Trying to teach in the age of the AI homework machine • Hacker News Board

Re: Trying to teach in the age of the AI homework machine

#631

Post by BobbyTables2 » Wed, May 28, 2025, 3:15 AM UTC

Balgair wrote: post #613

voilavilla wrote: post #604

Earlier quoted context omitted.

>> (where everyone cheated anyways) This is depressing. I'm late GenX, I didn't cheat in college (engineering, RPI), nor did my peers. Of course, there was very little writing of essays so that's probably why, not to mention all of our exams were in person paper-and-pencil (and this was 1986-1990, so no phones). Literally impossible to cheat. We did have study groups where people explained the homework to each other,…

Even before LLMs, if you walked into any frat and asked to see their test bank, you'd get thousands of files. Though not technically cheating, having every test a professor ever gave was a huge advantage. Especially since most profs would just reuse tests and HWs without any changes anyway. To my generation, it wasn't that cheating was a 'thing' as much as it was impossible to avoid. Profs were so lazy that any semi-…

Don’t know about frats, but I went to a lowly ranked “third tier” university and a “top 10” one.

While most of the classes were taught pretty well at both, the third tier ones were taught much better. Just couldn’t get an interview upon graduation despite near 4.0…

It is utterly bizarre that we use graduate research dollars to evaluate the quality of undergraduate education.

Re: Trying to teach in the age of the AI homework machine

#632

Post by thomastjeffery » Wed, May 28, 2025, 3:32 AM UTC

tgv wrote: post #535

thomastjeffery wrote: post #486

Earlier quoted context omitted.

Let me rephrase their point, then: The fact that AI can replace the work that you are measured on should tell you something about the measurement itself. The goal of learning should be to enrich the learner. Instead, the goal of learning is to pass measure. Success has been quietly replaced with victory. Now LLMs are here to call that bluff.

And learning does do that. It is an economic compromise, though. Most of us have average (or worse) teachers. I have the feeling that that's what your arguing against, not learning per se. > LLMs are here to call that bluff Students have been copying from e.g. encyclopedias for as long as anyone can remember. That doesn't mean that an encyclopedia removes the need to learn. Even rote memorization has its use. But it'…

The bluff I'm referring to here is the measurement. The notion that an educational experience can be meaningfully measured, and that such a measurement can be guaranteed well enough to prevent uneducated people from obtaining fraudulent certification.

I don't see learning as a compromise with economics. I see each as entirely irrelevant to the other. Certification is not a tool for learning; it is a tool for capitalism. A certification is nothing more than evidence of victory over the educational institution. Sure, the intended path to victory involves a lot of learning, but we humans can never be truly constrained to intended paths. That's a good thing: we shouldn't be.

Re: Trying to teach in the age of the AI homework machine

#633

Post by BrenBarn » Wed, May 28, 2025, 4:16 AM UTC

throwaway2037 wrote: post #299

avhception wrote: post #274

Earlier quoted context omitted.

I think one of the outcomes might be a devaluation of the certifications offered in the public job marketplace.

I can say from some working experience in the United States that way too many jobs require a university degree. I remember being an intern or my first job after uni (which I struggled a great deal to complete), looking around and thinking: "There is no way that all of these people need a uni degree to do their jobs." I couldn't believe how easy work was compared to my uni studies (it was hell). I felt like I was play…

I agree. I tend to think though that the best way forward is to ignore all of these education issues and just focus on raising the floor. The difference between a "good-paying job" and a "not-so-good-paying job" should be small, and everyone should be able to have a good life regardless of what job they have. Then people can choose to go to college if they want to learn about things, and maybe to learn about subjects related to a job they want, but not because they think it's a way to make more money.

Re: Trying to teach in the age of the AI homework machine

#634

Post by intended » Wed, May 28, 2025, 4:24 AM UTC

california-og wrote: post #509

intended wrote: post #503

Earlier quoted context omitted.

I get the feeling that you aren’t asking for the short version, because most people wouldn’t latch onto that point and create an account for it. Hmmm. An example - the inefficacy of Fact checking efforts. Fact checking is quintessentially counter speech, and we know that it has failed to stop the uptake and popularity of falsehoods. And I say this after speaking to people who work at fact checking orgs. However, this…

I think this 3-part essay might be relevant to your argument: https://www.e-flux.com/journal/147/623330/society-of-the-psy...

I’ve been recommending network propaganda recently. The book has the data that makes the case better than I can about structural issues in the information ecosystem.

Also started going through this legal essay (paper?) recently, Lies, Counter-lies, and Disinformation in the Marketplace of Ideas

https://www.repository.law.indiana.edu/cgi/viewcontent.cgi?a...

Re: Trying to teach in the age of the AI homework machine

#635

Post by voilavilla » Wed, May 28, 2025, 5:47 AM UTC

Balgair wrote: post #613

voilavilla wrote: post #604

Earlier quoted context omitted.

>> (where everyone cheated anyways) This is depressing. I'm late GenX, I didn't cheat in college (engineering, RPI), nor did my peers. Of course, there was very little writing of essays so that's probably why, not to mention all of our exams were in person paper-and-pencil (and this was 1986-1990, so no phones). Literally impossible to cheat. We did have study groups where people explained the homework to each other,…

Even before LLMs, if you walked into any frat and asked to see their test bank, you'd get thousands of files. Though not technically cheating, having every test a professor ever gave was a huge advantage. Especially since most profs would just reuse tests and HWs without any changes anyway. To my generation, it wasn't that cheating was a 'thing' as much as it was impossible to avoid. Profs were so lazy that any semi-…

Seems to me that studying a collection of every test over the years, without knowing what questions will be on the exam is... actually learning? >_<

Re: Trying to teach in the age of the AI homework machine

#636

Post by phito » Wed, May 28, 2025, 6:27 AM UTC

voilavilla wrote: post #635

Balgair wrote: post #613

Earlier quoted context omitted.

Even before LLMs, if you walked into any frat and asked to see their test bank, you'd get thousands of files. Though not technically cheating, having every test a professor ever gave was a huge advantage. Especially since most profs would just reuse tests and HWs without any changes anyway. To my generation, it wasn't that cheating was a 'thing' as much as it was impossible to avoid. Profs were so lazy that any semi-…

Seems to me that studying a collection of every test over the years, without knowing what questions will be on the exam is... actually learning? >_<

Not really. I had fellow students who understood nothing, could not program at all, but could tell you the answer to question 6 of the 2015 Java exam because they had memorized it all.

Re: Trying to teach in the age of the AI homework machine

#637

Post by throwaway290 » Wed, May 28, 2025, 6:28 AM UTC

hackable_sand wrote: post #609

throwaway290 wrote: post #492

Earlier quoted context omitted.

But that's very different with LLMs and that stuff. You don't need to know how to write an essay or write a song. That's kind of the point.

LLMs don't know how to write an essay or a song. To be able to direct an LLM you would need to know how to do those things yourself.

LLM don't "know" anything. It's a program. But those programs get fed so many (mostly stolen) essays and songs that it outputs a convincing essay or song.

So the end result is this. you don't need to know how to write an essay but you get an essay written for you. It's very different from using calculator and knowing math or using ProTools and knowing how to make a song.

Re: Trying to teach in the age of the AI homework machine

#638

Post by throwaway290 » Wed, May 28, 2025, 6:33 AM UTC

nkrisc wrote: post #497

throwaway290 wrote: post #362

Earlier quoted context omitted.

> If calculators are so good at math, is learning math itself useful? What's your answer? Surely it was proven to be "not useful"? I don't think I ever met a person who benefitted from knowing math now that everyone has a calculator in pocket. Other than maybe playing some games where if you do calculation on the fly you win

> Surely it was proven to be "not useful"? I don't think we're living in the same world. I have met plenty of people who, despite having a calculator, can't solve their own problems because they don't know what to do with it in order to solve their problem.

Examples pls. I know people who can do mental math and it's maybe only sometimes useful in some games.

Re: Trying to teach in the age of the AI homework machine

#639

Post by BrenBarn » Wed, May 28, 2025, 7:09 AM UTC

california-og wrote: post #294

halgir wrote: post #284

Earlier quoted context omitted.

> Some thesises had hallucinated sources, some had AI slop blogs as sources, the texts are robotic and boring. But should I fail them, out of principle on what the ideal University should be? No, you should fail them for turning in bad theses, just like you would before AI.

That's probably what should happen, but it's not what happens in reality. In grading I have to follow a very detailed grading matrix (made by some higher-ups) and the requirements for passing and getting the lowest grade are so incredibly low that it's almost impossible to fail, if the text even somewhat resembles a thesis. The only way I could fail a student, is if they cheated, plagiarised or fabricated stuff. The…

> In grading I have to follow a very detailed grading matrix (made by some higher-ups) and the requirements for passing and getting the lowest grade are so incredibly low that it's almost impossible to fail, if the text even somewhat resembles a thesis. The only way I could fail a student, is if they cheated, plagiarised or fabricated stuff.

This is another example of "AI is exacerbating existing problems". :-) That kind of grading policy is absurd and should never have existed in the first place, but now AI is really making that obvious.

Re: Trying to teach in the age of the AI homework machine

#640

Post by sshine » Wed, May 28, 2025, 7:57 AM UTC

TallonRain wrote: post #581

squigz wrote: post #416

Earlier quoted context omitted.

I hope by 'handwritten' you don't literally mean pen and paper?

Yes, pen and paper. The approach is to pseudocode the solution, minor syntax errors aren’t punished (and indeed are generally expected anyway). The point is to simply show that you understand and can work through the concepts involved, it’s not being literally compiled. Writing a small algorithm with pen & paper on programming exams in universities of all sizes was still common when I was in uni in the 2010s and ther…

Small algorithms, sure.

Outside of algorithms courses, the practice has diminishing practicality.

I once had to write a filesystem driver for a code, and it ended up being a little less than a thousand lines of C. I’m happy I didn’t need to do that on paper, it would have had limited value to my learning.