This is the peak of civilization, get ready for regress.
in India, it’s common to pay someone to take exams. congratulations Doctor
But their ability to make Terminator-Robocop porn in under an hour is unmatched
trig test day, surprise!. ‘no calculators’. gaah, after all that time spent keying identities into my ti-83. still passed with my broken memory
what they’re doing to push back against the tech.
Classify it as a cheating tool.
Works in theory, but not in practice, as there are no tools that can tell 100% reliably if something was written by AI. Best way IMO to test students is via oral exams. Let them explain certain things and topics they allegedly wrote about in their thesis; that way you can quickly see who actually bothered to learn and understand and not only let their thesis write by AI.
They should bring back oral exams and blue books.
This is it. Stop it with these take-home test. Homework was always bullshit.
But then professors and teachers would have to think themselves instead of regurgitating the same lesson plan and worksheets from two decades ago
Hot take incoming: good public school teachers are criminally underpaid. Most teachers are paid exactly what they’re worth.
There was this cool initiative by a professor who is a friend of mine. He would give a pretty standard homework, but then the additional instructions were to complete said homework using an LLM. Then, the students would have to write, by hand, an analysis of all that the LLM got wrong, or could’ve done better. They then proceeded to discuss their analysis in class. Participating in the discussion with actual meaningful arguments was half of the points, the other half being the quality of the handwritten analysis.
It was more work, but at least the fuckers quickly appreciated that the machine was actually shit at doing their homework, and even if it could pass, it would be with the bare minimum. It also pruned the students who actually wanted to learn from the slackers who were just wasting their parent’s money.
I was with you up to the writing by hand part.
It’s also important for parents to genuinly take an interest in their children’s education. Help them understand why we don’t use AI for school-work. And be there for them when they need help so they don’t have to resort to AI when it feels hopeless.
I remember a study we all read when I was working as a sub teacher. Ages 7 - 12. How much time does an average parent spend talking to their child on an average day. Giving commands is not talking for the purpose of the study.
5 minutes. It was 5-6 minutes. It explains a lot doesn’t it.
There are plenty of negligent parents out there, but also plenty who don’t have the time because they gotta pay those bills that crept up on them, especially in this economy.
They are fine with cheating as long as you are using microsoft peoducts.
My uni gives everyone multiple chatbot accounts. (Paid by our tuition money no doubt)
Also all programming general events since I became a student have been “Make app for X purposes but with AI in it.”
I seen a TA open up their grok history when I asked why my answer to an exam question was wrong.
Universities don’t actually give shit about “academic integrity,” they simply want to buy bunch of microsoft products and call it a day.
Related Rant:
Recently had one professor try a “group exam” and worked with people requiring chatgpt to do “that one recursive fibonacci assignment.”
I think I have changed as a person.
No shit. LLMs are the anti-thesis of education. The basis of which is hands on practice. Using a glorified autocomplete-my-assignments should be barred from education without exception.
This is the tip of the iceberg. The world is facing a critical collapse of profession. We’re nearing a point that was foretold ins sci-fi where humans no longer understand how the machine works.Just that they can use it and it works.
Except sci-fi was too optimistic. Science fiction has actual AGI. We have glorified autocomplete. A non-intelligence.
This is worse than that too, we’re already at a point where most people don’t understand how the machine works… But overuse of AI is going to make it so that most people don’t know how to do simple tasks that everybody used to know. They’ll be non-functional without an internet connection.
it’s the equivalent of trying become a marathon running while you sit on your ass on the couch and and flying FPV drone for your ‘training’ and saying it’s the same thing.
It seems like a huge pervasive part of modern culture is that success and fortune equal never having to get your hands dirty, never worry about the details, and never learning to figure shit out because you can just pay somebody else to do it (or ask the “AI” to).
Obviously some amount of specialization and delegation is good. That’s how you get a society.
But to just exist passively is something else. It’s bad for us, and I don’t mean that as a moral judgment. Nobody needs certain skills to justify their existence. I mean it in the clinical sense, like that you can be sedentary with more than just your physical body.
It’s absolutely terrifying. I am a returning student to uni in my thirties and the only person not using any AI. They literally depend on it.
I just had a classmate the other day turn to me, frustrated, saying “You ever ask chat(gpt) a question and it gives you a whole, like, paragraph you then have to read? like, why can’t it simplify it?”
Did I mention I am an electrical/computer engineering double major? So yeah, even reading is too much for these kids. Future workforce is fucking cooked.
It took me an embarrassingly long time to realize that the “chat” slang was short for ChatGPT.
no, it’s for the chat audience in video streams
Ah ok, didn’t realize that was the original source. But now I wonder if it’ll eventually be supplanted as a reference to ChatGPT.
It depends on the context, it is slang for both. That’s why I put the parenthesis. Don’t be embarrassed, it’s hard to keep up with slang.
not true , young person here
Future workforce is fucking cooked.
Yep, and I predict that programmers actually understanding code (and especially being able to quickly and thoroughly review code), are becoming increasingly valuable (again?) in the future, when someone really has to guarantee what the AI actually generated (and let me tell you there are still so many stupid things the AI does…).
The new age COBOL programmers are anyone who can actually program.
We’re already seeing it a little, I know IBM just had to start massively hiring just in Feb
I have a friend who was frustrated that his programming exam was too hard (Python) and stated " why do I need to learn this? I can just use ai and get the job done ". We’re absolutely fugged.
It’s honestly disheartening. I understand our education system is not well in that tests aren’t actually very condusive to learning, but they treat the idea of learning a skill like it’s some obnoxious chore they just want over with so they never have to do it again when its like… bruh civilization/tech grows exponentially, y’all gotta learn your whole lives and it should be something you ENJOY it should give you pride to be good at something or understand a subject thoroughly.
i don’t even know what to begin doing about this problem but even if you pretend the environmental impacts are fine/manageable, I can’t help but think this shits gotta be destroyed for the future of humanity.
learning isn’t fun for the vast majority of people, it’s painful and miserable. they want to avoid it as much as possible.
I don’t even know what kind of exam it could have been but python is the least problematic language to learn, except ruby and go lang maybe, that are fairly simpler and intuitive. I exclude LUA because I only used it in TIC-80 and for these minigames. So far only senior devs are using AI effectively, if that is possible.
Unsure on the specifics, something for one of those online colleges. Yeah that’s exactly why it’s a concerning statement. This isn’t assembly, just scripting basics
Oh yes adult students are still motivated to yeet that LLM out of the windows, but young ones are just there.
Oh yes! We are fucked.
That’s right, It goes in the SquArE hole!
It genuinely feels that way yeah like idk what we do about this problem it’s bad, however bad you think it is it’s worse
I’ve seen some videos of Costco brawls between gen z men over Pokemon cards they want to flip. Surely the bar can’t get lower then that
I think, unfortunately, we are being lead by grifters creating a society of grifters i.e. the future generation is learning the lesson that the way to get ahead is by grifting.
I know this is gonna sound dramatic but if it keeps going? We’re looking at apocalyptic outcomes for the future of civilization.
the future generation is learning the lesson that the way to get ahead is by grifting.
So basically how it always was?
There’s a reason why there are billionaires…
But yeah it’s likely getting worse, like an elite and otherwise brain-dead propaganda following society (to exaggerate a little bit…). Initially after reading Adorno years ago, I never thought that the rather negative way he wrote, and something like the third Reich will not happen again, yet here we are repeating mistakes, fueled by accelerating climate-change (and the resulting conflicts)…
We are leaving (or rather have left) the most peaceful era of humanity, back to smashing each others heads without a real reason, leaving a more positive holistic utopian future… Sad.
I’m almost 50 doing a doctorate in engineering. ChatGPT is very helpful to do homework but won’t really teach you to do things. That said, I’m very glad to have it because this program would be so much harder without something to check my answers and understanding.
Using AI to help with assignments in education is like using a forklift in the gym. You are taking stats for everyone else and undermining your opportunity to learn and train your brain.
You are paying to get the opportunity to train your brain and educate yourself, a d are squandering the chance by using AI. It’s not the best idea.
…except that it doesn’t really “check your answers and understanding”…
Sure it does. It’s a calculator and can tell me if I did my statistics homework correctly. I can also ask it to summarize a concept I find difficult.
No, it actually doesn’t. How LLMs work is that it takes in written words and makes a sentence based on the likelihood of what the next word will be based on human readable text. That’s literally it.
Hence, there’s absolutely no guarantee that ChatGPT’s ‘review’ of your homework will always be 100% correct because it is probable that the answer was written incorrectly in the billions of lines of text it has been fed.
On the other hand, a calculator has been superficially wired for it’s purpose to process an input. 1 + 1 will always equal 2.
I’d wager your supervisor will be horrified to learn that you’re getting an LLM to learn from rather than your peers. This is why i absolutely hate that it’s being used as a substitute what essentially makes us human: art, music, research, learning etc.
It’s a tool that needs a licence because you need to know how to use it to complement your existing skills, not supplement it.
We’re allowed to use them. We just have to provide the prompt used and the answer given.
Ah very interesting. Still, it’s a shame cutting out the peer to peer discussions to expand one’s own learning and understanding.
To do an in person degree I’d have to quit my job. I have been waiting for years for a program like this. That said, doing it in person would be a better experience and I wish I’d had the opportunity to do something like that years ago.
TLDR; should have had my doctorate years ago. I went to Argentina and had to abandon my work when the economy collapsed.
If you want a calculator, try an actual calculator or a website that actually does math like Wolfram alpha.
A chatbot can’t actually do math. It can only give you an answer that looks right because it has no fundamental understanding of what “right” actually is.
Love Wolfram. I believe Stephen is a big proponent of this stuff? Found his blog on it (turns three years old next Monday) and curious for your thoughts on it.
In Just Two and a Half Months…
Early in January I wrote about the possibility of connecting ChatGPT to Wolfram|Alpha. And today—just two and a half months later—I’m excited to announce that it’s happened! Thanks to some heroic software engineering by our team and by OpenAI, ChatGPT can now call on Wolfram|Alpha—and Wolfram Language as well—to give it what we might think of as “computational superpowers”. It’s still very early days for all of this, but it’s already very impressive—and one can begin to see how amazingly powerful (and perhaps even revolutionary) what we can call “ChatGPT + Wolfram” can be.
https://writings.stephenwolfram.com/2023/03/chatgpt-gets-its-wolfram-superpowers/
With that naivety you are setting yourself up for failure.
deleted by creator
How old are you and what doctorates do you have?
We’re not in a dick-swinging contest here. At least I am not :) What a dumb way to deflect from the point I am making. If you have any understanding of the matter, you know that LLMs are only statistically accurate, and mathematically, that is useless.
I have a long, successful career. I am doing this program for personal growth and to create a challenge for myself. Please stop projecting your impressions on my life, I am my own person and very confident in what I have accomplished.
Edit: I got downvoted for being confident in my successes. Really?
This is an echo chamber. Logical and rational thoughts will be downvoted if they go against the narrative. Don’t take it personally.
Yeah I didn’t realize just how deeply people were into the projected angst before now. Thank you.
I am not objecting to you to learn something new. The naivety is what you display assuming you can use LLMs to “check your homework”.
until it hallucinates
who the fuck wants a calculator that gives out the wring answer every 5th time
The thing is; you have the experience and knowledge necessary to understand when you are being fed wrong information when it hallucinates. You have come this far not needing a machine to think for you and are not useless without it.
I had a networking lab in which we were faced with using an Aruba layer 3 switch, after learning Cisco all semester. The point was for us to research Aruba commands, and rather than even try to look up any actual references or the information provided by our professor they immediately tried to ask chat, which proceeded to keep giving them CLEARLY Cisco commands. Despite me expaining this they chose to ignore me until a half hour of failed commands later lead then to give up and just wait for me to try and figured it out.
Btw I am, by no means a genius. I actually think I am of average intelligence, and I’m okay with that. But I have become one of the smartest/go to people in most of my classes because once chatgpt cannot figure it out they genuinely don’t know what to do. They just don’t understand how to think for themselves.
Learned helplessness. There is nothing more shameful then just giving up after the obvious route is exhausted.
It’s like this a lot with family, like show me you made an effort to troubleshoot, otherwise you’re just demonstrating you think your time is more valuable then mine to do the whole rigamorole
this program would be so much harder without something to check my answers and understanding.
Do students these days not have study buddies?
Not when you’re in an online program and have to do it fully alone. Even when I was in a traditional situation I always had a job and never could do study groups. It’s specifically why I never studied stem previously.
Gen Alpha can barely look people in the eyes when they’re talking.
Thanks for sharing! Good luck with your doctorate!
Thank you. I appreciate the support among a sea of personal attacks. I didn’t realize that this place had such strong tankie vibes.
It’s not really surprising, and there are already quite a few studies on cognitive decline related to AI usage out there. I guess the wide scale effects will only be visible in a few decades, but I suspect it will look a lot like Idiocracy.
The bubble needs to pop yesterday
If this happens, a new great depression will come.
A huge part is the tools they’re using to “detect” AI use. My sister in law fought with Grammarly, which she was required to pay for and use by the school, to prove her paper wasn’t written by AI.
She spent twenty hours trying to “de-ai” a paper that wasn’t written by AI. The only things that worked were using bad grammar and poor sentence structure. The class was pass/fail, and probably for that reason.
So much ai slop in the professional world too
I got my first obviously AI email from a boss recently. The tone didn’t match his normal cadence of writing, it was sterile, repetitive, and could have really been summed up as, “Do you have additional information about item X that can help explain this to our customer?”
It was three paragraphs long.
So, I was wondering, will you reply efficiently with your own words or, in order for them to learn to respect your time, will you use AI to reply?
I walked into his office and answered verbally, lol.
That’s even more efficient =)
Then fail them. Oh, that would look bad? The "system "won’t allow it? Remind me, what’s your fucking job again?
BINGO!
I’ve heard much of the same from my friends who teach middle and high schoolers: most alarmingly that they can put information up on the board, ask a question about it, and the students don’t even connect that the answer is already in front of their eyes.
And sadly, a very common question they get is: “If AI can do this for me, why do I need to learn it in the first place?”
The worst part is that, in the short-term, the only recourse people have is suing social media and LLM companies, who are awash in cash and happy to settle, or throw their weight behind age verification, which in its various forms poses a security risk. Parents, clearly, are parking their kids in front of screens and unwilling to parent, so that’s not something you can depend upon.
I’m just glad I never procreated, but this problem is going to affect us all when these kids try to enter the work force and can’t actually do anything.
Reminds me a lot of teachers lying that we wouldn’t have calculators with us at all times.
Valid point: One needs to know how arithmetic works in order to get a computer/calculator to do it for you. This is fair; I use CAD software to design furniture, I do it parametrically, I have to solve problems like “if the overall width of the table top is 24 inches, the top overhangs by two inches all the way around, and the legs are an inch and a half wide, how long does the apron board between the legs need to be? And how long do I cut the board to add 3/4” long tenons?" I have to keep order of operations in mind there. But I write the expression and allow the computer to solve it.
Also valid point: I haven’t once done long division since middle school, because guess what? I have machines for that. I have had use for the concept of quotients and remainders…but I had to learn for myself how to get computers to calculate them using modulo operators. 5 / 2 = 2, 5 % 2 = 1. five divided by two is two remainder one. The algorithm of drawing the sideways L and putting one number under it and the other number to the left of it and then doing long division is not something I needed an entire semester of practice doing. You can’t convince me that was designed for my benefit, that was designed to keep me quiet.
Modern school is framed largely as a series of assignments one needs to get good grades in in order to be allowed to do something. A high school diploma is required to…practically be a citizen. “You have to get a good grade on this essay because it’s required for you to pass this class, which is required for you to graduate and get your diploma that we are legally required to force you to get.”
Children aren’t stupid, they know what bullshit is and they don’t like having their time wasted anymore than adults do. Children as a demographic have dozens of millennia of experience growing up into adults, they’ve been playing house and playing job since the invention of houses and jobs, they played cave and played hunt before that. They can feel when school isn’t like house or job and won’t help them do house or job. And it’s gotten to the point where that describes most of school, because they focus more on the difficulty of a class or test or curriculum than its usefulness.
So, I don’t want to reach but, are you saying education should be a function of the usefulness of the knowledge in a given everyday job?
Yes; and for K-12 education that “job” should be “being a citizen.”
Ask an elementary school teacher what the point of school is, they’ll say it’s to prepare the child for their adult life. Throughout school, ask that question: How does this lesson prepare students to live in the world? Elementary school teachers give pretty good answers: we’re teaching them to read so they can glean knowledge from anything from road signs to research papers, it’s probably the most powerful skill that can be taught. We’re teaching them to add and subtract because that’s how basically everything works. A question you’ll ask or be asked millions of times in your life is “how many?”
No ask a middle school teacher why we’re spending so much time on graphing functions. You know why? Because Texas Instruments lobbied to have their own products legislated into curricula.
I don’t think we are from the same country. I find very strange the notion of recognizing the importance of reading, but not the importance of writing, which implies some analysis. The very same source of so-called AI are precisely written knowledge and maths. Wasn’t that the spark of this conversation?
What worries me is the idea of being educated as a function of usefulness for an everyday job. That notion just assumes that people must comply at being a part in the production machinery, without free will of the very same people. Personally, I’d kill myself before having to spend 48 hours a week assembling cars, for example. Preparing for adult’s life should also consider being completely out of what the State was expecting from you as a citizen, and from the world they projected for their future adults.
I’m not against the teaching of writing skills. I think it’s currently done badly and I hope AI becomes such a problem that they change how it’s done.
From high school up until I dropped out of college for the second time, scholarly writing was approached in a really stupid way: Assign a topic the student doesn’t care about and have them “do a paper” on it. The boundary conditions of the paper are given in word, sentence, paragraph and/or page numbers. The alleged purpose for this exercise is to develop research skills; constructing arguments, supporting those arguments with vetted sources, drawing logical conclusions. In practice, the skills being built are padding with non-statements, use of MS Word’s rich text features and paying for an MLA handbook. The teacher doesn’t grade the paper based on the validity of sources or the quality of the arguments, it’s graded on the correctness of grammar, punctuation, spelling and formatting. Because to actually grade student papers based on the merit of their ideas is a monumental task; you’re asking a high school teacher to peer review 60 to 120 essays a month. So it doesn’t happen; you get students throwing inappropriate sentences in the middle of essays to see if the professor notices and often they don’t.
Have you ever heard the phrase “practice makes perfect?” It’s referring to Thorndike’s principle of exercise. A behavior is most strongly established through frequent stimulus and response. It’s why actors rehearse and athletes practice. But! “Practice” requires a feedback mechanism to correct wayward behaviors, you need a director or coach to correct anything wrong. When I was in 7th grade, I was handed half a page of sheet music for all-county band tryouts and told to take it home and practice it. I had 14 months of trumpet playing experience at this point, I wasn’t perfect at sight reading, so I took it home and learned to play it wrong. I showed up to the audition, confidently played a piece of music that only vaguely resembled what was on the page and did not make all-county band that year.
Of all the students currently in writing classes whose essays won’t be graded or even read by their teachers…how many of them are learning how to research wrong?
I’m most of the way toward convinced that long essays are required specifically to be a massive opportunity cost for the students. The cult of academia reveres the shut-in scholar who eschews personal life in favor of work and research, so they force all students to cosplay as this arch-saint by assigning pointless tasks that look like what a scholar does. If AI can unceremoniously kill the entrenched ritual so that actual education can resume, so much the better.
What’s the problem with using AI to write a research paper? The problem schools have with AI is it makes completing the assignment too fast and too easy. My problem is AI blatantly makes shit up, it’s an outright bad tool.
I’ve got a better idea: decrease the importance of writing research papers from scratch, and start them out reviewing the research of others. Hand them a paper and have them follow up the sources and see if it’s bullshit or not. Peer review is a massive part of science, right? So why don’t they ever teach students to do it?
There’s a lot more to life than house and job and learning how to learn is the most important thing you can get from grade school.
Explain to me how a decade of busywork teaches children how to learn.
It doesn’t.
It’s a consequence of government departments and school administration lacking knowledge of pedagogy forcing underpaid and overworked educators to teach their pupils how to pass standardized tests. It’s a consequence of a cultural and societal de-prioritization and outright disdain for critical thinking.
It’s a consequence of shifting the focus away from learning how to learn and towards learning how to work, a focus that has clearly impacted your view of education. You were failed by a system that was designed to fail you.
It clearly teaches children how to learn in a way that serves the ruling class: uncuriously, only on command, and not beyond specifications.
It teaches them to be good employees.
The kids who were 12 when the pandemic happened are now 18 and will be having their own kids soon.
and will be having their own kids soon
that’s a stretch. iirc people have kids at 32 these days
Fuck im old.
