I have been doing this professionally for over twenty years, and I still learn something every day. Over the past couple years, with AI augmentation, I learn even more, move faster and go deeper into the systems that I work on.
A CS degree was never going to provide you with all of the tools that you need for a successful career. It provided you with the basics to get your foot in the door.
AI is not replacing you or an education. AI is raising the bar of what you can comprehend and retain. From someone who is technically minded (and self-taught), all I have ever wanted is to keep learning.
I was never a good student (not academically inclined at all) and I have learnt more with AI in past 6 months than the engineering degree taught me.
AI is, in my opinion, absolutely murdering the junior dev market. It's now much more efficient to pay for a senior or staff engineer and set them up with a subscription to claude or codex. I think things are going to remain fine for senior+ engineers, but I genuinely don't know what advice I'd give to graduating devs at this time.
AI + CS allows you to develop with leverage I've never seen before.
Too bad a big percentage wont get to do that professionally anymore due to AI.
>AI is not replacing you or an education. AI is raising the bar of what you can comprehend and retain.
No, it also makes the job market hell and cuts down on positions, on top of an already squeezed pre-AI environment.
And that's if one is still OK with the fact that coding work (something many of us enjoyed in itself) turned into "project manager for AIs".
Yes and no.
I concur that it was never about the tools, although I've seen plenty of people advocating it had a more "practical" approach, using the same tools as companies use, and in the same way.
On the other hand it's not only about the basics.
In there you can go at depth in aspects that you will have to fight to use, like proper engineering practices, and might only be vital on the later stages of your career. And I feel that those learnings, plus the possibility of acquiring a breadth of knowledge that is hardly possible otherwise, is what gives CS graduates a leg up to other practitioners of the art of software development.
YMMV and a leg up can be easily overtaken, yes.
Realistically, "AI is not working" is probably not a realistic scenario. He'll just stop working, just like if the network goes down.
The real risk of AI augmentation is to your value as an employee. Maybe the augmentation makes you feel and perform like Iron Man, but if the suit is doing the heavy lifting, and some cheaper guy could wear it and do pretty well, how are you going to keep that fat paycheck? It's not like any of us have tenure.
I do learn every day thanks to AI, I agree with that part. But I'm increasingly convinced that AI is likely to replace either me or some of my teammates in the near future.
In the last year, half of my team has been laid off (or left) already and won't be replaced, and we don't see any drop in productivity and the tools are getting better by the day. We need experts to drive all this, but not as many people as before.
AI is the tractor in The Grapes of Wrath, and you’re the one driving it.
Good for you! Many people just want to be able to pay their bills.
Your post is entirely too dismissive of the chaos this new technology is introducing.
Far too many people focus on passing an interview and then let legal protections keep their position, all while draining resources that could have been better allocated to someone who is actually passionate. They saw my generation do reasonably well and came for the paycheck. I got into this because there were very few people who could rise to the occasion of problem solving that the industry needed. I did not walk into the room with all of the skills or an idealized salary, I walked into the room with a mentality.
Perhaps the "chaos" being introduced is simply a long overdue market correction.
I'll let the downvotes tally how pervasive this is.
For some, the degree is the destination and that's more than fine as well. It's a marker that says "others agree I learned x to a certain level, and I had fun doing it". This tends to overlap well with those who just liked university as an extra method of learning rather than going primarily to form part 1(a) of their career or something.
TBH I’m also thinking about how to use AI at work for learning. (My work environment is 100% all-in on AI.) Everyone can see and admit that AI coding is making us lazier and that it produces too much code to do old-style line by line reviews. Some people are using agent teams to churn out millions of lines of code where there’s zero hope of reviewing it. We don’t yet know how to keep quality high and complexity low while doing agent coding, we’re just having them write lots of tests and hoping that when things break they’re easy to fix. Agents should get better at broad architecture in the future, but they’re definitely not there yet.
> The amount of work involved has also tripled
This has to be one of the big ironies of using AI. I’ve heard this from other students too. The problem is that old-style assignments where the level of difficulty was hand-crafted is way way too easy for an LLM, therefore teachers need to crank the level of difficulty so the student has to actually spend some time on the assignment. But LLMs are getting better every month and the difficulty is so high that no students can possibly do the assignment by hand and now small errors in assignment design can lead to large errors in effort and time required.
Oh, it's terrible. I'm a computer science instructor at a university (I have 20 years of industry experience prior) and it was already bad enough before trying to predict where the market was going to be 4 years in the future. But a year ago, it was tough to predict where the market was going to be in one month.
So yeah, we can teach all kinds of stuff. But is it the right stuff? With one month of visibility, who can tell?
But now I feel like it's settled down a little bit. And now I feel like I have more like 7 to 8 months of visibility. One thing that's helped is that we see the hiring differential between seniors and juniors. So now we can answer the question, what does the market want? It wants the expertise between junior and senior. And I'm confident it'll want that for at least the next 7-8 months.
Good luck cramming that into 4 years, though! I figure we have to try to bridge the gap as best we can. Eventually, seniors will become rare enough that the expense isn't worth it. It'll become better to hire cheaper most-qualified juniors.
There's still some seriously open-ended questions though. How much does coding matter, for example? If you ask senior devs who don't do any coding how much they rely on their coding expertise to make decisions, you'll get a zillion answers across the spectrum. This becomes even more true for senior devs who use AI for coding. Heck, ask 100 devs if it's important that you can even read code or not and see what comes out!
And then there's the question of how do you teach those systems level thinking skills? Historically, universities haven't taught those skills. You came out of the University with general problem solving skills in technology, and then at your job you were exposed to a small corner of the codebase to fix some bug. Gradually, over the years, you learn more and more of the codebase and become more and more senior. We have to get that into our curriculum. I'm working on that as we speak and will be working on that for the next several years, it looks like; even if my position vanishes, I'll keep attacking it—this is a tough problem that's fun to try to crack.
Anyway, I hope someday my forward visibility gets back up to 18 months. Won't that be amazing. I'd also like to buy some RAM. ;)
Implement more versions with different languages, frameworks, libraries and AI agents. Then you can do differential execution for testing, which is easier than imagining testing scenarios from pure code. You can fuzz or use production data (shadow mode execution) to get notified early of divergences.
Keeping a population of diverse replicas might be the simplest way to derisk AI code. When 2 replicas diverge an agent can decide - was it a bug in one of them, or was the spec unclear? Either way, if you fix a bug or clarify the spec you come ahead.
A separate class of problems is related to intent divergence. For that my solution is to keep a full log of all user messages during the lifetime of the project. Even 10,000 messages collected over weeks and months can be read and summarized in a few minutes. That means you can check for intent divergence with agents, but you always start from the raw source (your human messages) not from intermediate summaries and diverged code.
Combining the 2 ideas would make each replica agent derive the user intent separately from the same log.
The other thing I want to try more at work is, rather than programmers at desks individually talking to agents, do pair or group programming with an agent in the room where the people discuss the design while the agents listens, asks questions, and writes the code live. We’ve had a few meetings where we do this, and it feels much better than individual vibe coding, because as a group we review the design as we go and prevent going down paths I might have gone down by myself. It combines the coding and review into a single session and ensures everyone present is comfortable with the design.
That’s not what it’s about. It’s actually from someone who genuinely wants to learn for the sake of learning who’s talking about how AI is creating havoc in her studies.
The real villain in this story is the teaching staff. They need to get their act together and stop using generative AI for courseloads.
In other words, they need to do their job. The parts about AI mandates changing every 3 months reeks of an overly bureaucratic, admin-heavy structure where the students’ learning is a lower priority being performative about AI for your boss.
AI can teach you, you just have to put the effort in to tell it to teach you. In some contexts, I might want it to fire off a utility bash script and I have no interest in learning bash so I just run it and don't worry too much about the implementation. But if I'm getting it to help me write Java code for work, I review the output and ask it about parts that I don't understand. And it teaches me way better than the old way of reading dry/bad documentation.
It's what you make of it.
AI clearly means you need to change how you teach and evaluate students. We are only a few years in though, and it's still extremely unclear what effective teaching with these tools looks like.
The worst part is that the evaluation loop is measured in multiple years! If you start trying some new education mechanism today you won't get a robust idea of how well it works until your current cohort graduate in 3-6 years time.
Students effectively all become guinea pigs for untested new ways of learning and teaching.
I think we have a solution for the evaluation part: the overwhelming majority of a student's grade is computed from in-person assessments only. Some instructors allow repeats on exams as well.
The teaching part is as hard, but it has always been hard. The students you get one year are completely different the next year, for various reasons. As long as you have competent administrators, all is good--this is what we sign up for.
If tests evaluate understanding is up for debate. But that is what we switched to in the sciences.
Universities are going to have a reckoning moment, as most degrees 1) are not economically viable and 2) don't make sense to be taught in the current mode of instruction.
(Apparently German and Austrian universities already have a strong culture of oral assessment, so they're ahead of the pack.)
First off, congrats on spending time doing side projects in OpenProcessing. Keep working on projects that keep you motivated and keep you learning.
We've been there, lost in our learning paths. Even before AI there were times when we were overwhelmed with our CS journey. You'll overcome, don't you worry about that.
The CS degree was never the end of the learning journey. It was always _what you put into it_. You'll learn what you practice, and uni just happens to be a good place to practice things. Whether that's optimizing AI harnesses, understanding how CPU caches work, crunching pixels on a GPU or the principles of distributed computing... it will all make you a better thinker and good at what you do one day.
Stay curious.
I don't think programming becoming 'easier' has make understanding programs any easier or less important. Even if all of the above goes away, I think computer science has taught me a lot about how to think about problems.
At my uni they had us doing assignments in assembly on day one, before anything was actually taught. The lab handbook said "your first year programing courses will serve you well." lol. Apparently it used to be a 2nd year course and they just moved it to the first semester of the first year unmodified, for no apparent reason.
Had a similar experience at other universities and other courses. This was in the Netherlands so maybe it's different in other countries but my experience is that it functions mostly as a filter for people who already know the material rather than actually teaching you anything.
Over half of people fail the first year (and are banned from repeating it, for 5 years), which is considered normal and acceptable. So it's more of a selection mechanism than anything else.
So I said to a friend who's choosing a university right now, "just don't expect them to actually teach you anything", which is a pretty sad thing to say for an educational institution.
Despite the massive suck all around, I very much like the idea of university, and did enjoy many aspects of my time there, especially the projects. (And I've followed some lectures from other universities on YouTube, where I experienced actual learning, and -- god forbid -- actual joy!)
It is now, and always has been IMO, up to the individual whether or not they want to learn deeply the material being presented, or just do what is necessary to get the grade as a means to an end. In the past, accidently learning something deeply was a little more likely than having the answers at our fingertips via AI, but the possibility is still there depending on how you use the tool.
The best way we know of to learn something with AI is to do the work yourself, then watch AI blow you out of the water on the same task, learn its tricks and do the work again (better, even, since AI isn't that great at it yet)
Which means the teacher can use AI, and you can't.
This is also an entirely different subject to whether you should be learning that specific task in the first place or delegate it to AI.
Citation on this? There is something on HN front page right now that is contradicting.
I don't have a very high quality source but this does cover and expand on it: https://youtu.be/vFOnDsy58NU
Or "Karim lakhani how to learn with AI" and see Gemini's skim of the video.
The other thread is highly sensationalised and not really related. (Edit: sorry I thought it was a different thread, the one that's sublinked by OP in your thread. So disregard this paragraph. I can't contest either position, I'm a mere spectator. That being said, I don't see the contradiction, perhaps you could elaborate on that)
Sure, in the real world outside of school, you'll be able to get help from many external tools. But when the entire point is to learn something, any tools that prohibit you from learning should be avoided.
It's awkward, you're very unlikely to stop students from using AI for certain assignments (i.e: first programming class) for the good of their fundamentals, but there's also no other teaching tool being used to get them equivalent experience to solving assignments without AI.
It was interesting to see things change towards the end. Some was the chaos from the lack of consistency the article highlights just starting to emerge. Mostly though, it was just the rampant surge in cheating and the classes not being prepared for it.
I haven't seen a better example of "those who can't do, teach" than CS professors.
I keep getting excited because I can ask it questions and learn so much from it, but then I realize that a lot of what I'm learning is specific technical knowledge that I don't really need anymore.
That said, the important stuff in building systems has never been the coding. It's the wisdom to organize the system in such a way that it works well and is changeable. AI is still really bad at this. But it's super good at the for loops and if statements, which is actually the easy stuff.
At least his school is trying to find something that works, and reacting when they make a mistake.
No one has the answers yet.
Side note, advice to all college students: I learned a lot from my CS degree, but I also got a lot of unexpected value from the courses that prepare you to answer the big-picture questions about life, namely the humanities. YMMV. But having worked at all levels of our field, from junior dev to architect, manager, director, VP, startup co-founder, I can tell you that my college's liberal arts/humanities requirements only helped me more and more every year, becoming essential once I was responsible for larger teams. If I had to do it over again, I'd be more careful about selecting those courses and getting more from them. The question of "how to use AI in education" is a perfect example: what expertise do think will be required to figure this out?
They gave Wesley a laser scalpel, to teach him to chisel wood, with whatever his mind could picture.
Wesley complained, he said, "dood, this isn't art! i'm not even chiseling!"
And the mentor-kidnapper said, "No, Wesley, you are doing art. The art is in your mind. The tool is just doing the work."
And so then Wesley threw the perfect wooden dolphin at the wall, infuriated that all the pleasure had been removed from the art. But he lived the rest of his life on that planet, making perfect wooden laser sculpted dolphins, cause that was his job and his boss told him too bad.
For a little while Wesley made popsicle stick art during his free time at night to try and feel the sensation of choice and control, but eventually just gave up and played mariokart until he was an old man.
File under "old man yells at clouds."
I think it's worth asking what a CS degree is for. Is it
to teach you how to code and make you obviously hire-able
by industry? Maybe it's to teach you how to think about
data and code? If you think it's the former, then AI is
probably going to be a problem for you. If the latter,
it's neutral to somewhat good.
I've been hiring CS grads for 4 decades (yikes!) and
there's definitely been a shift. Not all existing
programs are bad. Not all young software devs are bad.
But... the percentage of recent grads that seem to
understand what data processing systems are about has
definitely fallen.
Us old farts in industry must share some of the blame
here. Our industry seems to be pretty boom or bust.
Someone comes in with a stack of money and tells people to
build { the web | blockchain | AI | virtual worlds | etc }
and suddenly there's a rush. At the start of the gold
rush, we're not sure how to build everything that's going
to be the next google [[1]] so we fill the room with young
kids who have seen some recent tooling and will work
cheap. We don't develop them. We want them to be
compliant drones dancing to a tune played by a CTO and
VP/Eng who got their jobs because they're the children of
the fund investors who gave the startup a bunch of money.
And then we go back to Universities and tell them "OMG!
Your not bar-raising your grads by teaching them __fill in
the blank__!" [[2]]
So... I don't doubt it looks like AI is destroying the
degree. But maybe it's just making it hard to ignore
previous problems with the uni to industry pipeline.
And I didn't mention the Mythical Man Month once in this
comment. I'm getting better.
1. I'm not sure too many people want to be Google at the
moment, but over the last 25 years, they've certainly
done well.
2. I'm intentionally misspelling "you're" because in my
time in industry, I've found only about 15% of people
would pass a 7th grade spelling test. What are you going to do with your new-found CS knowledge?
Are you going to make the next google? Are you going to
get a job at the local insurance company processing
claims? Teaching yourself without a degree is a bad path
if you want to do grad school.
Maybe you want to understand how software works so you can
have a better understanding of the world and participate
more fully in society? Well... I can't argue with that.
I'm bearish on using AI to teach yourself CS because AI
doesn't teach people CS. If you learn something about CS
because you're looking at the output of an AI model, it's
sort of a side-effect.
We probably have a very different idea about what "CS" is.
I think of it as an organized exploration of notation,
processes and proofs related to a branch of applied
mathematics whose subject is general purpose computing
devices. Asking an AI "give me a function to calculate
the fibonacci sequence in rust" isn't going to help you
understand a) where the fibonacci sequence came from, b)
what recursion is (and why and how it was introduced) and
c) how the lexemes in a programming language map to an
abstract process which when reified delivers the fibonnaci
sequence.
I've picked a pretty simple example as a straw-man, mostly
because I think it //should// be knocked down. If this is
what you're doing to "learn CS" then more power to you,
but you're probably not doing yourself any favours.
You probably can use AI with prompts like "generate
example code that implements the abstract factory pattern
and explain how each function relates to the pattern" and
get a fairly decent explanation. Modern agents can keep
enough of a context you can ask it some decent questions
if its examples aren't entirely clear.
I've used and seen other people use Claude Code to
interatively generate design work. Mostly APIs. It's
hardly ever right the first time and Claude Code sometimes
barfs if you ask it to explain why it did a particular
thing. But it's mostly decent and in a few cases it has
pointed out a few ragged edges I didn't notice.
But... I learned how to map between human requirements and
software on my own. I don't know if someone could learn
expert level design by examining the output of a tool that
was trained on mediocre code.
And Claude Code really doesn't understand how to interpret
my prompts about TLA+ and Coq/Roq. It doesn't generate
complete BS. I'm thinking there's just not enough
training data included.
In the old days, if you went to a decent school like
Harvard or Yale or CMU, you could be reasonably assured
the education wouldn't be bad. [[1]] You could go to a
cheaper state state school and get a decent education,
though there was always the chance you would get a crap
instructor or two.
There's an idea that you can teach most people to be
effective software developers by putting them in front of
a terminal with a man page and a c compiler. I believe
this is bullshit. This technique will work for //some//
people, but not as many as sitting you in a class-room and
having an expert explain what the important bits are and
stepping you through the reasoning of why.
I think "learning by AI" may work for //some// people, if
they figure out how to ask the right questions. I
recommend reading decent books like Gries' "The Science of
Programming" or Knuth's "Literate Programming." (You also
have to read SICP because it defines a base level
understanding that all those CS grads have supposedly
learned.)
But I'm reminded of the end of the movie Hidden Dragon,
Crouching Tiger (spoilers ahead) where it is revealed the
antagonist learned kung fu from a well regarded book, but
since they couldn't read, they only looked at the
pictures. They understood the forms but not the
motivations for why the forms were developed. The
antagonist was a formatable foe, but was eventually
defeated because that minor difference in context was
sufficient to lead to their defeat.
I don't think you can learn to be an expert computer
scientist exclusively by reading books / papers. But you
can come close. I don't think you can learn to be an
expert computer scientist exclusively by asking AI to give
you example code, explaining it and critiquing your code.
Mostly because "being an expert computer scientist" is
less about being a 10x coder and more about understanding
the relations between abstract concepts, notation and
symbolic manipulation, how concrete problems are
represented and manipulated and how this all relates to
concretization using computing machinery.
AI gives you an answer, not the understanding embedded in
the text from which the answer was derived.
But if you do it right and you ask AI the right questions,
that understanding can be revealed.
I think my sense is that it is very easy to use AI wrong
and difficult(ish) to use AI as a learning tool. I
sincerely wish you well and would love to read a blog
about your experience.
1. Do not ask about MIT. They're beyond salvaging.I'm also now reminded that this is ran by that same guy who complained about the MIT license allowing people to do whatever they wish.
1) Get the degree, do not drop out - In our society educational status signaling still counts and plays a good huge part in whether you can successfully get a job, do research, get tenure etc.
2) Learn AI to offload labour but not your brain - AI exhibits "Jagged Intelligence" and is "Probabilistic". So you must confine it within explicit guardrails. Always have the "Map of the entire Forest" in your Head and have AI contribute only in constructing "the Trees within the Forest" and that too where you know the peaks of its jagged intelligence lie.
3) Focus on end-to-end Software Engineering and learn how to do everything yourself with no external dependencies other than AI. Specifically, learn to map Requirements -> Formal Specifications and do Formal Verification against it. You need to become the overall "Systems Engineer" (as it used to designated in the old days).
4) In addition, use AI to learn about business functions viz. marketing/sales, accounting, legal etc. so you can setup yourself easily as a OPC. This is your Nirvana.
5) Because AI can go deep in any one particular area, cultivating cross and inter disciplinary knowledge, studies, skills are more important today. You must truly build a "M/T" shaped skillset.
6) Always keep in mind Grant Allen's dictum viz. "I have never let Schooling interfere with my Education". AI is an excellent Socratic dialogue partner to learn anything quickly. Practice Maieutic Prompting with AI so you can appropriately constrain it and accelerate your learning curve.
7) Never ever stop learning something actively (can be Mental, Verbal or Physical). The subject you are learning can be anything but the key is to do it with mindful self-effort. It has been shown that people who are curious, continuously learning and figuring things out for themselves keep their mental acuity well into old age.
PS: In order to motivate you, here is something for you to figure out and learn.
Set Theory/Propositional/Predicate Calculus - Curry-Howard Isomorphism - Lambda Calculus/Simple Type Theory - CoC/CIC
PPS: If you fully understand it, your professor will award you the degree immediately ;-)
Unis usually watch industrial revolutions from far and have several years to experiment and come up with an approach that mostly works before pushing students through those policies and coursework. This revolution threatens to destroy their business, so they feel an urgent need to do something, even if slapdash policies much better changed quickly.
AI is throwing a wrench in the normal workflows because:
(1) it can be seen as an existential threat to Unis (either their business model or the value of having their diploma),
(2) it is undercutting the value of a Uni degree because that diploma doesn't increase your odds of getting a good job right now,
(3) even people in the workplace using AI aren't working with consistent policies. Tokenmaxxing, then Token-efficiency. You might be prosecuted for a crime if you put confidential customer data into the LLM machine (my CEO sent an email warning about this to all employees), but all of your incentives are to move so fast that might be the only option available. Most of the time it has more to do with negotiating the uncertainty around what is the role of the human in the SDLC (ideation, planning, design, coding, testing, review, maintenance, iteration, coordination, etc) and what is the role of the many agents.
(4) Students are coming to Uni with AI as a habit and might need to change those habits to adhere to Uni policies. The little ones are getting access to LLMs early so the habits are formed early.
(5) Teachers / professors have always had to contend with cheating (how to detect and how to punish). Teachers can use a calculator to build the test (eg. Ensure the contrived numbers on a math equation are solvable and to verify the answers on their answer sheet). That doesn't mean that students should have access to a calculator if the purpose of that unit is to learn the concept without the use of tools.
As frustrating as it is for you, it is almost certainly as frustrating for the professors and department heads.
The cost cutting is a normal headache in all aspects of life. Welcome to capitalism.
Provide examples.
It's hard for readers to gauge whether you're complaints are legitimate or overreactions.
> The amount of work involved has also tripled
are hard to evaluate if I don't know what the work was before. Nor can I compare with my education to see if the workload is reasonable.
> I am here to learn, god damn it, not barely scratch the surface of the 15 deliverables I have for a 3,000 word assignment.
What type of CS courses require a 3000 word assignment?
The advice I give you is the same I'm trying to believe in myself: just chill dude... AI isn't human and while can't be one, you are "safe"... We are still figuring it out while it happen to find another way to "feel valuable" after code-writing is no longer the source of our self-value.
Hold onto the roller coaster hand rail, as we become the agrarian society we were all along.
Energy farmers in the desert.
Every farmer fixing his own tractor.
Thats where we come from.
The "everyone is special, and every product is specialized" period is ending.
Everyone is an expert on exactly the things they decided they wanted to be an expert on now.
Thats where we're headed.
If you listen carefully, you can hear the sound of humans being plugged into the matrix, and downloading their fill of cartridges, in ever increasing numbers.
Talk is cheap again.
See you at the replicator parts store. :}
No one needs AI in universities or software engineering. It only benefits the middle managers and mediocre people who jump on any bandwagon in order to help their careers.
No software has gotten any better since AI, except that some forgotten bugs from the 1990s were found, just as 20000+ bugs have been found by the Luddite OssFuzz system before AI.
EDIT: zamadatix is an administrator. They felt offended by the swipe at mediocrities.
> Whereas before, assignments were designed by a person who thought them through at least once, then approved by other people in the department, now I doubt they're getting read more than once (if that). Errors abound, often with corrections coming a week or two before the deadline.
This is entirely unacceptable. Professors who are using AI to generate assignments and course loads need to be discipline and removed if they can’t start doing their job.
From the title I thought I was going to be reading an article wondering how AI would devalue a CS degree, but this is actually a story about a CS program falling apart because the instructors are flailing about without AI.
Why? How many times have these same assignments been created, semester after semester for decades before we see how wasteful this is? These are solved problems in 99% of cases and just busy work for the staff to justify their existence. Their job should be to lecture and assist students, not create endless assignments that just regurgitate what was already done a thousand times already.