Tootfinder

Should we teach vibe coding? Here's why not.
Should AI coding be taught in undergrad CS education?
1/2
I teach undergraduate computer science labs, including for intro and more-advanced core courses. I don't publish (non-negligible) scholarly work in the area, but I've got years of craft expertise in course design, and I do follow the academic literature to some degree. In other words, In not the world's leading expert, but I have spent a lot of time thinking about course design, and consider myself competent at it, with plenty of direct experience in what knowledge & skills I can expect from students as they move through the curriculum.
I'm also strongly against most uses of what's called "AI" these days (specifically, generative deep neutral networks as supplied by our current cadre of techbro). There are a surprising number of completely orthogonal reasons to oppose the use of these systems, and a very limited number of reasonable exceptions (overcoming accessibility barriers is an example). On the grounds of environmental and digital-commons-pollution costs alone, using specifically the largest/newest models is unethical in most cases.
But as any good teacher should, I constantly question these evaluations, because I worry about the impact on my students should I eschew teaching relevant tech for bad reasons (and even for his reasons). I also want to make my reasoning clear to students, who should absolutely question me on this. That inspired me to ask a simple question: ignoring for one moment the ethical objections (which we shouldn't, of course; they're very stark), at what level in the CS major could I expect to teach a course about programming with AI assistance, and expect students to succeed at a more technically demanding final project than a course at the same level where students were banned from using AI? In other words, at what level would I expect students to actually benefit from AI coding "assistance?"
To be clear, I'm assuming that students aren't using AI in other aspects of coursework: the topic of using AI to "help you study" is a separate one (TL;DR it's gross value is not negative, but it's mostly not worth the harm to your metacognitive abilities, which AI-induced changes to the digital commons are making more important than ever).
So what's my answer to this question?
If I'm being incredibly optimistic, senior year. Slightly less optimistic, second year of a masters program. Realistic? Maybe never.
The interesting bit for you-the-reader is: why is this my answer? (Especially given that students would probably self-report significant gains at lower levels.) To start with, [this paper where experienced developers thought that AI assistance sped up their work on real tasks when in fact it slowed it down] (https://arxiv.org/abs/2507.09089) is informative. There are a lot of differences in task between experienced devs solving real bugs and students working on a class project, but it's important to understand that we shouldn't have a baseline expectation that AI coding "assistants" will speed things up in the best of circumstances, and we shouldn't trust self-reports of productivity (or the AI hype machine in general).
Now we might imagine that coding assistants will be better at helping with a student project than at helping with fixing bugs in open-source software, since it's a much easier task. For many programming assignments that have a fixed answer, we know that many AI assistants can just spit out a solution based on prompting them with the problem description (there's another elephant in the room here to do with learning outcomes regardless of project success, but we'll ignore this over too, my focus here is on project complexity reach, not learning outcomes). My question is about more open-ended projects, not assignments with an expected answer. Here's a second study (by one of my colleagues) about novices using AI assistance for programming tasks. It showcases how difficult it is to use AI tools well, and some of these stumbling blocks that novices in particular face.
But what about intermediate students? Might there be some level where the AI is helpful because the task is still relatively simple and the students are good enough to handle it? The problem with this is that as task complexity increases, so does the likelihood of the AI generating (or copying) code that uses more complex constructs which a student doesn't understand. Let's say I have second year students writing interactive websites with JavaScript. Without a lot of care that those students don't know how to deploy, the AI is likely to suggest code that depends on several different frameworks, from React to JQuery, without actually setting up or including those frameworks, and of course three students would be way out of their depth trying to do that. This is a general problem: each programming class carefully limits the specific code frameworks and constructs it expects students to know based on the material it covers. There is no feasible way to limit an AI assistant to a fixed set of constructs or frameworks, using current designs. There are alternate designs where this would be possible (like AI search through adaptation from a controlled library of snippets) but those would be entirely different tools.
So what happens on a sizeable class project where the AI has dropped in buggy code, especially if it uses code constructs the students don't understand? Best case, they understand that they don't understand and re-prompt, or ask for help from an instructor or TA quickly who helps them get rid of the stuff they don't understand and re-prompt or manually add stuff they do. Average case: they waste several hours and/or sweep the bugs partly under the rug, resulting in a project with significant defects. Students in their second and even third years of a CS major still have a lot to learn about debugging, and usually have significant gaps in their knowledge of even their most comfortable programming language. I do think regardless of AI we as teachers need to get better at teaching debugging skills, but the knowledge gaps are inevitable because there's just too much to know. In Python, for example, the LLM is going to spit out yields, async functions, try/finally, maybe even something like a while/else, or with recent training data, the walrus operator. I can't expect even a fraction of 3rd year students who have worked with Python since their first year to know about all these things, and based on how students approach projects where they have studied all the relevant constructs but have forgotten some, I'm not optimistic seeing these things will magically become learning opportunities. Student projects are better off working with a limited subset of full programming languages that the students have actually learned, and using AI coding assistants as currently designed makes this impossible. Beyond that, even when the "assistant" just introduces bugs using syntax the students understand, even through their 4th year many students struggle to understand the operation of moderately complex code they've written themselves, let alone written by someone else. Having access to an AI that will confidently offer incorrect explanations for bugs will make this worse.
To be sure a small minority of students will be able to overcome these problems, but that minority is the group that has a good grasp of the fundamentals and has broadened their knowledge through self-study, which earlier AI-reliant classes would make less likely to happen. In any case, I care about the average student, since we already have plenty of stuff about our institutions that makes life easier for a favored few while being worse for the average student (note that our construction of that favored few as the "good" students is a large part of this problem).
To summarize: because AI assistants introduce excess code complexity and difficult-to-debug bugs, they'll slow down rather than speed up project progress for the average student on moderately complex projects. On a fixed deadline, they'll result in worse projects, or necessitate less ambitious project scoping to ensure adequate completion, and I expect this remains broadly true through 4-6 years of study in most programs (don't take this as an endorsement of AI "assistants" for masters students; we've ignored a lot of other problems along the way).
There's a related problem: solving open-ended project assignments well ultimately depends on deeply understanding the problem, and AI "assistants" allow students to put a lot of code in their file without spending much time thinking about the problem or building an understanding of it. This is awful for learning outcomes, but also bad for project success. Getting students to see the value of thinking deeply about a problem is a thorny pedagogical puzzle at the best of times, and allowing the use of AI "assistants" makes the problem much much worse. This is another area I hope to see (or even drive) pedagogical improvement in, for what it's worth.
1/2

Joking aside, this advice is generally pretty good. Reducing social media usage *definitely* improves my mental health. The less time I use my phone, the better my focus is. Dopamine purges *really* help increasing creativity. Most of the things he's saying are pretty spot on, but I literally can't finish some tasks without help. My brain is just wired in a way that makes listening to a podcast while doing chores basically necessary. I just can't go to a normal gym. I go to a climbing gym because it make exercise actually interesting.
There are a lot of hacks ADHD folks have to use to do "normal" stuff. Neurotypical folks just can't seem to comprehend that.

The UK starts enforcing new online child safety laws, requiring websites that host porn, self-harm, suicide, and eating disorder content to verify users' age (Wired)
https://www.wired.com/story/the-age-checked-internet-has-arrived/

The Age-Checked Internet Has Arrived
Starting today, UK adults will have to prove their age to access porn online. Experts warn that a global wave of age-check laws threatens to chill speech and ultimately harm children and adults alike.

Granite Embedding R2 Models
Parul Awasthy, Aashka Trivedi, Yulong Li, Meet Doshi, Riyaz Bhat, Vignesh P, Vishwajeet Kumar, Yushu Yang, Bhavani Iyer, Abraham Daniels, Rudra Murthy, Ken Barker, Martin Franz, Madison Lee, Todd Ward, Salim Roukos, David Cox, Luis Lastras, Jaydeep Sen, Radu Florian
https://arxiv.org/abs/2508.21085

Granite Embedding R2 Models
We introduce the Granite Embedding R2 models, a comprehensive family of high-performance English encoder-based embedding models engineered for enterprise-scale dense retrieval applications. Building upon our first-generation release, these models deliver substantial improvements, including 16x expanded context length (8,192 tokens), state-of-the-art performance across diverse retrieval domains - text, code, long-document search, multi-turn conversational, and tabular data - and measurable speed…

Just read this post by @… on an optimistic AGI future, and while it had some interesting and worthwhile ideas, it's also in my opinion dangerously misguided, and plays into the current AGI hype in a harmful way.
https://social.coop/@eloquence/114940607434005478
My criticisms include:
- Current LLM technology has many layers, but the biggest most capable models are all tied to corporate datacenters and require inordinate amounts of every and water use to run. Trying to use these tools to bring about a post-scarcity economy will burn up the planet. We urgently need more-capable but also vastly more efficient AI technologies if we want to use AI for a post-scarcity economy, and we are *not* nearly on the verge of this despite what the big companies pushing LLMs want us to think.
- I can see that permacommons.org claims a small level of expenses on AI equates to low climate impact. However, given current deep subsidies on place by the big companies to attract users, that isn't a great assumption. The fact that their FAQ dodges the question about which AI systems they use isn't a great look.
- These systems are not free in the same way that Wikipedia or open-source software is. To run your own model you need a data harvesting & cleaning operation that costs millions of dollars minimum, and then you need millions of dollars worth of storage & compute to train & host the models. Right now, big corporations are trying to compete for market share by heavily subsidizing these things, but it you go along with that, you become dependent on them, and you'll be screwed when they jack up the price to a profitable level later. I'd love to see open dataset initiatives SBD the like, and there are some of these things, but not enough yet, and many of the initiatives focus on one problem while ignoring others (fine for research but not the basis for a society yet).
- Between the environmental impacts, the horrible labor conditions and undercompensation of data workers who filter the big datasets, and the impacts of both AI scrapers and AI commons pollution, the developers of the most popular & effective LLMs have a lot of answer for. This project only really mentions environmental impacts, which makes me think that they're not serious about ethics, which in turn makes me distrustful of the whole enterprise.
- Their language also ends up encouraging AI use broadly while totally ignoring several entire classes of harm, so they're effectively contributing to AI hype, especially with such casual talk of AGI and robotics as if embodied AGI were just around the corner. To be clear about this point: we are several breakthroughs away from AGI under the most optimistic assumptions, and giving the impression that those will happen soon plays directly into the hands of the Sam Altmans of the world who are trying to make money off the impression of impending huge advances in AI capabilities. Adding to the AI hype is irresponsible.
- I've got a more philosophical criticism that I'll post about separately.
I do think that the idea of using AI & other software tools, possibly along with robotics and funded by many local cooperatives, in order to make businesses obsolete before they can do the same to all workers, is a good one. Get your local library to buy a knitting machine alongside their 3D printer.
Lately I've felt too busy criticizing AI to really sit down and think about what I do want the future to look like, even though I'm a big proponent of positive visions for the future as a force multiplier for criticism, and this article is inspiring to me in that regard, even if the specific project doesn't seem like a good one.

How to tell a vibe coder of lying when they say they check their code.
People who will admit to using LLMs to write code will usually claim that they "carefully check" the output since we all know that LLM code has a lot of errors in it. This is insufficient to address several problems that LLMs cause, including labor issues, digital commons stress/pollution, license violation, and environmental issues, but at least it's they are checking their code carefully we shouldn't assume that it's any worse quality-wise than human-authored code, right?
Well, from principles alone we can expect it to be worse, since checking code the AI wrote is a much more boring task than writing code yourself, so anyone who has ever studied human-computer interaction even a little bit can predict people will quickly slack off, stating to trust the AI way too much, because it's less work. I'm a different domain, the journalist who published an entire "summer reading list" full of nonexistent titles is a great example of this. I'm sure he also intended to carefully check the AI output, but then got lazy. Clearly he did not have a good grasp of the likely failure modes of the tool he was using.
But for vibe coders, there's one easy tell we can look for, at least in some cases: coding in Python without type hints. To be clear, this doesn't apply to novice coders, who might not be aware that type hints are an option. But any serious Python software engineer, whether they used type hints before or not, would know that they're an option. And if you know they're an option, you also know they're an excellent tool for catching code defects, with a very low effort:reward ratio, especially if we assume an LLM generates them. Of the cases where adding types requires any thought at all, 95% of them offer chances to improve your code design and make it more robust. Knowing about but not using type hints in Python is a great sign that you don't care very much about code quality. That's totally fine in many cases: I've got a few demos or jam games in Python with no type hints, and it's okay that they're buggy. I was never going to debug them to a polished level anyways. But if we're talking about a vibe coder who claims that they're taking extra care to check for the (frequent) LLM-induced errors, that's not the situation.
Note that this shouldn't be read as an endorsement of vibe coding for demos or other rough-is-acceptable code: the other ethical issues I skipped past at the start still make it unethical to use in all but a few cases (for example, I have my students use it for a single assignment so they can see for themselves how it's not all it's cracked up to be, and even then they have an option to observe a pre-recorded prompt session instead).

LLM coding is the opposite of DRY
An important principle in software engineering is DRY: Don't Repeat Yourself. We recognize that having the same code copied in more than one place is bad for several reasons:
1. It makes the entire codebase harder to read.
2. It increases maintenance burden, since any problems in the duplicated code need to be solved in more than one place.
3. Because it becomes possible for the copies to drift apart if changes to one aren't transferred to the other (maybe the person making the change has forgotten there was a copy) it makes the code more error-prone and harder to debug.
All modern programming languages make it almost entirely unnecessary to repeat code: we can move the repeated code into a "function" or "module" and then reference it from all the different places it's needed. At a larger scale, someone might write an open-source "library" of such functions or modules and instead of re-implementing that functionality ourselves, we can use their code, with an acknowledgement. Using another person's library this way is complicated, because now you're dependent on them: if they stop maintaining it or introduce bugs, you've inherited a problem, but still, you could always copy their project and maintain your own version, and it would be not much more work than if you had implemented stuff yourself from the start. It's a little more complicated than this, but the basic principle holds, and it's a foundational one for software development in general and the open-source movement in particular. The network of "citations" as open-source software builds on other open-source software and people contribute patches to each others' projects is a lot of what makes the movement into a community, and it can lead to collaborations that drive further development. So the DRY principle is important at both small and large scales.
Unfortunately, the current crop of hyped-up LLM coding systems from the big players are antithetical to DRY at all scales:
- At the library scale, they train on open source software but then (with some unknown frequency) replicate parts of it line-for-line *without* any citation [1]. The person who was using the LLM has no way of knowing that this happened, or even any way to check for it. In theory the LLM company could build a system for this, but it's not likely to be profitable unless the courts actually start punishing these license violations, which doesn't seem likely based on results so far and the difficulty of finding out that the violations are happening. By creating these copies (and also mash-ups, along with lots of less-problematic stuff), the LLM users (enabled and encouraged by the LLM-peddlers) are directly undermining the DRY principle. If we see what the big AI companies claim to want, which is a massive shift towards machine-authored code, DRY at the library scale will effectively be dead, with each new project simply re-implementing the functionality it needs instead of every using a library. This might seem to have some upside, since dependency hell is a thing, but the downside in terms of comprehensibility and therefore maintainability, correctness, and security will be massive. The eventual lack of new high-quality DRY-respecting code to train the models on will only make this problem worse.
- At the module & function level, AI is probably prone to re-writing rather than re-using the functions or needs, especially with a workflow where a human prompts it for many independent completions. This part I don't have direct evidence for, since I don't use LLM coding models myself except in very specific circumstances because it's not generally ethical to do so. I do know that when it tries to call existing functions, it often guesses incorrectly about the parameters they need, which I'm sure is a headache and source of bugs for the vibe coders out there. An AI could be designed to take more context into account and use existing lookup tools to get accurate function signatures and use them when generating function calls, but even though that would probably significantly improve output quality, I suspect it's the kind of thing that would be seen as too-baroque and thus not a priority. Would love to hear I'm wrong about any of this, but I suspect the consequences are that any medium-or-larger sized codebase written with LLM tools will have significant bloat from duplicate functionality, and will have places where better use of existing libraries would have made the code simpler. At a fundamental level, a principle like DRY is not something that current LLM training techniques are able to learn, and while they can imitate it from their training sets to some degree when asked for large amounts of code, when prompted for many smaller chunks, they're asymptotically likely to violate it.
I think this is an important critique in part because it cuts against the argument that "LLMs are the modern compliers, if you reject them you're just like the people who wanted to keep hand-writing assembly code, and you'll be just as obsolete." Compilers actually represented a great win for abstraction, encapsulation, and DRY in general, and they supported and are integral to open source development, whereas LLMs are set to do the opposite.
[1] to see what this looks like in action in prose, see the example on page 30 of the NYTimes copyright complaint against OpenAI (#AI #GenAI #LLMs #VibeCoding