Failing grades soar with AI usage, dwindling math skills in Berkeley CS classes

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.

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.

[1] - https://archive.is/8zxfo

CS Professor here: just yesterday I did the discussion of a course projects' (Parallel Computing), and one of the three groups that I did yesterday have clearly gone the ChatGPT way. They couldn't even understand the choices the LLM made regarding the architecture, etc. The way to "catch" these students is similar to what we did in the past when students copied from other students which is "to give them rope to hang" - ask for clarifications until they follow unintended paths that lead nowhere.

To fellow professors, when you're suspicious my suggestion is to appeal to their honesty (like "let's be honest, how much of this code is yours, and how much is ChatGPT's?") and offer some empathy and understanding (like understanding they may had multiple deadlines in the same week, etc.). Nevertheless, don't miss the chance to give them the lesson on how is the correct way of doing things. The way to catch these students is to find the same signs of yesteryear copying from other students (which in essence is what copying from an LLM is, although the number has increased because they found us professors unprepared for the volume).

The other two groups also used LLM but in a high-level and architectural way. They were clearly responsible for the code (even if they didn't wrote it 100% manually) and could explain their reasoning and strategies used to solve the problems.

Me and my colleagues still have a lot of projects to review, and I asked them to keep the score of the number of projects like these, but so far, the score is 1 in 3 (33%).

It's a strange thing that as humans, we sleepwalk into every crisis, never agreeing on anything, and then when we're there, we also never agree on the causes. When we ge too the point where we can no longer "engineer" or "science" anything we will spend the next decade arguing that the issue was not really AI, or that if it was, it was inevitable and no one (or everyone) was to blame. Rinse, repeat. Yet we're here, today, looking at the bleak future, and taking yet another step forward.

Do we assume society just self regulates. I think it does, but the cost of letting it self regulate is really really high, with lots of suffering. Is it that we find this acceptable when there is a chance we won't be the first to feel the pain?

In unrelated news

"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

>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...

My daughter was struggling with her Math class back in January. I used Claude to build a tool that allowed me to generate very focused worksheets. The worksheets had problems designed to drill the concepts she was struggling with.

It worked, and it would have been MUCH harder to do this the traditional way.

The tool generates PDFs including an answer key and solution sets that solved the problems using a variety of techniques so I could check her work more easily and we could iterate quickly.

That's powerful. It comes back to how are you using the tool. Are you using it to make things better or to take shortcuts?

The grade data come from https://berkeleytime.com/grades

I was worried they may have cherrypicked courses that support their chosen narrative.

So I plotted the % of F grades (red line) for all CS courses still offered, and sorted the chart in descending order of the # grades given out (light blue vertical bars) in the most recent semester when the course was offered.

My worry was borne out. See the first few charts. No big increase in F % in the past few semesters.

https://x.com/rahimnathwani/status/2062431813143019525?s=61

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.

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.

They worry me. A lot.

My son is 15 and I use Google Family Link to control what he does on his phone: it's pretty open for the most part (I receive notifications of installs) but Gemini is a hard-ban.

We've spoken at length of the dangers.

He says his pals use LLMs frequently and I suspect that's the reason for their test scores: some of them are in the 20% - 40% range for tests whereas my son is 80%+ because he studies past-papers and answers questions in his revision.

I worry for the future coz you can be sure that the AI providers don't care if a schoolchild is using their LLM to answer the homework questions.

“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.

But ... It is exactly the right idea IMO

If you watch old videos of tradesmen using basic hand tools like hammers, you'll find examples of skill/dexterity with the tool that I think don't exist today at all except maybe in communities like the Amish.

I think it's true that we collectively lose something akin to beauty every time technology advances. But usually some new set of skills that have beauty emerge.

If LLMs end up being the pneumatic nail gun for the human mind, I personally think that's a fine thing for us to accept.

If they end up being more like some dark factory that autonomously does everything - then I think ultimately the thing that makes us human (our minds) will slowly decay and be lost, and that seems very sad. That's a version of the future we should try to prevent, I think.

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.

The obvious cognitive deterimental effects of using map apps, when we all realized we lose directional sense and our previous ability to navigate without the smartdevices, was society's canary in a coal mine and a headsup of what was coming.

WAL-E and Idiocracy. The future.

> “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.”

Grade curves are how you test your curriculum for good challenge - are you challenging people such that an A isn't a too-low threshold. When you force people into a curve, you haven't defined a threshold of mastery, you've defined a sorting function: A means "better than this year's peers". It is absolutely bananas to me that a tech/math oriented school would be doing any sort of curving.

Writing better exams, even if they're more expensive to grade, and removing homework from grading as far as possible addresses this problem well wherever it's applicable. Senior-level math courses at many universities are already like this: homework is ungraded, or counts for little, and it's possible for students to "cheat" on the homework by copying another student instead of struggling through the exercises. But the students who do that don't learn much, if at all, and predictably fail the exams. Professors warn students at the beginning of the class and tell them how this will work, something like:

> You can always ask me for feedback on your homework and I will mark up every part of it, but you won't receive a grade for homework. However, if you don't do the homework and take your time with it, you will fail the class. My office hours are in the syllabus and you're strongly encouraged to use them. There will be an early exam to give you a chance to know whether you are likely to fail this class before you lose your chance to drop it.

Correctness is harder to adjudicate in some humanities disciplines but the format of these exams is actually not super different from essay tests (when a math professor grades a proof, they're inspecting specialized prose for validity, coherence, persuasion in a way that also reveals knowledge).

When you don't rely on homework for determining whether or not a student passes the class, you make cheating on the homework into the student's problem instead of the professor's or the university's. Students have the right incentives to solve problems for which they are the ones responsible, and they figure it out after one failed (or ideally, dropped) class at worst.

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.

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.

I feel like a lot of the lament expressed in the comments is grounded in loss aversion. We really don't like losing things—regardless of the objective (or subjective) value of that loss. We know this intellectually, but we still feel it emotionally.

I'm feeling effects of using LLMs day in and day out, but I am not yet convinced that it's overwhelmingly negative, the way much of HN seems to lean.

I derive a lot of joy from shipping outstanding code for my clients and fixing problems they're experiencing. My joy has only increased as I can now ship better code, faster, with fewer bugs. No, I don't intimately understand the code the way I used to, but I understand it enough to accomplish the end goal.

The premise of this article takes a presumed position that the grades we were posting before really mattered a whole lot. I'm not convinced that's true.

It's incredibly difficult at this point to "skate where the puck is going" as Gretzky is said to have done. No one knows what knowledge work will look like in five years. People used to memorize log and trig tables, and no one would say that's part of being a competent mathematician at this point.

That said, assessments of poor critical thinking skills jump out at me more than the rest. That sort of thing seems likely to matter until machines can replace us completely.

The traditional sit down in a lecture hall, listen to professor speak, then come in for a pen and paper test, is outdated for this era of technology. What's the point of physically bringing +300 students into a room for a passive listening experience?

The future of education is rapid back and forth between the student and the teacher. You could image a future where a student sits down all day with an AI system that trains the student, and it won't let them pass until it's confident you know the material. This will happen at different rates for different people, but this is expected.

An ideal education would be getting those skills discs uploaded to you in the Matrix, but unfortunately it will take a bit longer to master them as we don't have full brain-system interfaces (yet).

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.

> 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.

It’s not just students; this affliction is cropping up among established academics. My wife is editor-in-chief of a journal and in some months has rejected 100% of the letters to the editor including 6 that came in from a single author because all scored 1.0 certainty of complete LLM fabrication. The author in question is no student. It’s a little more difficult to fabricate an entire original paper this way, I suppose.

It will have taken us less than 1000 years to go from scarcity of the printed word to the over-abundance, and finally to the uselessness of it.

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?

It’s not that they can’t think deeply, these are smart people.

It’s that there is no reward for doing so and in fact there is punishment.

The punishment is that for all the thinking you do, someone else will arrive at the same result as you in less time, or maybe even a better result. You don’t get rewarded for the effort of thinking, only for the end result.

Naturally, even if you are an intelligent individual, you can still be conditioned in this way to take the easy way out, unless you purposely like to suffer. But suffering is only worth it if you know in the end you come out ahead.

But now, you do not come out ahead. People will be using AI in the workforce for the rest of your life anyway, might as well just join the trend.

It’s like if everyone started taking a magical steroid and growth hormone to build muscle and look great instead of actually working out in a gym and possibly getting worse results anyway.

Hmm, meanwhile somewhere else on campus this study:

Artificial Intelligence and Grade Inflation

https://cshe.berkeley.edu/publications/artificial-intelligen...

Well, at least the faculty are actually giving out the Fs and not just lowering the bar, so kudos to them for that.

While this is worse in degree, it’s not new.

I TA’d in the early 2000s and the first day students were warned that we used automatic analysis to find programming assignments that were similar to previous submissions. And renaming things, moving them around etc would not help.

We caught and failed cheaters every term.

Well here we have a product market fit issue. See the market was lead to believe that if you got a degree you got a job.

So learning was never the actual goal.

Originally, at least in premis, it was to learn and advance the arts and sciences.

So what we need now is a college for llm's to advance the arts and sciences.

Even as a software engineer myself, I'm feeling a bit of cognitive decline having AI doing some/most of the thinking for me

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

Perhaps the future will belong to those who learn to use llms to enhance their capabilities. Neil Stephenson Diamond Age was an interesting take on this very same topic [1].

So the Claude web app has this “learn” option that turns the session into a Socratic dialog of sorts. One could easily imagine enforcing this on an age based or parental controls set up. Maybe it can be prompted around but at the very least the concept could be a path forward.

As others have said there is a way to use llms to increase learning, but autodidacts will always autodidact.

[1] https://en.wikipedia.org/wiki/The_Diamond_Age

At this point I would support a ban on generative AI by anyone under 18, or even perhaps 21 years of age.

A bunch of science fiction stories had "first connection to cyberspace" as a coming of age event, maybe those authors were on to something.

The most worrying part isn't the cheating it's that students seem to be skipping the struggle that learning requires, as one professor put it "Confusion is the sweat of learning"

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.

1. The article itself seems like an LLM summary of a conversation.

2. No US educational institution should ever grade on a curve. Your job is not to compare students but to educate them. Grade curves hide the performance of the educators and process of education in actually improving the skills of students.

3. Both AI and the cognitive and emotional overload from social media taking away brain space may be to blame. Idea: let students report screen time statistics at the beginning of each semester and weekly or at the end. See if and how it correlates with academics.

> nearly 30 students in CS 10 were caught cheating on take-home exams in spring 2026.

You'd have to be really lazy and disrespectful, to get caught cheating on a take-home exam...

I agree that AI is likely a driving force here, but it is also likely not the only driving force. COVID likely played a devastating role, along with curriculum changes in high school, reactionary cultural shifts towards anti-intellectualism, and broader declines in literacy that have been in progress for a while now. It would be interesting to see data for the past 5-10 years or so.

One thing I’ve used in interviews is to write some code that looks like it was written by an overly enthusiastic engineer who just discovered some new concept (e.g. “trees are the ultimate data structures”) then have the candidate review the code. I wonder if this could work for education: orient the entire class around who can give the AI the best corrections.