I have some sympathy for these kids. If LLMs were around when I was a student, I would've also used them to "speed up" my homework assignments then proceed to fail all my tests.
Now I work mostly with PhDs who were at the top of every academic environment they've ever been in. And yet I can see their thinking skills rapidly declining as well; many of them can no longer brainstorm, code, think deeply, or write without an LLM present doing 90% of the work. Many of them can no longer sit quietly for even 30 minutes just thinking on their own, which is a required skill for producing original thought.
For adults the cognitive decline won't be as measurable since there's no exams, and overall output volume will still be fine due to LLM help. But I do believe it's already happening absolutely everywhere around us. Honestly, I wanted to be in denial about it before but it's too obvious to ignore now.
I’m not noticing the decline in my own abilities any more than I had before using them. I finished undergrad 20 years ago and my once sharp math skills had been severely diminished within only 5-10 years. Just simple arithmetic and percentages that I could rapidly do in my head became dependent on calculators/spreadsheets. For all other trivia type knowledge, my brain has offloaded it to the internet RAM in my pocket. It’s a familiar feeling of when some question comes up and I think “oh, I used to know that, let me look it up”. Maybe I just already hit my personal floor of stupidity before LLMs.
However, I personally feel a huge mental burden of the state of communication. The contemporary version of it where I have a million threads and conversations im juggling at any given time. Emails, voicemail, chat, online, texts, personal, business, home, children, other family, friends, then there’s the variants like Messages, Messenger, WhatsApp, etc. And as overwhelming as it is for me, I’m super under connected than everyone else I know. I quit following most news and all sports, as I just don’t have the bandwidth for it.
My brain was molded preinternet and I feel like it’s reaching its max on the analog to digital conversion. Or at least it’s just a really lossy process.
Yeah, I'm 45 and I'm like you - no social media, relatively under connected, and still feel swamped constantly by emails and calls and especially texts. They eat up half my productive time every day, and most of them are things I'm looped in on that I don't even need to respond to.
Okay so let's say that's the new cognitive burden. The new escape hatch is "AI". Now you don't need to read your mail or write responses! Let an LLM handle that for you! And now your friends and coworkers will send you AI generated mail anyway, so if you're actually taking the time to read and respond to it yourself you're a chump, right?
Noise machines. Humans are noise machines. Ever try to sleep till noon and notice that everyone else seems like they can't feel alive unless they wake up and make the maximum amount of noise and racket possible? What could be better for a gibbering species of ground dwelling apes than a miraculous machine that gibbers for them, to point back and forth at each other?
I'm dumb as a rock and I don't have a PhD, but since ~1 year ago I started forcing myself to do small bits of coding and math manually.
I'm not noticing a "cognitive decline" per se, but I do see I'm a lot "lazier", even stuff that used to be routine when I started coding now feel heavy.
I do a similar version of this, where if I notice a mistake in generated code, I fix it manually (or at least attempt to) instead of telling Claude to fix it.
Same, but also because it feels like it takes longer for an LLM to do it. I think that's something people who are into gathering personal metrics should do - measure how long it takes to type a prompt / have the LLM fix things vs just doing it yourself.
I use an agent to generate a first-pass attempt, and then (deadlines willing), I manually read every line at least once so I understand what the code actually does.
Then I manually fix the inevitable slop that is mixed in with the good stuff, and only once the code is up to my personal standards do I send it.
This probably reduces my “AI performance boost” to 30-50% instead of the huge gains reported by others. But I retain the ability to reason about the codebase and use AI much more precisely when I’m trying to troubleshoot production outages or subtle bugs — something I notice the rest of my team struggles with, since adopting “agentic workflows” everywhere.
I think actively working to retain some cognitive flexibility and “muscle memory” around coding tasks is going to be rather advantageous in the long run.
This likely varies person by person or the way people adapted AI. For me AI replaced the boring part of writing code, but has not replaced the fun part of thinking about code and problem solving.
I'd argue that this is an adjustment period that society has to go through. The way we are using electronic devices today, in some years it will probably be looked at like smoking cigarettes. And I'd argue that a lot of the "decline" is due to a shift of skills away from things that mattered more in the past toward other things that are not measured/perceived by the older generation.
Interesting analogy. I believe regarding addictiveness they may be compared.
> a shift of skills away from things that mattered more in the past toward other things that are not measured/perceived by the older generation.
Do you have any ideas what these things might be? As someone in his twenties, I’m sometimes saddened by observing that some of the skills I acquired over a long time (e.g., writing, coding) may become obsolete or won’t be respected anymore just now that I‘m finally getting good at them.
Ages ago I had similar thoughts. Everything changed when I came to terms with the concept of change being the only constant. A bit of a cliché, perhaps, but profoundly true.
Eh, I think it's less like a cigarette and more like the car. We're not going back. Americans are famously less healthy the more car dependent they are, and now people walk/run as an explicit task to be healthy. People will start going to a "thinking" gym, or engaging in additional manual mental activities for sport, like we do with chess today.
This is an age-old argument actually, the same one was raised when the printing press was invented and reading became a more generally available skill.
Funny that you mention that. A month ago I started the Duolingo chess course, and just yesterday I noticed that my brain is clearer, more capable of deep thought than it has been in years. It's like stepping out of a fog. I also started CPAP recently, so it's hard to attribute the change to either, but I feel certain that the chess helped.
The interesting thing about jogging is I do my best thinking while jogging. I've found it impossible to do deep thinking while driving, as driving evidently requires higher functions of the brain. Jogging doesn't require any of that, I can jog deep in thought and have no recollection of the previous mile.
> I'd argue that this is an adjustment period that society has to go through.
I used to think like this until social media proved there are some tech innovations we just can’t adjust to. 10 years ago you would’ve never caught me supporting any sort of age based social media ban. Now? I don’t think it goes far enough. Fake news (actual fake news) and misinformation has only gotten worse with it as well. It’s so destructive.
The human is designed to interact with small groups, to understand several smaller groups, and perhaps to imagine a big group of smaller groups. In a literal sense, let's say 100 people per group. At that level the human can actually know and interact with them still. In a city of 100.000 it's still managable to feel you are related and involved to this group-of-groups. In a city of a million, you'll revert to only your own small group and have lost the connection to the collective.
The same goes for speed and quantity of input, as to what the human is designed for (not literally designed). Be it social media with it's infinite scrolling, cars racing by as opposed to looking out the window a few times per hour because you see someone/something, constant sound input if you live anywhere remotely busy or work in a busy office.
The point I'm trying to make is that the world used to be comprehensible for the human. Some understood a little complexer things, some only the simpler things. Now there is an overload of everything. So, most humans are in survival mode wether they know it or not. Hence the many seekin mindfullness etc
No matter, it's an observation, not a judgement or opinion on it. The world will just keep rushing forward. Some have a slight hand in the direction it goes for better (never) or for worse, but spiral it will.
I think there’s a major part of this conversation being omitted, though I am not saying you did it intentionally: “the attention economy.” We have gone from advertising to a system of creating addicts for profit
Definately agree, that was included in the "or for worse" in this sentence I wrote. As for creating the addicts, nobody had a masterplan. It's all the pieces spiraling together,
>> The world will just keep rushing forward. Some have a slight hand in the direction it goes for better (never) or for worse, but spiral it will.
The systems are too large and self-propulsing for anyone to really control. Consider the rainforest. How many millions of variables interact, nobody is in charge, everything influences everything in a billion different ways. You might say, well we can cut it down, so kind we can control it. Allright, let's continue to spiral. You might build a city there after a few years. Still in charge right. But it get's too hot because there's no vegitation, so you have to change again. And then we find that people keep getting strangely sick, and scientists find some special mushroom that survived and apparantly thrives on the mix of cut trees and diesel fumes and their spores in the air are poisonous. I made that up, but you get the idea hopefully.
LLMS didn’t invent cheating just made it easier. When you cheat you’re the one who cheats yourself because the point of an education is to learn, not complete the assignments and get high marks on tests alone. No one benefits and no one other than you is materially hurt by cheating, but you are absolutely the one who is hurt.
There’s no way to learn than to force the brain into adaptation which it is resistant to do through challenge and stress, just like your muscles. Similarly you can’t play e sports and get into physical condition any more than you can use LLMs to do your homework and learn.
It’s going to be a hard adjustment for a lot of people to recognize that letting the machine think for you is as healthy as smoking brain cigarettes.
The smart student uses the LLM as a proctor or provide challenges and feedback on attempts rather than an easy button. They make great tools for learning if they’re used as an adversarial or editorial tool. The future belongs to those who work to use the tools in ways that make themselves more efficacious, not those who use efficacious tools so they don’t have to work.
The problem with AI in an educational setting is when one is graded versus their students on things and things genuinely depend on those grades. Group projects also force those willing to do things without AI to go along with others in their group who'll use it regardless.
>The smart student uses the LLM as a proctor or provide challenges and feedback on attempts rather than an easy button.
Yeah, this is how we used wolframalpha for Math as students. Whatever we had to do, we did it ourself as a group of three. Afterwards we checked with Wolframaplha to see if we were correct. If there were any difference between us, we went line by line to find where the error appeared.
It was helpful, because we did it ourself, but because the work was graded, we had the security, that it is not a total failure.
But I like to add artwork to my presentations. My artistic skills have not advanced beyond 2nd grade. So I'll make a line sketch, and give to AI to "fix" it.
The results are nice and I use them.
I have no interest in learning how to do art well myself, so using AI for it is appropriate.
I have observed this in myself when I began to over-leverage AI in my workflows. I've since become more deliberate with what kinds of tasks I will use it for, although I still slip up.
With writing:
Things like brainstorming a plot line for a book with a custom GPT or Claude project that has all of my prior books in its knowledge? Works great.
Things like asking it to write a paragraph or chapter for me - I can rapidly feel my own writing skill, motivation, vocabulary, and ability to grasp/remember the resulting plotlines deteriorating. I don't use it for that anymore.
With studying:
I've been taking a couple of evening uni courses and the thing I found so great is that I've been forcing myself to think through the problems, and take my own notes in every lecture. I may then still get ChatGPT to help explain and reason through some of the concepts with me. And I have it review and 'grade' my assignments. But I refuse to ask it to start drafting answers.
With programming:
This one is tougher. When I am not very personally invested in a problem or codebase it becomes too easy to offload more parts to Claude, and when the company encourages 'vibing' to speed up velocity and you're reviewing and writing a higher influx of lower quality PRs, investment goes down. I still sometimes catch myself committing solutions I only _mostly_ grasp and the rest is hand-waving. A big part of it is a work culture thing.
For my own projects I make sure to understand and have a back-and-forth with the planning agent for each task, or write the first plan myself to go off of. When it comes to producing the code, I have to admit it is much easier to properly review parts of the codebase I am extra interested and knowledgeable in (backend in my case). The frontend I'm less well versed in and also admittedly less interested in, so I do sometimes fall into the trap of "Ehh it works, just commit it" with the goal of doing a thorough quality pass before actual release.
With all of the above, I can feel my ability to think, plan, reason, focus (and my vocabulary) suffer if I go over the line too much into agent offloading. For me keeping that balance is as much about maintaining my own long-term brain health as it is about producing good output. I imagine younger people growing up with AI today won't even know what that more capable (in my opinion) brain state feels like - to them, the AI-using brain will be the norm.
I've been wondering if there would be a benefit to inverting how we teach subjects now. Previously we would teach from the bottom, and build up. Semi-colon goes here, curly brace goes there, and then build up to architecture, systems, etc.
But this doesn't seem to make sense when someone comes to a topic with an LLM in-hand. They need to know high-level techniques, architecture, best practice, etc. As they pursue the topic they start to get down into the details, although probably never learn to do it fully independently.
I quite like this view because it paints a somewhat optimistic way forward from where we are now.
You can ask LLMs about high-level techniques, and their answers will usually be good enough.
What you can't get from LLMs is the taste and judgment, which you can only obtain by having a strong CS base and coding manually for years.
High-level techniques were never a problem. You could Google tens of articles on this topic. They are useless too, it's like learning how to drive a racing bicycle from reading a book. Sure, you will know a lot about nuances, but you will fail miserably when it comes to a real race.
The other day I just wanted to loop through characters in a std::string to copy data to a new string with a few escape characters (sending to peripheral device). Simple enough task for AI. I got a coroutine monstrocity back, with copies to std::array and a range based iterator, since I specified C++23. If I specified C++11, I would have received a:
char p = src.data();
while (p)
{
…
p++;
}
I had the experience to keep calling out AI to simplify and downgrade the solution to something primitive, which ended up smaller, faster, easier to maintain. Juniors with real world experience would not bother, they’ll take the first working AI result.
> which you can only obtain by having a strong CS base and coding manually for years.
I hope this isn’t the case. It is the route I took, but it also doesn’t seem to be a likely route going forward. Strong CS grounding is feasible for sure, but I have a hard time believing that a meaningful number of people will be spending the requisite years coding manually.
taste and judgment, which you can only obtain by having a strong CS base and coding manually for years.
I disagree, the definers of taste; art and food critics, movie and book reviewers, don’t need to have learned the craft by doing. Taste is a separate skill.
I can't speak for other disciplines, but for math and CS, both with a really heavy focus on abstraction, the final result of learning is to build a nice intuition on top of the abstractions we find useful/expressive. And to build the intuition, the old, usual, and perhaps the only way is to see and practice a lot of concrete examples, after which the motivation of building some abstraction can be understood, and after which the abstraction itself can be fully grasped.
e.g. The "group" abstraction requires one see a lot of int, polynomial, modular arithmetic etc. before knowing why we want such a thing. It's unskippable.
This idea sounds good at first, but if you look closer, it would just make workers, not experts who really understand. What we could do, and already do, is tweak the learned abstractions. In our field, it's easy to see: most of us first learned about computing abstractions, not how processors actually work, or started with Java, not assembler.
Plus, you can't teach math from top to bottom.
I keep trying to convince people that English majors and Philosophy majors will benefit the most from LLMs. English majors in particular, have been trained to be VERY exact in how they word things.
That awareness of how to structure the English language, it will benefit those who use LLMs.
Then again, maybe someone will just make a LLM that’s built to turn poor English and poor reasoning into excellent English and excellent reasoning. Maybe this is just a technical puzzle that needs solving.
I don't think you can learn high level techniques or architectures without first understanding the basics first. This means boring boiler plate coding.
I’m not sure. We’ve always had to pick the level of abstraction we start teaching at. Voltages, transistors, registers, assembly, C, etc. This feels like it could just be a progression of that.
We are already remote sensors and manipulators for the corporate and economic structures we operate under. You can't see it, but we are ants in a superorganism.
I think it varies tremendously from one role to the next. I'm a senior software engineer and LLMs, the way I'm using them, improve almost everything I do. I use them to write most of my code now, but first I spent twenty years writing code before LLMs came into existence and second writing code is like 5% of my job. Most of my job is research, investigation, and architecture. I treat LLMs just like a junior engineer. I give them clearly defined jobs that I could do on my own just fine, that I already spent years doing. The problem here is that students are using LLMs to automate everything BEFORE they become proficient at it themselves. Letting college students use LLMs for homework is like letting kindergarteners use calculators instead of counting on their fingers.
You cannot tell me that letting anyone do something for you does not affect the skills that you outsourced, unless you are some sort of a superhuman.
As an example, I have been drawing portraits for quite a few years now, and whenever I go on a hiatus and come back after a few months, I can notice my skill not being anywhere close to where it was before I stopped using it.
Sure, after 2 or 3 portraits they mostly come back because of the previous experience, but skill rust is a real thing, and if you think your coding skills are the same because you used to code 20 years but haven't coded for some time, you are probably just lying to yourself.
I recently switched back from a Tesla to an older car without permanently having a map visible. Suddenly my brain has to think about routes again and it definitively feels like my brain has to put in more effort again to handle it.
On the contrary, with the amount of times I went to ask for help and was failed pedagogically, plus not being able to afford tutoring like my peers had, I think access to an LLM would have genuinely boosted my grades.
I still did well, but I had gaps for which there was no help outside of the internet available.
The risk or difference is that tutoring helped people learn which they can use to do the work, whereas with only one or two different words an LLM will do the work (that proves you have learned) for you. A tutor has limits, but an LLM needs to be asked to set limits. And especially younger people are less likely to "punish themselves" like that.
Yes, I can churn out a lot more stuff as can most of my peers. Experiments etc are all way faster to run with coding agents. But I think the overall creativity and originality is a lot lower. I think this is what many people are facing, if you don't use LLMs your short term productivity is worse.
They're incredibly more productive. LLMs are amplifiers, so where they'd have branched and tried out N things, they can easily try 5N pathways of RnD. LLMs are extending the frontiers of science fast -- math -> phy -> chem -> bio in that order.
In my own experience, the only path I truly gain intellectual benefits is the one where I work closely with the LLM, test very narrow hypotheses, and leverage it for learning over producing.
Trying 5N paths is useful and sometimes yields interesting insights I’ll retain, but it’s not the rich, challenging, deeply engaging kind of process I find I need in order to develop useful knowledge and skills.
So yes it’s an accelerant for people who want stuff from me, but that doesn’t map directly to learning and building skills. I think that mismatching is really important.
To help learn I use LLMs to generate practice exams for whatever I'm trying to learn, then on the questions I struggle with have the LLMs explain the logic and point out my mistakes. I haven't been in college for over a decade, this is just for topics I'm curious about and want to learn. For any serious topic I recommend auditing the practice exams with a different LLM than the one used to generate to help reduce hallucinations. Seems to work well for me. I quite like reading the "thought" processes shown by DeepSeek.
I'm hearing different from PhDs. The bottleneck with much research isn't "trying out ideas" so much as it's all the bureaucratic minutiae, grants, mentoring PhD candidates, collaboration with other researchers, etc.
I've heard LLMs can be helpful in limited targeted ways. But not as some kind of "game changing" accelerant.
It’s creating a daemon and machine spirit filled world of Warhammer 40k. We already scarcely understand how the world works, but LLM use actively degrades cognitive ability that way it is used by a majority of people (The bringing a forklift to gym analogy).
To me it is crazy that you are being downvoted. My experience in academia was that an incredible amount of time was devoted to data cleansing analysis, coding, etc., which were completely non-core to the actual underlying academic pursuit.
There's an unnecessary feeling of fear that permeates any factual conversation on LLM's impact on science and engineering. You can just view the practitioner over the shoulder and see all the things they're able to do in a minute that would have taken days.
The downvotes are just a sign of the times. It's also something to observe and think about..
Yeah, it's a scary thought. I feel the pull of it every time I'm stuck on a code problem that I don't want to search solutions for and hand-code... and I also feel myself wanting to reach for the crutch of an LLM when I just have something boilerplate and easy to do. It's incredibly tempting to just ask the question and have the "thinking" done for you. Until you have actual skin in the game and realize that it doesn't reason, and its "thinking" is utter shit. Then it's like: you got addicted to cigarettes and now you have to quit, because this habit is poisonous. It really does lead very quickly to cognitive decline if you rely on them, or even think about asking them while you're writing code.
I doubt it. I'm stupid and I use LLMs a lot but I can still meditate for 30 minutes.
But apparently some of the smartest people in the world have lost the skill? But the commenter haven't, because why, they're 15 years older and thus immune to the same LLM-effects?
Plus, the issue with people having trouble sitting still for 30 minutes precede LLMs with decades.
I didn't say I'm immune to those effects, I'm including myself in this as well. (also, I'm not older than my colleagues).
Most people definitely can't meditate for 30 minutes, so if you can do this, it's very impressive. Regardless, being able to think about poorly-defined problems and build completely new mental models from nothing is genuinely a really hard and uncomfortable task. If you don't use the skill you'll lose it.
> Most people definitely can't meditate for 30 minutes, so if you can do this, it's very impressive.
Maybe not traditional meditation, but I have no problem taking a 30 minute plus walk with nothing but my thoughts. It’s actually when I do most of my thinking. The other is in the shower/sauna where devices don’t work anyway.
Why is it so hard to believe? The young adults now have grown up with short form media and instant gratification / dopamine hits from apps. It's vastly different than people of the same age just a few years ago.
Not saying everyone else is immune, but those a few years older have also had a period without it.
> I'm stupid and I use LLMs a lot but I can still meditate for 30 minutes.
> apparently some of the smartest people in the world have lost the skill?
> But the commenter haven't
> why?
Perhaps because a correlation you assumed was there (more smartness = more ability to sit still alone with one's thoughts), is not actually as strong as you thought? If one does not start with that assumption, there is no inherent conflict in the 3 pieces of evidence you cited.
Or perhaps because you are smarter than you give yourself credit for :)
>Now I work mostly with PhDs who were at the top of every academic environment they've ever been in. And yet I can see their thinking skills rapidly declining as well;
tomorrow most regular people's thinking skills will definitely be weaker than those of the LLMs of tomorrow. And physical skills in most cases will be weaker than those of the robots. That leads to the question - what would most people do?
Are you saying you leave it up to the LLM to judge whether your idea is good or not? Are you even human anymore?
(I am not saying LLMs can't be a good tool in evaluating ideas. To me, it sounds like you're firing off ideas all over, letting the LLMs judge what's good and what's not. Insane.)
The likely 'real' reason is hidden in one paragraph within the article and has nothing to do with the implication of the eye-catching title: "Both Garcia and Ranade have joined more than 1,300 UC faculty in signing a petition calling for the reinstatement of ACT and SAT standardized testing scores for STEM admissions in the UC system. The petition and its accompanying open letter detail similar concerns with students’ mathematical preparation."
Around COVID times many top universities experimented with removing test requirements from admissions, under an argument largely related to equity. It's been a failure everywhere, with many, if not most, universities already reversing it. As Yale put it, "Yale’s research from before and after the pandemic has consistently demonstrated that, among all application components, test scores are the single greatest predictor of a student’s future Yale grades. This is true even after controlling for family income and other demographic variables, and it is true for subject-based exams such as AP and IB, in addition to the ACT and SAT." [1]
That link is for an archive because that page has been removed. That's because they briefly experimented with a new 'test flexible' strategy where they allowed students to submit test scores or not, but then scrapped that altogether and went back to simply requiring test scores.
I'm having a difficult time imagining how an admissions event in 2021 materializes in the spring semester of 2026 in a class largely taken by first-year students.
In addition to overreliance on AI, Garcia also pointed out that many students are underprepared mathematically, a concern echoed by campus associate teaching professor Gireeja Ranade.
From the article discussed the other week:
Over three years — from fall 2021 to fall 2023 — the letter said, at least 20% of Berkeley first-semester calculus students who took a diagnostic exam showed deficits. “Basic mathematical fluency is analogous to literacy; without it, success in university-level STEM becomes structurally unattainable for students,” faculty wrote.
It's been steadily getting worse. The current article only looks at F's which conveniently hides if there has been a slope down. Additionally, kids entering HS in 2021/2022 would just now be hitting college.
SAT/ACT math is incredibly simplistic and at worst maybe contributed by not filtering as many out. Math scores have been declining nation wide for decades now, that’s been a big issue for a while.
If it's a lagging effect, then why is the year-over-year spike in failure rates happening not just in 1st/2nd year classes, but also in a 3rd/4th year class at the same time?
Berkeley chancellor told students to vote for 2020 California Proposition 16, which would've repeated 1996 Proposition 209 that banned race-based admission in public universities. Prop 16 failed. Subsequently, Cal started ignoring SAT/ACT scores. I have to think this was their alternative way of taking fewer Asian students, who average highest on that. Soon after I got an email from the same chancellor praising the change for bringing more racial diversity. The email included before and after numbers where % Asian decreased and all others increased.
I worked in admissions for an Ivy League. We used a vibes based system based on how compelling a sob story you could concoct whilst staying on the good side of fraud. Ie you wanna aim for drug addict father black trans queer activist with autism etc whatever is on the NYT frontpage today.
I know two companies whole line of business is just optimizing your sob story.
Things are changing a bit now though, it did get a bit ridiculous around the years 0 - 4 AF (Anno Floyd).
Core of the issue was always protecting legacy admissions and other grifts vs the deluge of Asian students with perfect test scores.
I wish I was lying
A famous MIT professor did a sabatical at our AI lab. He said it was "a joy to teach here, as you can rely on students being proficient in basic math as opposed to the US where you have to teach those explicitly or lose the class completely".
That was in the 1980s.
My first math exam as a CS undergraduate, 123 out of 129 students failed. The math department professors refused to dumb down their classes for CS students.
Math was core to the CS curicullum in those days. It would fade away over the next few decades to almost nothing. The main reason being the CS department wanted to popularize its uptake, and remove barriers that kept students from passing. There was also a major dose of interdepartemenral rivalry and academic politiking involved.
"More than 600 University of California faculty members, led by mathematicians at UC Berkeley, are calling on the system to reinstate standardized testing requirements for science, technology, engineering and mathematics applicants, saying that six years of test-free admissions has not reliably assessed readiness and professors are often teaching middle school math to incoming students."
>In addition, the guidelines state that “a typical GPA for a lower division course will fall in the range 2.8 – 3.3.” In spring 2026, both classes’ average grades were C-pluses, according to Berkeleytime, corresponding to a 2.3 GPA.
As a Cal alum, I am actually really glad to see they are holding the line on grade inflation. I worked my butt off to achieve the GPA I did, and it would really suck to see my labor devalued if Cal went the direction of e.g. Yale and started handing out 79% A's and A-minuses: https://yaledailynews.com/articles/professors-face-grading-d...
Unpopular opinion: turning public universities into an academic hunger games is diametrically opposed to their purpose for existing, which is to create an educated populace. Intentionally lowering the quality of instruction, as well as deliberately trying to trip students up on exams, is not improving educational outcomes for anyone. People who complain about "grade inflation" have completely lost sight of why public education exists in the first place.
Obviously a balance would be best, but as someone who went to a very grade-inflated school, I do believe that grade inflation gets in the way of education substantially. When you can get through classes with very little effort and understanding and know you will get a sufficient grade, many people will simply not learn the material deeply.
The material outcome is what should be the goal. Tests are a relatively brute way to try and determine how well the student understands the material, but conversations about grade inflation and "back in my day getting a grade was hard", and professors purposefully putting difficult questions (not in content but in presentation of the question) all betray the inherent goal being pursued.
Its all Goodhart's law problem, but we are missing the forest for the trees talking about grades and tests when what we want is people to be educated, and critical thinkers and competent in their area and due to a comprehensive way to evaluate that we end up talking about grade inflation or how Yale vs Berkeley gives letters at the end of a semester
Some of the exams in Berkeley were brutal, but they never felt like trick questions, they did on occasion require a level of mastery of the material which was extreme, but it never felt like someone was just trying to make the questions obtuse for the sake of it.
There are 10 different public universities in the UC system and 23 in the CSU system. The majority of them are not difficult to graduate. If you don't want a demanding education, don't go to a demanding university.
>Intentionally lowering the quality of instruction, as well as deliberately trying to trip students up on exams
I was happy with the quality of the instruction, and I didn't feel I was being "tripped up" on exams.
It's not about "hunger games", it's about challenging students to learn a lot of material and learn it well. Again, if that's not what you want, just don't attend.
It's interesting that it's specifically math-within-CS being discussed here. I can imagine a lot of students "just want to learn programming" (or similar), and see the math as a tedious distraction.
As a naturally curious person, nothing will stop me from learning about the topics that interest me. But school also taught me a lot of things that didn't interest me, and a lot of those things turned out to be useful anyway. I think if I had access to AI from a younger age, I'd have used it to skip learning the things I didn't care about, which would not have done me any favours.
Back in the university, I took both math and CS courses and a significant percentage of students seemed interested in neither math nor programming but rather in the jobs they would get afterwards. I didn't notice the same thing with math majors.
What a terribly ambiguous title. "Failing grades soar after xyz" makes it sound like xyz has helped what were previously terrible, failing grades become good ones.
I suspect the ambiguity might be part of making it "clickbaity", as it naturally causes you to wonder which meaning it's about and become more interested in reading.
My experience (n=1) is that while I'm definitely lazier on certain tasks, AI has opened up some much more complex tasks. There are many tasks which I still carry out which I don't trust AI with. Maybe it's a result of the codebase I work with being fairly complicated and math heavy, but I'd say the overall outcome for me has been: lazier application on the easy tasks, mind opening on the harder tasks.
Pity. I recently started a fun activity to rebrush my math my where I tries to solve problems while asking Gemini Live mode for confirmation and suggestions, sometimes step by step.
It kinda was fun, like a very patient professor stand right besides you. It was the one of the best math learning experience I've ever had, and you don't even need to send bribe/gift to Gemini to keep you in it's favor.
On the other hand, if you ask a LLM to completely finish the work without thinking it through by yourself, then it sounded like cheating, to yourself.
The main thing I use as a fallback is to keep thoughts connected in a Zettelkasten. This interacts well with AI assisted information gathering, while firing synapses whenever a connection can be made. I use Tiago Forte's method of organizing as needed within a loose org mode confederation of atomic notes.
Maths skills have been slowly falling even before the advent of LLMs. I have a story but this is anecdotical so take it with a grain of salt.
I was in my 3rd bachelor's year studying physics (France) and overheard a conversation between two of my teachers. They were discussing how they should modify the 1st year program to now include math, because he had been noticing how more and more students were failing the more math-heavy subjects like body and newtonian mechanics. He said that they should now teach (or re-teach) calculus to 1st year students, which was not taught when I entered college (it was assumed that you learned it in high school and we would only cover linear algebra in 1st year).
I can imagine things are only getting worse with students that can now get under the illusion that they know math because they have a tool that can do it for them. Which raises the question: should programs adapt to this, like we adapted to having calculators?
AI + Education is really interesting but also pretty tough to get right. Working on something that is hopefully going in the right direction: https://knowable.ca
The exams need to change. Now that we have LLMs the value a human can bring to a task has changed and it’s that new value that has to be tested.
It’s like testing your drawing ability in a photography class. The difference is that now nearly have subject and testing method we have has become obsolete. Drawings courses still exist as will traditional courses, but the main stream has changed and exams and schools need to adapt.
> Some of the numbers that you saw from the number of students who receive failing grades were because we caught them (cheating) and prosecuted them and are sending their cases to the center for student conduct,” Garcia said. According to Garcia, nearly 30 students in CS 10 were caught cheating on take-home exams in spring 2026.
In my uni, rates of honor code violations in introductory CS classes were high even before AI. I was a section-leader for the CS106 series at Stanford, and the honor code violations were common. In 2015, ~20% of one intro class was suspected of an honor code violation [1]. Often, the CS department comprised the majority of honor code violations in a given quarter.
There are several reasons for this:
1. Cheating in CS is easier to detect. MOSS [2] (authored by CS professor Alex Aiken) is a very effective tool at detecting plagiarism in coding assignments. Personally I witnessed more honor-code violations in math problem sets, but there was no feasible way for professors to detect this.
2. Problems in programming assignments are (usually) very tangibly wrong. I can bullshit my way through an essay with shoddy research, I can hand-wave a proof that is definitely wrong but will probably garner at least some points. But when your program is crashing or not compiling, and the due date is approaching, it produces a very immediate and undeniable sense of failure and pressure to cheat. The thing is, many students would get a decent chunk of credit even for failing code, but this is not immediately obvious.
3. The ability to cheat is more available. Math problem sets tend to change quarter by quarter. It's basically impossible to cheat on a prose essay short of straight up paying someone to write it for you, or fabricating sources. But for CS classes, especially at prominent universities, there are plenty of solutions online. Much of it is people who aren't event at Stanford implementing the assignments for fun or self-learning, and sharing it with their peers. Which, to be clear, isn't unethical or bad - it's the responsibility of Stanford students to refrain from looking at those solutions. But nonetheless, it's a contributing factor.
> MOSS [2] (authored by CS professor Alex Aiken) is a very effective tool at detecting plagiarism
He apparently also makes (I would assume a satisfying amount of) money selling the same technology to law firms for copyright/patent analysis: https://www.similix.com
(I love these ultra minimal HTML sites, ex. https://www.hwaci.com (SQLite commercial licensing) for another example. It just has this subtle smugness, like you either don't need any new clients or virtually all of the market is your client.)
I believe it’s still a single section, so probably around 250 (at least that’s about what it was when I was there a long time ago). Compared to the 1000+ who take 61A.
And if cheating was triggered using AI detectors, was it real?
AI detectors are pretty mid in practice - they tend to have a lot of false positives for "B" students who are okay, but can still be struggled to be more coherent than AIs are. There are some specific triggers that AIs are way more likely to do than students, but a lot of AI detectors will trigger on this "almost there, but you're still struggling" level of essay writing that might get a B, B-.
I could expect the same might be true for CS students even though I haven't seen how AI detectors work for CS/math homework.
You'd be amazed at how many students we know are obviously cheating because the logs reveal that they copy pasted a long, complete answer within seconds of opening a problem for the first time, full of sophisticated code constructs that we didn't teach them, and lot's of nicely formatted comments. Sometimes they even copy/paste the entire GPT output and then format it down.
This has been my wife’s experience as a college math professor. Instead of code it’s extremely formal problems with way more steps than the student normally performs using notation never taught in class.
It’s not that students didn’t cheat before, LLMs have just lowered the bar so far many can’t complete a live test in a class that requires effort.
AI has a way of exposing people. In this example, students who are there to get a degree from a prestigious institution, rather than to learn, are prone to take perceived shortcuts and proceed to come unstuck when their AI isn't there to do their work for them, such as in an exam.
It's too damn tempting to not use. You have a magical machine that, on command, will spit out the answer to your question in 10 seconds, whereas you'd need to spend hours to do the assignment the Good Old Fashioned Way. Even students who aren't just there for the prestigious degree are falling victim to this.
When you're up against a deadline - and unless you're very good at time management you're frequently up against a deadline - it's going to be an irresistible lever to pull.
In times past, cheating would mean copying an answer off the Internet or off a friend, both of which are easy to detect. More sophisticated cheaters might spend an hour rewriting the solution to make it less obvious they cheated, but at some point the cost of cheating (time + risk of getting caught) starts exceeding the cost of just doing the assignment. AI changes this - you get a customized answer that doesn't show up in a database with no extra work.
The thing is, students fail to realize just what using AI robs them of. Struggling with the assignment is the entire point. You don't learn if the assignments are too easy; you need to have some challenge to push your brain to understand the material more deeply and to build those pathways to apply the knowledge in novel ways. You become more efficient and effective over time as that knowledge settles in and you get more proficient - one of the reasons why time-bounded exams still make sense (being fast is also a proxy measure for understanding).
With AI, they fail later (during the exams), where as without using AI previously, they'd fail early and either course-correct, or drop out early (and suffer less of the consequences).
Not sure what the solution is - there's no possibility of stopping students using AI to complete their homework/assignments etc. But let me flip the question - do they need to be stopped? Why not let them fail at the exam? As long as the exam acts as a filter, their usage of AI to "cheat" their learning is inconsequential to anyone but themselves.
You are wrong. Some would have failed before, but not in the larger numbers. Before when they couldn’t complete an assignment they would try different things, seek a professor, or seek out friends to help explain. You could find answer keys to many assignments online, but that doesn’t feel like learning and wouldn’t even always answer your actual misunderstanding. It wasn’t perfectly tailored to your issue all the time.
Now the barrier to an answer is zero. They are basically watching a YouTube video on how to X, seeing step by step instructions feeling like they are doing it, and the moment they swing a real hammer they are whacking themselves in the crotch. It might get better after a few years, but this stuff is just now hitting mainstream for the masses. ChatGPT has only been in mainstream use for about 3 years.
This tracks, I have read that this generation is the first one since the 1800s that performs worse academically than the previous ones. Experts blamed screens and anything digital in the classroom.
Students need to be taught how to use AI apps efficently to learn. Their goal is not to solve problems, but to learn how to solve them. Let alone, they instead use AI apps to solve problems for them.
AI apps are very powerful for teaching. You just need to tell them to do that, and not to directly solve your problem.
It's funny that GP mentioned science fiction as a negative because what immediately springs to mind, for me, is Neal Stephenson's The Diamond Age. We literally have the tools to build his "Young Lady's Illustrated Primer" today. We just have to give today's AI a lesson plan to follow and ensure that it never gives the student the answers, and only keeps explaining the concepts in different ways until they click. Wrap that in an iPad app and you've essentially got the exact self-paced learning tool that Stephenson envisioned changing the world.
It depends how you use it. You can either get it to explain a concept, or do your homework for you. Its a bit like the decision students have to make as to whether to review their material before exams or go out partying.
Overall it just seems like a huge waste of money to piss away the huge tuition cost your parents probably paid.
You can use an llm to get out of doing homework but you can also use it to ask every question you would ever wanted in a 1-1 tutoring session. The problem is kids will use it to cheat on their homework. If we can’t deal with that problem then a ban is necessary. But these things can be phenomenal teachers if you use them properly.
As an educator, this is exactly what I struggle with. I'm pulling out all the stops to give students every chance to do the hard work and not lean on AI. But there's a good chunk of the class who don't listen to reason. I haven't figured it out yet. They know, logically, they can't pass an interview, but that's apparently a "tomorrow" problem.
The smart ones either use it not at all, or use it to positive effect, like you're saying.
They are great for self-teaching and great to cheat and not learn anything, depending on how you use them.
Main problem is that the technology was very disruptive for education and nobody has figured out yet how to utilize it at scale for schools and universities.
I dont think ai is good enough for it coding or any other work once i told ai a problem and he generated an entire solution which i used and it was broken. You should never use ai like it treat is like a helper write a function for code and then ask if everything is correct and if something can improve read documentations understand how its working under the code if everything is correct then only deploy or build.
“I’m a strong, strong opponent of what Harvard is doing to say that only a fraction of students can earn A’s,” Garcia said. “I think you should have clear standards for what an A means, and then give tons of opportunity for people … to get to that A bar without lowering the standard. So everybody who’s curving is hiding that effect. It’s completely hiding that effect, and it’s pretending as if nothing’s wrong, and something is definitely wrong.”
To do this, you have to be a professor who has a strong idea of what subject mastery looks like. Not available to most.
I'm confused by Garcia's statement as well because CS@Cal traditionally uses a bell curve which is even stricter than Harvard's changes, because Harvard doesn't have the same stringent GPA requirements to declare a concentration unlike declaring an impacted major at L&S Cal.
Anyone with a pulse can declare a CS concentration at Harvard and muddle by (you actually need to try in order to get a C/C-). Of course, GPAs are calculated differently at Harvard compared to other universities, as a B- is treated at a 2.67 but most other programs treat that as a C+.
In a broad sense, this distinction between Harvard and Cal is the distinction between an old money Ivy and a flagship state school. One exists to propagate a social hierarchy, and the other aims to allow all entrants to succeed.
Ironically, the techniques of the latter yield the results of the first, but everybody gets to keep a pure heart.
Grades only matter as much as being able to transfer just to the real world.
People can use AI to outsource their learning, but if they use ai to outsource their understanding they just set themselves up to fail even more.
From what I’ve seen, how students are using ai (not that they are using ai) is making them less prepared for the real world, which unfortunately is changing faster than ever at the same time to create double impact.
The kids don't care about the integrity of the systems or their educations because they can see that all the benefits of a traditional education and career are predicated on a future that probably won't exist.
It's a rational response to entrenched elites that prevent realization of the very social contracts they push on the youth (hard work will equal success, home ownership is a fundamental, etc).
Combined with the looming specter of climate doom, and watching the adults do nothing about it, treating preparation for a conventional career as a scam to be counter-scammed makes a certain sense.
Average school system has been lacking for a very long time, overhauling it to focus on kids current interests, while sneaking in the other stuff, might now be possible and cheaper to realize with our new tech.
It's not just 'a person' or 'a student', we as a collective become more dumb. Very simple example to highlight this: Most developers use(d?) stackoverflow. Everything related to software development is stored there. The LLM's trained on it. Now a huge set of developers now longer go to stackoverflow to get answers. Or add to the collective. Stackoverflow is losing money (ad revenue). If / when stackoverflow goes away we will lose a huge amount of collective information on software development. We, as a group, will take a huge step back.
All of this makes me selfishly excited for my own future. It's glaringly obvious that anyone who's a heavy user of LLMs is atrophying their skills in real-time. I have yet to meet a single person for whom it's not the case.
But I essentially completely stopped using them for software engineering (why isn't really relevant, but it's not because od this skill atrophy). So as the skills of everyone else is diminishing, mine is proportionally raising.
It has never been easier to get better than others. You don't need to put in more effort, just the same effort as you always have, and others will do the job of losing their skills for your own benefit.
Sorry, but I don't think AI is entirely to blame here. When I graduated from a CS program at a top-10 school, I felt frustrated that the professors didn't ever teach. They had slides. They read off slides, verbatim. They explained things sometimes if you asked them, but most often in a very elitist and condescending tone. Like in the movie Good Will Hunting, you could have learned nearly all of it and more by borrowing those books for free from the library. Or, just opening a complex OSS project and learning to contribute.
And quite honestly. It shows in the CS grad population too. A lot of us are condescending toward anything that doesn't make sense to us. But, I digress.
The best engineers I've worked with are all non traditional backgrounds, non degree or degree holders from non elite schools. They think differently, they tinker, they are incredibly nice and patient, and do it for the love of connecting humans to technology.
Look up the names mentioned in the article. Garcia, Ranade, Nelson. All of them are involved with highly theoretical mathematics and scientific computing. Just because you're good at 1 thing does not mean you are qualified to teach. And none of these professors are trained or taught or graded or performance managed on how they teach. For most of them, its just required that they spend 10% of their time in the classroom lecturing.
Let's be honest about another thing. 99% of EECS graduates, even from elite schools, are wrangling objects and their relationships to a graph. Simply put, we're all just a bunch of glorified JSON massage therapists. It just so happens that we get paid well for it, and we hold that over people. The same happens in the classroom.
I think in order to facilitate a healthy, educational environment for young adults, we as adults must encourage, motivate and make that environment fun and practical. We force feed binary trees and the compiler AST's, but we need to make it fun. It's like the commonly accepted saying: Schools kill creativity :(.
University education is weird. Research profs (who make up a large fraction of all profs in a typical R1 institution), are hired for research ability and are only minimally evaluated on teaching ability. Furthermore, few research profs actually receive any kind of mandatory training on how to teach; a typical research prof might be assigned a course to teach and then just let loose to do so on the first day of the semester. If a prof actually cares they may attend some optional teaching training - but I stress that these are optional at many of the institutions I know of. (I suppose if someone gets really bad teaching evals they may be advised to attend said trainings - but for a tenured prof, that's just advice).
Worse, a decent chunk of research profs will treat teaching as a burden that just has to be done - a distraction from their exciting world-changing research. So, you get attitudes like the ones you mentioned.
I'm actually not sure why the system is set up to assume that profs who are good at research are automatically suited to teach classes, but that is how it's setup.
I really wonder if it's important to learn all that low-level stuff at this point. Most programmers today will never write a binary tree or a hash table. Modern high-performance ones are generic components you get from libraries. Even MIT gave up on teaching from Structure and Interpretation of Computer Programs.
I got all that stuff. I've wired up a 4-bit adder on a solderless breadboard for an architecture class. I used to have a well-thumbed copy of Knuth handy. I've designed and built a switching power supply. But I'm not up to date on using Claude Code, and should be.
I think it is important to learn how to implement it because it gives the student an opportunity to learn precisely because it's been done countless times and debated over to death. There are many analyses and if one doesn't click, maybe another one will. A student can learn how to analyze the algorithms and try out different implementations to assess differences in performance.
Of course, if a student just breezes through it then I would agree. That would make no sense.
IMHO, I think it's good to have some exposure to low-level stuff. There's a good amount of work you can't do without understanding the low-level stuff, but there's more work you can't do well without having at least an idea of the low-level stuff.
Start the kids off with high level stuff, but make them do some embedded systems on their way through. At least for an engineering degree. Also, do a bit of lower level communications somewhere in there; expose them to tcpdump/ wireshark, but they need not develop expertise.
> According to Berkeleytime, 35.3% of CS 10 students and 10.6% of CS 61A students received F’s in spring 2026. In spring 2025 and spring 2024, the percentage of F’s did not exceed 10% for either class.
I don't think instruction would've changed drastically in the last year though.
The fact that you are talking about Dan Garcia, a huge figure in computing education research and an excellent teacher, and the Beauty and Joy of Computing curriculum makes this hilarious. You should look up some details about both.
> I felt frustrated that the professors didn't ever teach. They had slides. They read off slides, verbatim. They explained things sometimes if you asked them, but most often in a very elitist and condescending tone
+10000. The goddamn slides. If I were a student now going to engineering school, I'd basically take the slides and throw them into NotebookLM and get way better lectures. Then I'd ask claude or GPT all my hard questions. Hell, I'd get the PDF version of my textbooks and do the same.
The number of lectures actually worthy of your time was so low.
I try to lecture as little as possible. No slides. Quick highlights discussion of the reading, maybe a coding demo, and then students work on coding challenges in class, in groups if they want. I circulate and help out. I'm lucky to have small class sizes at this university. I couldn't pull it off in a class of 300.
Garcia and Ranade are Teaching Professors. Their primary responsibility is to teach, develop curriculum, and do pedagogical research. This job posting explains: https://saberbio.wildapricot.org/Job-board/12919068
Professors suddenly realized everyone was cheating and started paying attention, but the cheating isn't new ... A lot of faculty are happy when their students get good grades because they interpret it as I'm such a good teacher instead of I should pay more attention to how they cheat. AI woke some of them up to reality.
A reckoning is coming for school. Learning the rote stuff is no longer essential. Now they need to learn, how to teach "how to think". How to invent, how to be creative. Art++, Woodshop++, Math--
"According to Berkeleytime, 35.3% of CS 10 students and 10.6% of CS 61A students received F’s in spring 2026"
Alternatively, more students are taking CS10 and CS61A irrespective of aptitude.
Anyone can code, but not everyone can become an employable SWE.
Anyone who has first or second hand experience with Cal or any other university knows how impacted CS majors have become, and how everyone is attempting to become a CS major because it's the easiest path to multiple high paying white collar careers.
And in all honesty, it's not like CS@Cal never had weedout classes (I remember CS70, CS61B, and Math54 had reputations of being the L&S weedout classes).
The question comes sooner than the students being tested on the job market. Another possibility is that dropping standardized testing was a net bad idea.
At UC Berkeley L&S, students are undeclared by default, and everyone is incentivized to take the intro CS classes (CS10, CS61A) irrespective of aptitude because worst case they can declare a CS minor or use the classes for other adjacent degrees (eg. Applied Math, Data Science).
Additionally, while Cal doesn't require standardized tests, most students who applied and attended already took the SAT, ACT, and APs becuase they cross-applied to other universities as well. This is reflected in UC Berkeley's HS Weighted GPA being in the 4.31-4.65 range [0], which means most students will have taken at least 6 AP classes.
Hell, I attended an Ivy and even then Cal was a target program for me, as well as my peers. If I didn't get into my Ivy I would have ended up at Cal and ended up in the same position.
Spring 2026 saw a marked shift in student performance. We saw it in intro physics courses on the East coast too. I bet anyone who cared to look saw it.
I'm not denying that. I'm just wondering if anyone measured if there is a correlation effect being induced by CS major declaration requirements.
Barely over a decade ago, CS tended to be a large but not too large major by enrollment in most universities yet nowadays it is the most in-demand major in most universities. You can see this at Stanford [0], but most other programs as well.
I read something interesting yesterday on the subject of AI in education (though, it has consequences to broader society too):
The goal of education is to impart knowledge in the student, preferably correct knowledge. The goal of an LLM is to produce an output that is convincingly human. It's not even that they're opposed, as much as they're ships for whom Polaris is in a completely different direction.
"Hallucinations" as they're called, or more plainly stated when the machine makes some shit up, are perfectly understandable in this context, as are the struggles of every single AI firm to get rid of them. Namely: the machine is functioning exactly as it is designed to, so how can you possibly fix it? It's working. The goal of an LLM is to produce text that passes for human, and apart from the obvious LLM tells, it largely does. Like say what you will about their lack of intelligence, the writing is solid. It's grammatically correct, spelling is dead on, what have you.
It reminds me of the famous phrase from Chomsky: Colorless green ideas sleep furiously. A sentence which is perfectly grammatically valid but is also completely devoid of meaning. An LLM would write that sentence, and it would be working correctly.
All of that to say: for all the things they CAN do and CAN be used for, I think we have to draw a hard line at education. I just don't think AI has a place in it. Of course that presumes that the goal of education is to, well, educate people, and especially here in the States but also abroad, we have been putting other interests, especially capital, far ahead of that for decades. I expect no different here.
And before someone comes in to go "WELL HOW DO YOU THINK YOU'RE GONNA STOP IT LUDDITE IT'S THE FUTUUUUUURE" yes, I'm sure as long as these exist and are available to people tech literate enough to access and use them, whatever that means into the far flung future, they will be a factor. Just like cheating, just like plagiarism, just like everything else that will get you kicked out of school. And the answer is the same: it will be stopped by institutions, imperfectly, and it will also happen anyway and with the same consequence: those responsible will mostly be harming themselves for short-term gains.
Respectfully, I disagree. I think there's absolutely a case for AI being encouraged in younger people, and there's room for these tools. I've been leaning on LLMs for side learning in side projects, and it has concretely helped me with conceptual questions about math and Vulkan as I've been trying to learn some graphics basics with side projects.
I would grant: I was not the most studious kid, I could definitely stand to learn how to read code a lot more effectively than I do; but I have found being able to ask a computer, "what portions of the Vulkan Programming Guide are less relevant with Vulkan's design changes since the release" pointing me to the dynamic rendering extensions and placing it into context, with inline code and links out to useful blog posts for additional reading, that sort of thing is very helpful.
Working on a prototype before I was trying to learn Vulkan, I was using it to explore SDL_GPU's API which definitely had some gaps in its documentation. Granted again, I could have referenced the sample code - I am sure you'll prefer I'd have done that - but it helped to get information about what each piece of the API was doing, and gave reasonable results that made sense and did inform me enough to understand what I was doing, turning much of that into an interactive learning of basic GPU programming for graphics. Where the AI hallucinated, it was often on things like method names, which I was able to read through and find the methods it was intending to name. (This only occurred once or twice when I was learning).
Unrelated, but adding the C macro syntax and nesting macros, which I could have an LLM explain inline and link the GNU manual. Never got that taught to me in a C course. Man, computers are complicated!
These have not replaced textbooks; I have been using them alongside textbooks and handwriting code for practice, and they work as a very good complement. I also sometimes use them to unblock me - I don't know CMake very well and lean on AI to do CMake, so I can focus on learning C++ and graphics, which is my primary objective right now.
I would add too, I have for fun given it prompts about various topics I learned in university, and I often will get answers that are bang-on what I learned in university undergraduate courses - the topics I tried were welfare state taxonomies, distributed systems, disk storage performance, filesystem layouts and internals.
Boy, this would've been cool for me as a kid. There's just so much information right there, and pointing you to topics and textbooks a couple questions away, I wish I had these tools. I was a curious kid in a terrible MAGA-esque family that was deeply uncurious about the world, had no knowledge of any advanced subject and basically mocked me for trying to learn more about stuff. And you go to the school library and it's all kids shit, not even an option to try and reach out for more. Now smart kids might be able to go just learn shit very freely and be pointed to textbooks, and go pirate them off some Russian site, and start learning and go tutor themselves, as I'm doing today as an adult.
At least knowing myself and knowing if there's another kid like me, I think they would deeply enjoy having a natural language encyclopedia, if we can get it as close to that as possible. I think even with some error inherent, if the tools can be often and directionally correct, that would be a plus. I went to university, and the professors there hallucinated some things so embarrassing it should bar them from teaching, for the standards people hold LLMs to! i.e., sanitizing conspiracy theories that Android records all language through the microphone therefore iOS is better, Apple Silicon is more battery efficient because it is RISC and not CISC. Got a terrible history of computer graphics technology you'd know was slanted if you watch the 8 Bit Guy on YouTube. Rubbish.
The thing that worries me, and what this article really talks about, are the kids that just don't give a shit. They are not new - when I went to high school, before AI, stupid kids would copy code off the internet. I think AI probably makes it worse because it makes it harder to call out and enforce against it, and agreed, that should be stopped. But to me, that is mainly a cultural problem. Too many Americans are completely uncurious and just spout garbage; there are a lot of kids who grow up in that cesspool and are going to grow up uncurious, and then AI acts as a shortcut rather than a vehicle of curiosity.
And granted, maybe AI is less useful when you are in a structured environment - but the structured environment has its downsides. Even in that environment many of the TAs were clueless and unhelpful, or just too damn busy or already too knowledgeable to meet students where they were at. Again, talk about hallucinations with TAs! Many times in my experience. And that's all to say nothing about getting people to not just do homework but actually go get curious about things and try stuff that isn't required of them.
I think there will be some culture that remains curious, and has these tools, will come to grips with where they can help, where they go wrong, how to balance it with other learning methods; and I think they are going to have kids that absorb a lot more knowledge and get to play with topics and learn things, faster, to each kids' interest, perhaps even individualized tutoring at better scale - I hope that is possible.
I hope the United States as well, but maybe not, because holy cow our culture and attitudes are plainly terrible these days. Your comment is pretty representative of how most people react if I suggest this or talk about my own experiences I'm describing here. But I hope at least I'm arguing something comprehensive here. There is too little conversation beyond hyperbolic nonsense on the internet; I consider "FUTURE LUDDITE" etc. to be in that realm.
I will add, too, although less relevant to education than just generally - for all the talk that these tools must be useless and incorrect, that just plainly does not map to my experience using these tools. AI can chew through a debug log on a custom system and pick out root causes on behaviors very effectively, in my experience.
It is just hard to reconcile that denigration of AI with the typical experience I have using these tools in the real world. It is not omnipotent or God, but it can effectively assist in work. There is a certain cognitive dissonance I feel when I walk away from using the tool to help accomplish particular tasks, then hear over and over people say the technology is fundamentally useless and fundamentally does not work. I guess I am just not enough of an academic to understand how something can accomplish work yet fundamentally isn't, somehow.
why would I as a child ever develop the imagination needed to actively engage with AI tools in the manner you describe? those AI tools take care of the imagining for me.
We're going to find that LLM usage has even worse effects on the mind than the horrific effects we're just starting to be certain of from social media. I'm just not going to use either. See you lads on the other side.
Probably not a bad thing, the coursework is antiquated and meeting students with new advanced tools and the awareness of AI's impact on things in the coming future
I imagine there is some apathy and laziness here but idk how unjustified it is
"Noooooo you need to manually code on paper in assembly"
Alright, well maybe the CS grads need to, but why expect that of everyone else
Now I work mostly with PhDs who were at the top of every academic environment they've ever been in. And yet I can see their thinking skills rapidly declining as well; many of them can no longer brainstorm, code, think deeply, or write without an LLM present doing 90% of the work. Many of them can no longer sit quietly for even 30 minutes just thinking on their own, which is a required skill for producing original thought.
For adults the cognitive decline won't be as measurable since there's no exams, and overall output volume will still be fine due to LLM help. But I do believe it's already happening absolutely everywhere around us. Honestly, I wanted to be in denial about it before but it's too obvious to ignore now.
However, I personally feel a huge mental burden of the state of communication. The contemporary version of it where I have a million threads and conversations im juggling at any given time. Emails, voicemail, chat, online, texts, personal, business, home, children, other family, friends, then there’s the variants like Messages, Messenger, WhatsApp, etc. And as overwhelming as it is for me, I’m super under connected than everyone else I know. I quit following most news and all sports, as I just don’t have the bandwidth for it.
My brain was molded preinternet and I feel like it’s reaching its max on the analog to digital conversion. Or at least it’s just a really lossy process.
Okay so let's say that's the new cognitive burden. The new escape hatch is "AI". Now you don't need to read your mail or write responses! Let an LLM handle that for you! And now your friends and coworkers will send you AI generated mail anyway, so if you're actually taking the time to read and respond to it yourself you're a chump, right?
Noise machines. Humans are noise machines. Ever try to sleep till noon and notice that everyone else seems like they can't feel alive unless they wake up and make the maximum amount of noise and racket possible? What could be better for a gibbering species of ground dwelling apes than a miraculous machine that gibbers for them, to point back and forth at each other?
I'm not noticing a "cognitive decline" per se, but I do see I'm a lot "lazier", even stuff that used to be routine when I started coding now feel heavy.
The funny thing is, maybe not noticing one can be the actual sign of it :)
And I'm just afraid this is what cognitive decline feels like from inside the deteriorating mind.
I use an agent to generate a first-pass attempt, and then (deadlines willing), I manually read every line at least once so I understand what the code actually does.
Then I manually fix the inevitable slop that is mixed in with the good stuff, and only once the code is up to my personal standards do I send it.
This probably reduces my “AI performance boost” to 30-50% instead of the huge gains reported by others. But I retain the ability to reason about the codebase and use AI much more precisely when I’m trying to troubleshoot production outages or subtle bugs — something I notice the rest of my team struggles with, since adopting “agentic workflows” everywhere.
I think actively working to retain some cognitive flexibility and “muscle memory” around coding tasks is going to be rather advantageous in the long run.
> a shift of skills away from things that mattered more in the past toward other things that are not measured/perceived by the older generation.
Do you have any ideas what these things might be? As someone in his twenties, I’m sometimes saddened by observing that some of the skills I acquired over a long time (e.g., writing, coding) may become obsolete or won’t be respected anymore just now that I‘m finally getting good at them.
But now we delegate thinking itself, so I wonder what is left.
A proper nights sleep is massive! I'd put 99% down to this..
I used to think like this until social media proved there are some tech innovations we just can’t adjust to. 10 years ago you would’ve never caught me supporting any sort of age based social media ban. Now? I don’t think it goes far enough. Fake news (actual fake news) and misinformation has only gotten worse with it as well. It’s so destructive.
The same goes for speed and quantity of input, as to what the human is designed for (not literally designed). Be it social media with it's infinite scrolling, cars racing by as opposed to looking out the window a few times per hour because you see someone/something, constant sound input if you live anywhere remotely busy or work in a busy office.
The point I'm trying to make is that the world used to be comprehensible for the human. Some understood a little complexer things, some only the simpler things. Now there is an overload of everything. So, most humans are in survival mode wether they know it or not. Hence the many seekin mindfullness etc
No matter, it's an observation, not a judgement or opinion on it. The world will just keep rushing forward. Some have a slight hand in the direction it goes for better (never) or for worse, but spiral it will.
>> The world will just keep rushing forward. Some have a slight hand in the direction it goes for better (never) or for worse, but spiral it will.
The systems are too large and self-propulsing for anyone to really control. Consider the rainforest. How many millions of variables interact, nobody is in charge, everything influences everything in a billion different ways. You might say, well we can cut it down, so kind we can control it. Allright, let's continue to spiral. You might build a city there after a few years. Still in charge right. But it get's too hot because there's no vegitation, so you have to change again. And then we find that people keep getting strangely sick, and scientists find some special mushroom that survived and apparantly thrives on the mix of cut trees and diesel fumes and their spores in the air are poisonous. I made that up, but you get the idea hopefully.
There’s no way to learn than to force the brain into adaptation which it is resistant to do through challenge and stress, just like your muscles. Similarly you can’t play e sports and get into physical condition any more than you can use LLMs to do your homework and learn.
It’s going to be a hard adjustment for a lot of people to recognize that letting the machine think for you is as healthy as smoking brain cigarettes.
The smart student uses the LLM as a proctor or provide challenges and feedback on attempts rather than an easy button. They make great tools for learning if they’re used as an adversarial or editorial tool. The future belongs to those who work to use the tools in ways that make themselves more efficacious, not those who use efficacious tools so they don’t have to work.
Yeah, this is how we used wolframalpha for Math as students. Whatever we had to do, we did it ourself as a group of three. Afterwards we checked with Wolframaplha to see if we were correct. If there were any difference between us, we went line by line to find where the error appeared.
It was helpful, because we did it ourself, but because the work was graded, we had the security, that it is not a total failure.
But I like to add artwork to my presentations. My artistic skills have not advanced beyond 2nd grade. So I'll make a line sketch, and give to AI to "fix" it.
The results are nice and I use them.
I have no interest in learning how to do art well myself, so using AI for it is appropriate.
But I still write my code myself.
With writing:
Things like brainstorming a plot line for a book with a custom GPT or Claude project that has all of my prior books in its knowledge? Works great.
Things like asking it to write a paragraph or chapter for me - I can rapidly feel my own writing skill, motivation, vocabulary, and ability to grasp/remember the resulting plotlines deteriorating. I don't use it for that anymore.
With studying:
I've been taking a couple of evening uni courses and the thing I found so great is that I've been forcing myself to think through the problems, and take my own notes in every lecture. I may then still get ChatGPT to help explain and reason through some of the concepts with me. And I have it review and 'grade' my assignments. But I refuse to ask it to start drafting answers.
With programming:
This one is tougher. When I am not very personally invested in a problem or codebase it becomes too easy to offload more parts to Claude, and when the company encourages 'vibing' to speed up velocity and you're reviewing and writing a higher influx of lower quality PRs, investment goes down. I still sometimes catch myself committing solutions I only _mostly_ grasp and the rest is hand-waving. A big part of it is a work culture thing.
For my own projects I make sure to understand and have a back-and-forth with the planning agent for each task, or write the first plan myself to go off of. When it comes to producing the code, I have to admit it is much easier to properly review parts of the codebase I am extra interested and knowledgeable in (backend in my case). The frontend I'm less well versed in and also admittedly less interested in, so I do sometimes fall into the trap of "Ehh it works, just commit it" with the goal of doing a thorough quality pass before actual release.
With all of the above, I can feel my ability to think, plan, reason, focus (and my vocabulary) suffer if I go over the line too much into agent offloading. For me keeping that balance is as much about maintaining my own long-term brain health as it is about producing good output. I imagine younger people growing up with AI today won't even know what that more capable (in my opinion) brain state feels like - to them, the AI-using brain will be the norm.
But this doesn't seem to make sense when someone comes to a topic with an LLM in-hand. They need to know high-level techniques, architecture, best practice, etc. As they pursue the topic they start to get down into the details, although probably never learn to do it fully independently.
I quite like this view because it paints a somewhat optimistic way forward from where we are now.
High-level techniques were never a problem. You could Google tens of articles on this topic. They are useless too, it's like learning how to drive a racing bicycle from reading a book. Sure, you will know a lot about nuances, but you will fail miserably when it comes to a real race.
I had the experience to keep calling out AI to simplify and downgrade the solution to something primitive, which ended up smaller, faster, easier to maintain. Juniors with real world experience would not bother, they’ll take the first working AI result.
I hope this isn’t the case. It is the route I took, but it also doesn’t seem to be a likely route going forward. Strong CS grounding is feasible for sure, but I have a hard time believing that a meaningful number of people will be spending the requisite years coding manually.
I disagree, the definers of taste; art and food critics, movie and book reviewers, don’t need to have learned the craft by doing. Taste is a separate skill.
e.g. The "group" abstraction requires one see a lot of int, polynomial, modular arithmetic etc. before knowing why we want such a thing. It's unskippable.
That awareness of how to structure the English language, it will benefit those who use LLMs.
Then again, maybe someone will just make a LLM that’s built to turn poor English and poor reasoning into excellent English and excellent reasoning. Maybe this is just a technical puzzle that needs solving.
As an example, I have been drawing portraits for quite a few years now, and whenever I go on a hiatus and come back after a few months, I can notice my skill not being anywhere close to where it was before I stopped using it.
Sure, after 2 or 3 portraits they mostly come back because of the previous experience, but skill rust is a real thing, and if you think your coding skills are the same because you used to code 20 years but haven't coded for some time, you are probably just lying to yourself.
I still did well, but I had gaps for which there was no help outside of the internet available.
Trying 5N paths is useful and sometimes yields interesting insights I’ll retain, but it’s not the rich, challenging, deeply engaging kind of process I find I need in order to develop useful knowledge and skills.
So yes it’s an accelerant for people who want stuff from me, but that doesn’t map directly to learning and building skills. I think that mismatching is really important.
I've heard LLMs can be helpful in limited targeted ways. But not as some kind of "game changing" accelerant.
The downvotes are just a sign of the times. It's also something to observe and think about..
Other fields may be different. YMMV
Asking suggesting or arguing to go deeper is impossible. There is a new path of least resistance and it saddens me.
Sorry, but I highly doubt that. Has a very "old man yells at clouds" vibe.
But apparently some of the smartest people in the world have lost the skill? But the commenter haven't, because why, they're 15 years older and thus immune to the same LLM-effects?
Plus, the issue with people having trouble sitting still for 30 minutes precede LLMs with decades.
Most people definitely can't meditate for 30 minutes, so if you can do this, it's very impressive. Regardless, being able to think about poorly-defined problems and build completely new mental models from nothing is genuinely a really hard and uncomfortable task. If you don't use the skill you'll lose it.
Maybe not traditional meditation, but I have no problem taking a 30 minute plus walk with nothing but my thoughts. It’s actually when I do most of my thinking. The other is in the shower/sauna where devices don’t work anyway.
Not saying everyone else is immune, but those a few years older have also had a period without it.
> apparently some of the smartest people in the world have lost the skill?
> But the commenter haven't
> why?
Perhaps because a correlation you assumed was there (more smartness = more ability to sit still alone with one's thoughts), is not actually as strong as you thought? If one does not start with that assumption, there is no inherent conflict in the 3 pieces of evidence you cited.
Or perhaps because you are smarter than you give yourself credit for :)
tomorrow most regular people's thinking skills will definitely be weaker than those of the LLMs of tomorrow. And physical skills in most cases will be weaker than those of the robots. That leads to the question - what would most people do?
(I am not saying LLMs can't be a good tool in evaluating ideas. To me, it sounds like you're firing off ideas all over, letting the LLMs judge what's good and what's not. Insane.)
Around COVID times many top universities experimented with removing test requirements from admissions, under an argument largely related to equity. It's been a failure everywhere, with many, if not most, universities already reversing it. As Yale put it, "Yale’s research from before and after the pandemic has consistently demonstrated that, among all application components, test scores are the single greatest predictor of a student’s future Yale grades. This is true even after controlling for family income and other demographic variables, and it is true for subject-based exams such as AP and IB, in addition to the ACT and SAT." [1]
That link is for an archive because that page has been removed. That's because they briefly experimented with a new 'test flexible' strategy where they allowed students to submit test scores or not, but then scrapped that altogether and went back to simply requiring test scores.
[1] - https://archive.is/8zxfo
It was already discussed on HN.
https://news.ycombinator.com/item?id=48309233
Could you explain?
From the current article
In addition to overreliance on AI, Garcia also pointed out that many students are underprepared mathematically, a concern echoed by campus associate teaching professor Gireeja Ranade.
From the article discussed the other week:
Over three years — from fall 2021 to fall 2023 — the letter said, at least 20% of Berkeley first-semester calculus students who took a diagnostic exam showed deficits. “Basic mathematical fluency is analogous to literacy; without it, success in university-level STEM becomes structurally unattainable for students,” faculty wrote.
It's been steadily getting worse. The current article only looks at F's which conveniently hides if there has been a slope down. Additionally, kids entering HS in 2021/2022 would just now be hitting college.
Works the other way too - if you introduce something positive in grade 1, you'll only see the results a few years later.
"Failure to complete the qualification" is the prediction.
There are many countries, especially in Europe, where entrace/admission tests are not a thing.
That was in the 1980s.
My first math exam as a CS undergraduate, 123 out of 129 students failed. The math department professors refused to dumb down their classes for CS students.
Math was core to the CS curicullum in those days. It would fade away over the next few decades to almost nothing. The main reason being the CS department wanted to popularize its uptake, and remove barriers that kept students from passing. There was also a major dose of interdepartemenral rivalry and academic politiking involved.
"More than 600 University of California faculty members, led by mathematicians at UC Berkeley, are calling on the system to reinstate standardized testing requirements for science, technology, engineering and mathematics applicants, saying that six years of test-free admissions has not reliably assessed readiness and professors are often teaching middle school math to incoming students."
https://archive.ph/18spS
As a Cal alum, I am actually really glad to see they are holding the line on grade inflation. I worked my butt off to achieve the GPA I did, and it would really suck to see my labor devalued if Cal went the direction of e.g. Yale and started handing out 79% A's and A-minuses: https://yaledailynews.com/articles/professors-face-grading-d...
Its all Goodhart's law problem, but we are missing the forest for the trees talking about grades and tests when what we want is people to be educated, and critical thinkers and competent in their area and due to a comprehensive way to evaluate that we end up talking about grade inflation or how Yale vs Berkeley gives letters at the end of a semester
>Intentionally lowering the quality of instruction, as well as deliberately trying to trip students up on exams
I was happy with the quality of the instruction, and I didn't feel I was being "tripped up" on exams.
It's not about "hunger games", it's about challenging students to learn a lot of material and learn it well. Again, if that's not what you want, just don't attend.
The number of places where this environment exists is getting smaller every year: https://xcancel.com/CJHandmer/status/2060144837157118307#m
I'm glad the professors at Cal are working to preserve it there.
Maybe we can use AI to create new exams that grade people on professional capability, and then gate entry into other professional degrees?
Hmm, Where would the teachers come from, and how good would the education actually be?
As a naturally curious person, nothing will stop me from learning about the topics that interest me. But school also taught me a lot of things that didn't interest me, and a lot of those things turned out to be useful anyway. I think if I had access to AI from a younger age, I'd have used it to skip learning the things I didn't care about, which would not have done me any favours.
It kinda was fun, like a very patient professor stand right besides you. It was the one of the best math learning experience I've ever had, and you don't even need to send bribe/gift to Gemini to keep you in it's favor.
On the other hand, if you ask a LLM to completely finish the work without thinking it through by yourself, then it sounded like cheating, to yourself.
I was in my 3rd bachelor's year studying physics (France) and overheard a conversation between two of my teachers. They were discussing how they should modify the 1st year program to now include math, because he had been noticing how more and more students were failing the more math-heavy subjects like body and newtonian mechanics. He said that they should now teach (or re-teach) calculus to 1st year students, which was not taught when I entered college (it was assumed that you learned it in high school and we would only cover linear algebra in 1st year).
I can imagine things are only getting worse with students that can now get under the illusion that they know math because they have a tool that can do it for them. Which raises the question: should programs adapt to this, like we adapted to having calculators?
It’s like testing your drawing ability in a photography class. The difference is that now nearly have subject and testing method we have has become obsolete. Drawings courses still exist as will traditional courses, but the main stream has changed and exams and schools need to adapt.
You do need to be good at math to do e.g. physics (or math itself!), nomatter the tools at your disposal.
There are several reasons for this:
1. Cheating in CS is easier to detect. MOSS [2] (authored by CS professor Alex Aiken) is a very effective tool at detecting plagiarism in coding assignments. Personally I witnessed more honor-code violations in math problem sets, but there was no feasible way for professors to detect this.
2. Problems in programming assignments are (usually) very tangibly wrong. I can bullshit my way through an essay with shoddy research, I can hand-wave a proof that is definitely wrong but will probably garner at least some points. But when your program is crashing or not compiling, and the due date is approaching, it produces a very immediate and undeniable sense of failure and pressure to cheat. The thing is, many students would get a decent chunk of credit even for failing code, but this is not immediately obvious.
3. The ability to cheat is more available. Math problem sets tend to change quarter by quarter. It's basically impossible to cheat on a prose essay short of straight up paying someone to write it for you, or fabricating sources. But for CS classes, especially at prominent universities, there are plenty of solutions online. Much of it is people who aren't event at Stanford implementing the assignments for fun or self-learning, and sharing it with their peers. Which, to be clear, isn't unethical or bad - it's the responsibility of Stanford students to refrain from looking at those solutions. But nonetheless, it's a contributing factor.
1. https://stanforddaily.com/2015/03/29/increase-in-cs-106-hono...
2. https://theory.stanford.edu/~aiken/moss/
He apparently also makes (I would assume a satisfying amount of) money selling the same technology to law firms for copyright/patent analysis: https://www.similix.com
(I love these ultra minimal HTML sites, ex. https://www.hwaci.com (SQLite commercial licensing) for another example. It just has this subtle smugness, like you either don't need any new clients or virtually all of the market is your client.)
Did they use AI to detect AI using cheaters?
AI detectors are pretty mid in practice - they tend to have a lot of false positives for "B" students who are okay, but can still be struggled to be more coherent than AIs are. There are some specific triggers that AIs are way more likely to do than students, but a lot of AI detectors will trigger on this "almost there, but you're still struggling" level of essay writing that might get a B, B-.
I could expect the same might be true for CS students even though I haven't seen how AI detectors work for CS/math homework.
It’s not that students didn’t cheat before, LLMs have just lowered the bar so far many can’t complete a live test in a class that requires effort.
It's not AI, its a deterministic program that analyzes compiled code for similarity.
When you're up against a deadline - and unless you're very good at time management you're frequently up against a deadline - it's going to be an irresistible lever to pull.
In times past, cheating would mean copying an answer off the Internet or off a friend, both of which are easy to detect. More sophisticated cheaters might spend an hour rewriting the solution to make it less obvious they cheated, but at some point the cost of cheating (time + risk of getting caught) starts exceeding the cost of just doing the assignment. AI changes this - you get a customized answer that doesn't show up in a database with no extra work.
The thing is, students fail to realize just what using AI robs them of. Struggling with the assignment is the entire point. You don't learn if the assignments are too easy; you need to have some challenge to push your brain to understand the material more deeply and to build those pathways to apply the knowledge in novel ways. You become more efficient and effective over time as that knowledge settles in and you get more proficient - one of the reasons why time-bounded exams still make sense (being fast is also a proxy measure for understanding).
Not sure what the solution is - there's no possibility of stopping students using AI to complete their homework/assignments etc. But let me flip the question - do they need to be stopped? Why not let them fail at the exam? As long as the exam acts as a filter, their usage of AI to "cheat" their learning is inconsequential to anyone but themselves.
Now the barrier to an answer is zero. They are basically watching a YouTube video on how to X, seeing step by step instructions feeling like they are doing it, and the moment they swing a real hammer they are whacking themselves in the crotch. It might get better after a few years, but this stuff is just now hitting mainstream for the masses. ChatGPT has only been in mainstream use for about 3 years.
AI apps are very powerful for teaching. You just need to tell them to do that, and not to directly solve your problem.
A bunch of science fiction stories had "first connection to cyberspace" as a coming of age event, maybe those authors were on to something.
Plagiarism isn't new, and those things enabled it too.
It's funny that GP mentioned science fiction as a negative because what immediately springs to mind, for me, is Neal Stephenson's The Diamond Age. We literally have the tools to build his "Young Lady's Illustrated Primer" today. We just have to give today's AI a lesson plan to follow and ensure that it never gives the student the answers, and only keeps explaining the concepts in different ways until they click. Wrap that in an iPad app and you've essentially got the exact self-paced learning tool that Stephenson envisioned changing the world.
Overall it just seems like a huge waste of money to piss away the huge tuition cost your parents probably paid.
The smart ones either use it not at all, or use it to positive effect, like you're saying.
Main problem is that the technology was very disruptive for education and nobody has figured out yet how to utilize it at scale for schools and universities.
The solution? I'm not sure but possibly use AI as more of a collaborate partner to discuss with rather than letting it give you the answers
To do this, you have to be a professor who has a strong idea of what subject mastery looks like. Not available to most.
But ... It is exactly the right idea IMO
Anyone with a pulse can declare a CS concentration at Harvard and muddle by (you actually need to try in order to get a C/C-). Of course, GPAs are calculated differently at Harvard compared to other universities, as a B- is treated at a 2.67 but most other programs treat that as a C+.
Ironically, the techniques of the latter yield the results of the first, but everybody gets to keep a pure heart.
People can use AI to outsource their learning, but if they use ai to outsource their understanding they just set themselves up to fail even more.
From what I’ve seen, how students are using ai (not that they are using ai) is making them less prepared for the real world, which unfortunately is changing faster than ever at the same time to create double impact.
* Tom Lehrer: New Math (1965) https://www.youtube.com/watch?v=W6OaYPVueW4
It's a rational response to entrenched elites that prevent realization of the very social contracts they push on the youth (hard work will equal success, home ownership is a fundamental, etc).
Combined with the looming specter of climate doom, and watching the adults do nothing about it, treating preparation for a conventional career as a scam to be counter-scammed makes a certain sense.
But I essentially completely stopped using them for software engineering (why isn't really relevant, but it's not because od this skill atrophy). So as the skills of everyone else is diminishing, mine is proportionally raising.
It has never been easier to get better than others. You don't need to put in more effort, just the same effort as you always have, and others will do the job of losing their skills for your own benefit.
And quite honestly. It shows in the CS grad population too. A lot of us are condescending toward anything that doesn't make sense to us. But, I digress.
The best engineers I've worked with are all non traditional backgrounds, non degree or degree holders from non elite schools. They think differently, they tinker, they are incredibly nice and patient, and do it for the love of connecting humans to technology.
Look up the names mentioned in the article. Garcia, Ranade, Nelson. All of them are involved with highly theoretical mathematics and scientific computing. Just because you're good at 1 thing does not mean you are qualified to teach. And none of these professors are trained or taught or graded or performance managed on how they teach. For most of them, its just required that they spend 10% of their time in the classroom lecturing.
Let's be honest about another thing. 99% of EECS graduates, even from elite schools, are wrangling objects and their relationships to a graph. Simply put, we're all just a bunch of glorified JSON massage therapists. It just so happens that we get paid well for it, and we hold that over people. The same happens in the classroom.
I think in order to facilitate a healthy, educational environment for young adults, we as adults must encourage, motivate and make that environment fun and practical. We force feed binary trees and the compiler AST's, but we need to make it fun. It's like the commonly accepted saying: Schools kill creativity :(.
Worse, a decent chunk of research profs will treat teaching as a burden that just has to be done - a distraction from their exciting world-changing research. So, you get attitudes like the ones you mentioned.
I'm actually not sure why the system is set up to assume that profs who are good at research are automatically suited to teach classes, but that is how it's setup.
I got all that stuff. I've wired up a 4-bit adder on a solderless breadboard for an architecture class. I used to have a well-thumbed copy of Knuth handy. I've designed and built a switching power supply. But I'm not up to date on using Claude Code, and should be.
Of course, if a student just breezes through it then I would agree. That would make no sense.
Start the kids off with high level stuff, but make them do some embedded systems on their way through. At least for an engineering degree. Also, do a bit of lower level communications somewhere in there; expose them to tcpdump/ wireshark, but they need not develop expertise.
I don't think instruction would've changed drastically in the last year though.
+10000. The goddamn slides. If I were a student now going to engineering school, I'd basically take the slides and throw them into NotebookLM and get way better lectures. Then I'd ask claude or GPT all my hard questions. Hell, I'd get the PDF version of my textbooks and do the same.
The number of lectures actually worthy of your time was so low.
Artificial Intelligence and Grade Inflation
https://cshe.berkeley.edu/publications/artificial-intelligen...
Alternatively, more students are taking CS10 and CS61A irrespective of aptitude.
Anyone can code, but not everyone can become an employable SWE.
Anyone who has first or second hand experience with Cal or any other university knows how impacted CS majors have become, and how everyone is attempting to become a CS major because it's the easiest path to multiple high paying white collar careers.
And in all honesty, it's not like CS@Cal never had weedout classes (I remember CS70, CS61B, and Math54 had reputations of being the L&S weedout classes).
At UC Berkeley L&S, students are undeclared by default, and everyone is incentivized to take the intro CS classes (CS10, CS61A) irrespective of aptitude because worst case they can declare a CS minor or use the classes for other adjacent degrees (eg. Applied Math, Data Science).
Additionally, while Cal doesn't require standardized tests, most students who applied and attended already took the SAT, ACT, and APs becuase they cross-applied to other universities as well. This is reflected in UC Berkeley's HS Weighted GPA being in the 4.31-4.65 range [0], which means most students will have taken at least 6 AP classes.
Hell, I attended an Ivy and even then Cal was a target program for me, as well as my peers. If I didn't get into my Ivy I would have ended up at Cal and ended up in the same position.
[0] - https://admissions.berkeley.edu/apply-to-berkeley/student-pr...
Barely over a decade ago, CS tended to be a large but not too large major by enrollment in most universities yet nowadays it is the most in-demand major in most universities. You can see this at Stanford [0], but most other programs as well.
[0] - https://stanforddaily.com/2020/04/25/stanford-in-the-2010s-t...
The goal of education is to impart knowledge in the student, preferably correct knowledge. The goal of an LLM is to produce an output that is convincingly human. It's not even that they're opposed, as much as they're ships for whom Polaris is in a completely different direction.
"Hallucinations" as they're called, or more plainly stated when the machine makes some shit up, are perfectly understandable in this context, as are the struggles of every single AI firm to get rid of them. Namely: the machine is functioning exactly as it is designed to, so how can you possibly fix it? It's working. The goal of an LLM is to produce text that passes for human, and apart from the obvious LLM tells, it largely does. Like say what you will about their lack of intelligence, the writing is solid. It's grammatically correct, spelling is dead on, what have you.
It reminds me of the famous phrase from Chomsky: Colorless green ideas sleep furiously. A sentence which is perfectly grammatically valid but is also completely devoid of meaning. An LLM would write that sentence, and it would be working correctly.
All of that to say: for all the things they CAN do and CAN be used for, I think we have to draw a hard line at education. I just don't think AI has a place in it. Of course that presumes that the goal of education is to, well, educate people, and especially here in the States but also abroad, we have been putting other interests, especially capital, far ahead of that for decades. I expect no different here.
And before someone comes in to go "WELL HOW DO YOU THINK YOU'RE GONNA STOP IT LUDDITE IT'S THE FUTUUUUUURE" yes, I'm sure as long as these exist and are available to people tech literate enough to access and use them, whatever that means into the far flung future, they will be a factor. Just like cheating, just like plagiarism, just like everything else that will get you kicked out of school. And the answer is the same: it will be stopped by institutions, imperfectly, and it will also happen anyway and with the same consequence: those responsible will mostly be harming themselves for short-term gains.
"Enlightenment is man's emergence from his self-imposed nonage. Nonage is the inability to use one's own understanding without another's guidance."
https://www.columbia.edu/acis/ets/CCREAD/etscc/kant.html
I would grant: I was not the most studious kid, I could definitely stand to learn how to read code a lot more effectively than I do; but I have found being able to ask a computer, "what portions of the Vulkan Programming Guide are less relevant with Vulkan's design changes since the release" pointing me to the dynamic rendering extensions and placing it into context, with inline code and links out to useful blog posts for additional reading, that sort of thing is very helpful.
Working on a prototype before I was trying to learn Vulkan, I was using it to explore SDL_GPU's API which definitely had some gaps in its documentation. Granted again, I could have referenced the sample code - I am sure you'll prefer I'd have done that - but it helped to get information about what each piece of the API was doing, and gave reasonable results that made sense and did inform me enough to understand what I was doing, turning much of that into an interactive learning of basic GPU programming for graphics. Where the AI hallucinated, it was often on things like method names, which I was able to read through and find the methods it was intending to name. (This only occurred once or twice when I was learning).
Unrelated, but adding the C macro syntax and nesting macros, which I could have an LLM explain inline and link the GNU manual. Never got that taught to me in a C course. Man, computers are complicated!
These have not replaced textbooks; I have been using them alongside textbooks and handwriting code for practice, and they work as a very good complement. I also sometimes use them to unblock me - I don't know CMake very well and lean on AI to do CMake, so I can focus on learning C++ and graphics, which is my primary objective right now.
I would add too, I have for fun given it prompts about various topics I learned in university, and I often will get answers that are bang-on what I learned in university undergraduate courses - the topics I tried were welfare state taxonomies, distributed systems, disk storage performance, filesystem layouts and internals.
Boy, this would've been cool for me as a kid. There's just so much information right there, and pointing you to topics and textbooks a couple questions away, I wish I had these tools. I was a curious kid in a terrible MAGA-esque family that was deeply uncurious about the world, had no knowledge of any advanced subject and basically mocked me for trying to learn more about stuff. And you go to the school library and it's all kids shit, not even an option to try and reach out for more. Now smart kids might be able to go just learn shit very freely and be pointed to textbooks, and go pirate them off some Russian site, and start learning and go tutor themselves, as I'm doing today as an adult.
At least knowing myself and knowing if there's another kid like me, I think they would deeply enjoy having a natural language encyclopedia, if we can get it as close to that as possible. I think even with some error inherent, if the tools can be often and directionally correct, that would be a plus. I went to university, and the professors there hallucinated some things so embarrassing it should bar them from teaching, for the standards people hold LLMs to! i.e., sanitizing conspiracy theories that Android records all language through the microphone therefore iOS is better, Apple Silicon is more battery efficient because it is RISC and not CISC. Got a terrible history of computer graphics technology you'd know was slanted if you watch the 8 Bit Guy on YouTube. Rubbish.
The thing that worries me, and what this article really talks about, are the kids that just don't give a shit. They are not new - when I went to high school, before AI, stupid kids would copy code off the internet. I think AI probably makes it worse because it makes it harder to call out and enforce against it, and agreed, that should be stopped. But to me, that is mainly a cultural problem. Too many Americans are completely uncurious and just spout garbage; there are a lot of kids who grow up in that cesspool and are going to grow up uncurious, and then AI acts as a shortcut rather than a vehicle of curiosity.
And granted, maybe AI is less useful when you are in a structured environment - but the structured environment has its downsides. Even in that environment many of the TAs were clueless and unhelpful, or just too damn busy or already too knowledgeable to meet students where they were at. Again, talk about hallucinations with TAs! Many times in my experience. And that's all to say nothing about getting people to not just do homework but actually go get curious about things and try stuff that isn't required of them.
I think there will be some culture that remains curious, and has these tools, will come to grips with where they can help, where they go wrong, how to balance it with other learning methods; and I think they are going to have kids that absorb a lot more knowledge and get to play with topics and learn things, faster, to each kids' interest, perhaps even individualized tutoring at better scale - I hope that is possible.
I hope the United States as well, but maybe not, because holy cow our culture and attitudes are plainly terrible these days. Your comment is pretty representative of how most people react if I suggest this or talk about my own experiences I'm describing here. But I hope at least I'm arguing something comprehensive here. There is too little conversation beyond hyperbolic nonsense on the internet; I consider "FUTURE LUDDITE" etc. to be in that realm.
It is just hard to reconcile that denigration of AI with the typical experience I have using these tools in the real world. It is not omnipotent or God, but it can effectively assist in work. There is a certain cognitive dissonance I feel when I walk away from using the tool to help accomplish particular tasks, then hear over and over people say the technology is fundamentally useless and fundamentally does not work. I guess I am just not enough of an academic to understand how something can accomplish work yet fundamentally isn't, somehow.
I imagine there is some apathy and laziness here but idk how unjustified it is
"Noooooo you need to manually code on paper in assembly"
Alright, well maybe the CS grads need to, but why expect that of everyone else