are truly obsolete. It is quite tragic.
You know, I was reading an article about falling labor force participation rates, and one thing was pointed out was there seems to be a supply mismatch. Plenty of jobs are available.
I suspect the only retirement I’ll ever see will have to be via a shotgun.
Professor denounces mass AI fraud on an exam at Brown - https://news.ycombinator.com/item?id=48708991 - June 2026 (728 comments)
Academia has a long history of lagging behind acceptance of new cognitive tools where they claim to want to defend the students, but instead defend the assignments of the past at the expense of the students. Calculators were treated as threats to learning, even though they ultimately freed students to focus on higher-level math and provably improved their abilities across many different studies. Internet sources were dismissed as less legitimate than books, as if “published in an outdated book from the 70s” magically made it more trustworthy than the most scrutinized reference sources online.
It is not clear from the article exactly how much of this course falls into that category, but if the answers can be produced trivially with a prompt and chatgpt, then maybe memorizing that material is no longer the right educational target. Academia desperately needs to redesign itself around AI as a cognitive tool students should be trained to leverage. If a question is trivially answered by a prompt with it, then you need harder questions that actually require students to push beyond that. Simply removing AI from the equation, calling it cheating, and pretending that it isn't an ever-present asset people are expected to leverage in real life is naive and just repeats the mistakes of the past.
Of course, not all testing is good, but the written exam has survived and proven useful despite the internet age, I'm not sure an even better search engine really changes that.
Ask an LLM to "clean up this code applying SOLID principles" is much more prescriptive than "clean up this code".
AI can just give you the details like an encyclopedia article. You still have to know at a high level how something works and which article to look for.
AI is just a more forgiving way to ask something, compared to the searches of the past (e.g. searching a scientific journal database). And (for now at least), AI is a good way to avoid the ad-ridden web.
It's very clear from the (excellent) article linked by dang [1] what the exams required:
> This year, the economist decided that both the midterm and the final exams for his course would be of the take-home, closed-book type (there is a certain tradition of this at Ivy League schools). “It’s a very nice kind of exam, because as you’re giving students practically unlimited time to complete it, it lets you make it harder than normal, to see how far they can go.” In this case, Serrano changed some of the model assumptions they had seen in class, and asked students to demonstrate whether certain statements were true or false under the new assumptions.
[1] https://english.elpais.com/education/2026-06-28/ai-fraud-at-...
1a. Yes, college isn’t right for everyone, and testing the traditional way with paper and pencil certainly disadvantages some students who would be star performers in a real world setting but are not a good fit for college classes. Think “Good Will Hunting” type people. They do exist.
1b. However, there are certainly more people who only imagine themselves as Will Hunting type people and that they are just too smart for college, but the reality is that they are dumb, or didn’t learn the material. For every person who fails a test because they’re a genius who is a bad fit in the system, there are at least 10 idiots who imagine themselves geniuses, and they would have passed the class if only that PhD professor with all his book smarts had actually written the right kind of exam. Traditional schooling and testing doesn’t work for the extreme upper tail of intelligence, but it exists because it did quite well at educating and sorting the masses to support the Industrial Revolution.
1c. If college were only about the knowledge, you could learn most of it with internet access and a library card for much cheaper. The vast majority of people are in college for the credential, and the institution has to protect the signal of the credential.
2a. Most college assignments and exams are not a good reflection of full time employment. College credentials serve mainly as a networking aid and a signal to employers that you are compliant and competent enough to follow a professor’s instructions, and will likely be a compliant and competent employee, although the instructions might be different. If it is the AI who is the one who followed the professor’s instructions you water the signal down, and employers don’t want that. It is irrelevant that a student can copy and paste things into ChatGPT, and that ChatGPT can get the answer right on this test. That isn’t what college is supposed to signal.
2b. A problem is that I literally can’t write a test in most subjects now that I would expect a student to complete that can’t be completed by ChatGPT better and faster. I teach undergraduate math and a while ago we thought that since GPT-4o could get a C in calculus that we’d just raise the standard. Now Fable and GPT-5.5 can cruise to an A in literally every math course in our catalog, and they can also catch every tiny issue in an exam written by a human. But I have to teach these undergraduate subjects so that some students can go on to PhD studies so they can contribute to the field. If we just stop teaching undergraduate subjects then PhD production and novel research grinds to a halt and only a few fields will progress where an AI is capable of self improvement.
2c. I’ve seen that my best students know how to do do something by hand and use a computer to complement/increase their capabilities, not to cover over entire gaps. When you have literally zero skill in an area, you can’t spot when you got a totally bad output from the AI (these days usually because of a lack of context or bad prompting because the student didn’t understand the material, not because the AI wasn’t capable). Somehow I have to incentivize students to learn the material on their own, so that they can be a better user of AI in the future. And a really effective way to do that is a graded test in which AI is not available to help them.
So to try to tie it together, is that there is still some value to a college degree, at least until something better comes along. But that college degree is only useful when it is a signal about the person and not some other tool. And although AI is getting very capable, somehow we have to teach the lower stuff to build up to the higher stuff, so we do need to restrict the use of AI in some educational settings so that we can build a good foundation for future learning.
As to the point about old paper sources being considered more reliable than an internet site, I agree...
It's also why you can't just throw kids in front of Khan Academy.
I know. The desire to get a degree or the "threat" of wasted tuition and disappointed family members is really what gets most students to crack a book. Most people aren't self-motivated to learn. If people were self-motivated to learn the material, as opposed to getting a credential or some other reason for being in college, I wouldn't need to worry about cheaters.
This is how it has been for years and AI doesn't really stunt that as long as students actually care about learning the material. Those kids will be fine. The kids that are fulfilling a prereq will be fine, for the most part, but will walk on crutches their whole lives (use AI) which might be a reasonable choice.
What we lose is the signal of which kids are interested in learning and which kids aren't. Grades and effort used to be the signal, now I think it has to be more about engagement" who asks questions, who comes in with strange ideas, who actually is struggling at times (actually grappling with the material) or learning things outside of the assignments.
Big picture, I think teachers should focus on teaching the learners and trying to motivate more students to join them. That has been the real job all along, AI just lays it bare.
Oh fuck that bullshit. I was a dumbass to believed shit like that in school, and it didn't fucking free me. What happened was I was always calculator-dependent, which made higher-level math harder, because I was always distracting myself by operating the goddamn thing, and I never developed a very good intuition for arithmetic.
I fucking hate calculators in math classes.
Oh.. he’s blind.
Professors need to step up and teach value beyond what LLMs know (very possible with or without LLMS). Or get out of the way from those building on the field with LLMs.
If you’re teaching students something that LLMs can score 100 on you are not adequately teaching them something useful for them in the future.
"we cannot choose to become idiots"