Even Professors Are Cutting Corners with Claude…
Jay reporting from Aofeisi
QbitAI | WeChat Official Account QbitAI
AI is ruining my computer science degree.
The one who wrote this complaint was a sophomore girl who had dreamed of studying computer science since childhood, Katie.
With no prestigious university to her name, she had only managed a B in high school math; she had received no formal professional training before, and the only “project experience” she could point to was a few rather crude web mini-games.
I had originally thought that once I entered the ivory tower of academia, I could finally study computer science properly!
Who would have thought—
The course syllabi were generated in bulk by Claude, the final assignments were automatically graded by an LLM, and even the caring emails sent by my advisor were probably run by backend automation scripts...
Faced with this absurd scene, she was completely crushed (just kidding):
Because of AI, my university seems to be undergoing a life-or-death crisis.
Less than 24 hours after the blog post was published, it shot straight to No. 1 on Hacker News’ front-page trending list..
The Completely Broken Computer Science Department
Throughout her time at school, ever since this wave of generative AI erupted, the school’s academic policies never stopped flip-flopping.
At first, of course, they embraced it wholeheartedly:
Grading assignments and writing feedback, generating course syllabi, sending study-reminder emails to everyone... large models were brought in for everything.
The result?
After less than three months of tinkering, they found it was all hallucinations and garbage, so they withdrew it at lightning speed..
The entire academic administration system remained in an utterly schizophrenic state—
You absolutely must not use AI!
Oh no, wrong—everyone must use AI!
No, no... wait, it seems we still can’t use AI...
And then something truly surreal happens.
For the previous module’s assignment in the same course, the professor mandated that you use AI, saying it was an important criterion in the course grade.
Hearing this, you grit your teeth and pay out of pocket for a Claude subscription.
By the next module, you confidently uploaded the code that “Stand Up and Pedal” had coaxed out of Opus 5.5—and then—
No way?
What do you mean, academic misconduct—using an LLM gets points deducted???
But honestly, if it were merely the rules changing from day to day, that would still be manageable.
What hit the author hardest was that they had worked so hard to get into university and were paying such high tuition, all in the hope of being able to study systematically under expert professors.
The result was that you, a university professor, are fooling me with Claude too???
Yes. The instructor at the podium sternly warns students not to cut corners, then turns around, throws the requirements straight into Claude, and generates the teaching materials with one click.
In the past, assignments, at least, were carefully devised by professors in line with the teaching objectives and reviewed by the teaching research group. The workload was moderate and built up gradually.
Now? The girl suspected that not one of those professors had ever even opened a Markdown file.
The whole thing was packed with LLM-generated nonsense. A 3,000-word assignment brief, with 15 deliverables to submit.
Do the math: the workload is now three times what it was before...
The upshot was that ordinary students with weak mathematical foundations who had switched majors and entered through the cross-major entrance exam were forced to speedrun through the knowledge points without actually learning them.
I came here to learn something real and properly, damn it—not to go through the motions on 15 utterly impossible tasks in a 3,000-word assignment!
Many students spent more than ten days painstakingly working on it, but for some reason it simply wouldn't run.
Then, one week before the DDL, the instructor suddenly sent an errata email:
Two places in the outline were wrong and have now been corrected. Please resubmit your work.
Give me back all those Tokens I burned for nothing😭
So, why on earth are they doing this?
Katie says that it can't all be blamed on the professors. The real culprit is still the management.
Do you know what the investors love to hear? Learning at scale—and better yet, automated!
If LLMs can automatically generate exam questions and grade assignments in seconds, why should the school spend a fortune hiring teaching assistants or expanding the lecturer headcount?
Technology became an excuse for the school administration to cut teaching spending without limit, squeezing lecturers endlessly and piling on so much work that it could not possibly be completed through human attention alone.
Katie also mentioned that she hardly ever sees her supervisor around.
The only contact was that, whenever she went several days without logging into the online learning platform, she would receive an email from her supervisor—
She nervously opened it, only to find that it had not been written by her supervisor at all; it was an automatically triggered alerto(╥﹏╥)o.
The hottest post on Hacker News
After the post went live, it quickly exploded across the internet and shot to No. 1 on Hacker News's front page.
The comment section was also flooded with lots of CS guys.
But to my surprise, even among fellow CS students, not everyone was sympathetic.
The entire computing and technology community was polarized into two camps.
One camp was made up of those who had already established themselves in the industry—the seasoned veterans.
From their perspective, all the chaos unfolding now is merely a transient pain in the evolution of the technological paradigm.
I have been in this profession for more than twenty years, and there is always something new to learn every day. Over the past few years, as artificial intelligence has advanced, I have learned faster and gained a deeper understanding of the systems I work on.
In the end, AI is nothing more than a handy tool. If you use large models skillfully to guide someone through Socratic questioning, today's learning resources are countless times more powerful than they were back then.
Artificial intelligence will not replace education or traditional ways of learning.
On the contrary, artificial intelligence will enhance your ability to understand and remember. As a technically minded, self-taught professional, what I have always longed for is to keep learning.
Therefore, many experienced veterans actually look upon the new generation of newcomerswith a certain degree of envy:
Becoming a senior engineer does not really depend on how long you have been doing software development. I have seen some exceptionally talented 21-year-old developers whose programming abilities even surpass those of people who have been working for more than 20 years.
I think the reason they perform so well is that they have a solid grounding in the fundamentals of computer science.
But in the end, this is a story for only a select few.
At the other end of the spectrum, the vast majority of newly graduated young people,only feel suffocated.
A commenter on the thread posted the findings of a quantitative study:
Researchers at Stanford University found that the employment trend for entry-level workers that they projected last year continues to persist and is expanding.
Specifically, employment levels for employees aged 22 to 25 in industries most exposed to artificial intelligence were 19% lower than for their peers in industries that do not frequently face the impact of AI.
In response, some professionals who regularly participate in recruitment also frankly corroborated this reality:
Indeed, instead of spending time developing a new graduate from scratch, it is more efficient to pair a senior engineer with a top-of-the-line Coding Agent.
I genuinely do not know what advice I should give to developers about to graduate.
After spending four years of your youth studying computer science, you still cannot even find a job...
Then what, exactly, is the point of a CS degree?
Rediscovering the warmth of programming
For this question, I think perhaps we can find an answer in Katie's blog post.
At the end of the article, she wrote the following:
I do not care whether computer science and mathematics have reached the end of the road! I still want to learn!
Okay, I do care about my future career development, but that does not mean I want to forget everything I have already learned or stop exploring.
There are still so many interesting directions worth exploring, some related to AI and some not. Yet the flood of headlines keeps saying things like “Software engineering is dead” and “You will be left behind without AI,” making it impossible to breathe.
Can't a girl simply learn statistics for fun?
After all this,Katie's answer was—
I can continue, but I won't. It would give me indigestion.
Maybe I'll go back to OpenProcessing and make some even sillier games.
At least, now I really know how to do some programming.
Then, she posted several crude little web pages she had written on OpenProcessing in her early years:
a simple space shooter set to catchy music;
a Mogu that meows when clicked;
and an interactive demo that uses microphone sounds to scare little fish away.
Honestly, they were very crude.
Any one of them was leagues apart from what Astra and Fable can generate today with a single prompt.
They might even be less popular than a new model’s pelican-on-a-bicycle demo...
But the moment you open one, you can feel something a model will never generate—
vitality.
That was what many programmers were like when they first fell in love with writing code.
Not to ace hundreds of LeetCode problems in the job-application pipeline, not to report in a weekly update how much they had saved on token usage bills, and certainly not to cater to efficiency metrics conjured up from thin air by mid-level managers at big companies.
Before all of it had been disciplined by utilitarian incentives, simply typing a few lines of characters into a black terminal with white text and watching the pixels light up on the screen and operate according to your own wishes was enough to make someone so excited they could stay up all night.
We ourselves created an almost omniscient supermachine.
Yet somehow, without our noticing, we shut humanity’s most primal curiosity and sense of immersion outside the door...
Original blog post: https://lordtoots.bearblog.dev/ai-is-destroying-my-cs-degree/
