AI Coding
Posted4 months agoActive4 months ago
geohot.github.ioTechstoryHigh profile
heatedmixed
Debate
85/100
AI CodingLlmsProgramming Productivity
Key topics
AI Coding
Llms
Programming Productivity
The article by geohot critiques the effectiveness of AI-assisted coding, sparking a lively debate among HN commenters about the benefits and limitations of LLMs in programming.
Snapshot generated from the HN discussion
Discussion Activity
Very active discussionFirst comment
1h
Peak period
136
0-12h
Avg / period
26.7
Comment distribution160 data points
Loading chart...
Based on 160 loaded comments
Key moments
- 01Story posted
Sep 13, 2025 at 5:28 AM EDT
4 months ago
Step 01 - 02First comment
Sep 13, 2025 at 6:32 AM EDT
1h after posting
Step 02 - 03Peak activity
136 comments in 0-12h
Hottest window of the conversation
Step 03 - 04Latest activity
Sep 19, 2025 at 2:41 PM EDT
4 months ago
Step 04
Generating AI Summary...
Analyzing up to 500 comments to identify key contributors and discussion patterns
ID: 45230677Type: storyLast synced: 11/20/2025, 8:18:36 PM
Want the full context?
Jump to the original sources
Read the primary article or dive into the live Hacker News thread when you're ready.
Adderall is similar. It makes people feel a lot more productive, but research on its effectiveness[0] seems to show that, at best, we get only a mild improvement in productivity, and marked deterioration of cognitive abilities.
[0] https://pmc.ncbi.nlm.nih.gov/articles/PMC6165228/
If legitimate research had found it to be drastically better, that study would definitely have been published in a big way.
Unscientifically, I personally know quite a number of folks that sincerely believed that they couldn’t function without it, but have since learned that they do far better on their own. I haven’t met a single one that actually had their productivity decline (after an adjustment period, of course), after giving up Adderall. In fact, I know several, that have had their careers really take off, after giving it up.
"Antibiotics don't improve your life, but can damage your health" would likely be the outcome on 13 randomly selected healthy individuals. But do the same study on 13 people with a bacterial infection susceptible to antibiotics and your results will be vastly different.
They'll need to learn, the same way I see lots of people learn.
It's been around long enough, though, that all the Adderall-eating people should have established a Gattaca-like "elite," with all us "undermen," scrabbling around at their feet.
Not sure why that never happened...
It's not that they make you superhuman - I don't think I'm any "smarter" on them. It's just that without them, my "productive" bursts would be at really random times (11PM or 2AM), and make it very hard to fit into a "normal" schedule with the rest of society (I would frequently fall asleep in classes or meetings).
So it's more that it allows you to "rearrange" the same (or slightly larger) amount of work but into a more reasonable, traditional schedule. And for certain things, helps more than that. But it's not a miracle drug by any means.
It seems like that with such small groups and effects you could run the same “study” again and again until you get the result that you initially desired.
https://pmc.ncbi.nlm.nih.gov/articles/PMC3489818/table/tbl1/
People without ADHD take it, believing that it makes them “super[wo]men.”
That said, I'll leave the conclusions about whether it's valuable for those with ADHD to the mental health professionals.
The ADHD was caught early, and treated, but the dyslexia was not. He thought he was a moron, for much of his early life, and his peers and employers did nothing to discourage that self-diagnosis.
Since he learned of his dyslexia, and started treating it, he has been an engineer at Intel, for most of his career (not that I envy him, right now).
A lot of it. Other stuff, too, that makes it look like a cup of weak tea, but it’s been 45 years.
I’m pretty familiar with the illusion of productivity.
>I’m pretty familiar with the illusion of productivity
II do not have ADHD and I'm familiar with successfully using Aderrall for actual productivity. YMMV. Turns out not everyone's brain chemistry is the same.
I just make sure there's a place to go, if the wheels fall off.
Cheers.
Their benefits when used as intended are solidly documented in research literature.
Researching concepts, for one, has become so much easier, especially for things where you don’t know anything yet and would have a hard time to even formulate a search engine query.
LLMs are really valuable for finding information that you aren't able to formulate a proper search query for.
To get the most out of them, ask them to point you to reliable sources instead of explaining directly. Even then, it pays to be very critical of where they're leading you to. To make an LLM the primary window through which you seek new information is extremely precarious epistemologically. Personally, I'd use it as a last resort.
I use LLM to do things like brainstorm, explaining programming concepts and debug. I will not use it to write code. The output is not good enough, and I feel dumber.
I only see the worst of my programming collegues coding with AI. And the results are actual trash. They have no actual understanding of the code "they" are writing, and they have no idea how to actually debug what "they" made, if LLM is not helpful. I can smell the technical debt.
You know, aside from AI making it super easy and fast to generate this tech debt in whatever amounts they desire?
AI removes that need. You don't need to know what the function does at all, so your brain devotes no energy towards remembering or understanding it.
I used to be a bit more open minded on this topic but im increasingly viewing any programmers who use AI for anything other than brainstorming and looking stuff up/explaining it as simply bad at what they do.
Not a billion dollar business yet, maybe, but 300 cars generating five or six figures revenue per year each isn't far off.
(And I say this as someone who is skeptical that totally autonomous cars worldwide will ever be a thing, but you can get to £10Bn far, far before that point. Become the dominant mode of transport in just ONE major American city and you're most of the way there).
Because geo fenced driving in a few select cities with very favourable conditions is not what was promised. That's the crux. They promised us that we have self drive anywhere at anytime at the press of a button.
> Not a billion dollar business yet, maybe, but 300 cars generating five or six figures revenue per year each isn't far off.
I'm not sure how you get to 6 figures revenue. Assuming the car makes $100 per hour for 24x7 52 weeks a year we still fall short of 1 million. But let's assume you're right $300M revenue (not profit, are they even operating at a plus even disregarding R&D costs?) on investment of >10 billion (probably more like 100), seems like the definition of hype.
> (And I say this as someone who is skeptical that totally autonomous cars worldwide will ever be a thing, but you can get to £10Bn far, far before that point. Become the dominant mode of transport in just ONE major American city and you're most of the way there).
What I don't understand with this argument, how are you proposing they become the dominant mode of transport. These services are competing with taxis, what do they offer over taxis that people suddenly switch on mass to self driving taxis? They need to become cost competitive (and convenience competitive) with driving your own car, which would significantly drive down revenue. Secondly if robotaxi companies take over transport, why would the public continue to build their infrastructure and not demand that these robotaxi companies start to finance the infrastructure they exclusively use?
So yeah, right now they'd have to be at ten cities x 300 cars each to hit 300M revenue, but there's still plenty of room for growth. Or should be, assuming the Waymo model isn't maxed out supporting the current handful of cities.
But I'm not convinced they have to hit cost parity with personal cars, because the huge advantage is you can work and drive (or be driven). If NYC and LA rush-hour congestion time becomes productive time, there's your billions.
I drive but prefer to take transit for this reason - some of my colleagues are able to join work calls effectively while driving, but for whatever reason my brain doesn't allow that. Just paying attention to calls is enough, you want me to pay attention to the road AND the call?
Specifically, natural language is:
Having something at our disposal that you can write code specs in, that is as easy as natural language yet, more concise, easy to learn and most of all not so anal/rigid as typical code languages are would be fantastic.Maybe LLMs can be sued to design such a thing ?
nice misspelling (or a joke?), related to all the lawsuits around LLMs.
Joking aside, it’s already there in a sense. Several times I started with a brief outline of what the prototype should do (an HTML/CSS/JS app), and sure enough, refinements and corrections followed. When the final version worked more or less as expected, I asked the LLM to create a specification (a reproducing prompt) of everything we made together. Even if the vibe-coded prototype is dropped, the time wasn’t wasted, I probably would never have come to the same bullet list specification without having an actual working app at my disposal to test and evaluate. So paradoxically this specification even might be used by a human later
True in the long run. Like a car with a high acceleration but low top speed.
AI makes you start fast, but regret later because you don't have the top speed.
Chill. Interesting times. Learn stuff, like always. Iterate. Be mindful and intentional and don't just chase mirrors but be practical.
The rest is fluff. You know yourself.
I am not defending we should drop AI, but we should really measure its effects and take actions accordingly. It's more than just getting more productivity.
However, the challenge has shifted to code review. I now spend the vast majority of my time reading code rather than writing it. You really need to build strong code-reading muscles. My process has become: read, scrap it, rewrite it, read again… and repeat until it’s done. This approach produces good results for me.
The issue is that not everyone has the same discipline to produce well-crafted code when using AI assistance. Many developers are satisfied once the code simply works. Since I review everything manually, I often discover issues that weren’t even mentioned. During reviews, I try to visualize the entire codebase and internalize everything to maintain a comprehensive understanding of the system’s scope.
And you get to pay some big corporation for the privilege.
In the general case, the only way to convince oneself that the code truly works is to reason through it, as testing only tests particular data points for particular properties. Hence, “simply works” is more like “appears to work for the cases I tried out”.
I’m could have used an LLM to assist but then I wouldn’t have learned much.
But I did use an LLM to make a management wrapper to present a menu of options (cli right now) and call the scripts. That probably saved me an hour, easily.
That’s my comfort level for anything even remotely “complicated”.
Personal experience (data points count = 1), as a somewhat seasoned dev (>30yrs of coding), it makes me WAY faster. I confess to not read the code produced at each iteration other than skimming through it for obvious architectural code smell, but I do read the final version line by line and make a few changes until I'm happy.
Long story short: things that would take me a week to put together now take a couple of hours. The vast bulk of the time saved is not having to identify the libraries I need, and not to have to rummage through API documentation.
> Long story short: things that would take me a week to put together now take a couple of hours. The vast bulk of the time saved is not having to identify the libraries I need, and not to have to rummage through API documentation.
One of these is not true.
With libraries, it's either you HAVE to use it, so you spend time being acquainted with it (usually a couple hours to make sense of its design, the rest will come on a needed basis) or you are evaluating multiple ones (and that task is much quicker).
Of course the latter. And of course I ask the AI to help me select a libray/module/project/whatever that provides what I need. And I ask the AI to classify them by popularity/robustness. And then I apply whatever little/much I know about the space to refine the choice.
may go as far as looking at examples that use the API. And maybe rummage through the code being the API to see if I like what I see.
The whole thing is altogether still way faster than having to pick what I need by hand with my rather limited data ingestion capabilities.
And then, once I've picked one, connecting to the API's is a no-brainer with an LLM, goes super fast.
Altogether major time saved.
I'm not.
I can build anything, but often struggle with getting bogged down with all the basic work. I love AI for speed running through all the boring stuff and getting to the good parts.
I liken AI development to a developer somewhere between junior and mid-level, someone I can given a paragraph or two of thought out instructions and have them bang out an hour of work. (The potential for then stunting the growth of actual juniors into tomorrow's senior developers is a serious concern, but a separate problem to solve)
Shhh, WIP blog post (on webpipe powered blog)
https://williamcotton.com/articles/introducing-web-pipe
Yes, I wrote my own DSL, complete with BDD testing framework, to write my blog with. In Rust!
My blog source code written in webpipe:http://github.com/williamcotton/williamcotton.com
In some cases, especially with the more senior devs in my org, fear of the good parts is why they're against AI. Devs often want the inherent safety of the boring, easy stuff for a while. AI changes the job to be a constant struggle with hard problems. That isn't necessarily a good thing. If you're actually senior by virtue of time rather than skill, you can only take on a limited number of challenging things one after another before you get exhausted.
Companies need to realise that AI to go faster is great, but there's still a cognitive impact on the people. A little respite from the hardcore stuff is genuinely useful sometimes. Taking all of that away will be bad for people.
That said, some devs hate the boring easy bits and will thrive. As with everything, individuals need to be managed as individuals.
I find this hilarious. From what I've seen watching people do it, it changes the job from deep thought and figuring out a good design to pulling a lever on a slot machine and hoping something good pops out.
The studies that show diminished critical thinking have matched what i saw anecdotally pairing with people who vibe coded. It replaced deep critical thinking with a kind of faith based gambler's mentality ("maybe if i tell it to think really hard it'll do it right next time...").
The only times ive seen a notable productivity improvement is when it was something not novel that didnt particularly matter if what popped out was shit - e.g. a proof of concept, ad hoc app, something that would naturally either work or fail obviously, etc. The buzz people get from these gamblers' highs when it works seems to make them happier than if they didnt use it at all though.
To pair this with the comment you're responding to, the decline in critical thinking is probably a sign that there's many who aren't as senior as their paycheck suggests. AI will likely lead to us being able to differentiate between who the architects/artisans are, and who the assembly line workers are. Like I said, that's not a new problem, it's just that AI lays that truth bare. That will have an effect generation over generation, but that's been the story of progress in pretty much every industry for time eternal.
Is it really? Or is it a refusal to do actual software engineering, letting the machine taking care of it (deterministically) and moving up the ladder in terms of abstraction. I've seen people describing things as sludge, but they've never learned awk to write a simple script to take care of the work. Or learned how to use their editor, instead using the same pattern they would have with Notepad.
I think it's better to take a step back and reflect on why we're spending time on basic stuff instead. Instead of praying that the LLM will generate some good basic stuff.
Put differently, I go back to my original comment, where AI is essentially a junior/mid dev that you can express what needs to be done with enough detail. In either case, AI or dev, you'd review and/or verify it.
> Or is it a refusal to do actual software engineering, letting the machine taking care of it (deterministically) and moving up the ladder in terms of abstraction.
One could say the same of installing packages in most modern programming languages instead of writing the code from first principles.
I disagree, because libraries define an interface with (ideally) precise, reproducible semantics, that you make use of. They provide exactly what the grandparent is saying, namely a formal abstraction. When you have the choice between a library and an LLM, requiring equal effort, the library is clearly preferable.
When an LLM is more time-efficient at a given coding task, it can be taken as an indication of a lack of a suitable library, tooling, or other abstraction for the use case.
Ive never found this to be true once in my career.
I know a lot of devs who looked down on CRUD or whatever it was they were doing and produced boilerplate slop though.
Code isnt like pottery. There arent "artisanal" devs who produce lovely code for people to look at like it's a beautiful vase. Good code that is hooked into the right product-market fit can reach millions of people if it works well.
The world was replete with shitty code before AI and mostly it either got tossed away or it incurred epic and unnecessary maintenance costs because it actually did something useful. Nothing has changed on that front except the tsunami of shit got bigger.
1) you try to explain what you want to get done
2) you try to explain what you want to get done and how to get it done
The first one is gambling, the second one has very small failure rate, at worst, the plan it presents shows it's not getting the solution you want it to do.
Very true. I think AI (especially Claude Code) forced me to actually think hard about the problem at hand before implementing the solution. And more importantly, write down my thoughts before they fleet away from my feeble mind. A discipline that I wished I had before.
These days the only difference is that I feed my ideas to a few different LLMs to have "different opinions". Usually they're crap but sometimes they present something useful that I can implement.
That looks like plenty of hours of fun! Thanks for the link :)
There's also the fact that, while you're coding the easy stuff, your mind is thinking about the hard stuff, looking things up, seeing how they articulate. If you're spending 100% of your time on hard stuff, you might be hurting these preliminaries.
I know brilliant people who took up knitting to keep their hands busy while they think over their difficult problems. But that only works if you can knit in your work hours. Sadly, despite clearly improving the productivity of these people, this is a fireable offense in many jobs.
I'm not saying that the only way to think through a hard problem is to work on boilerplate. If you are in a workplace where you can knit, or play table soccer, by all means, and if these help you, by all means, go for it.
What I'm thinking out loud is that if we're dedicating 100% of our time to the hard problems, we'll hit a snag, and that boilerplate may (accidentally) serve as the padding that makes sure we're not at these 100%.
That being said, I'm not going to claim this as a certainty, just an idea.
The ones who are excited about this are the ones who are motivated by the product. When AI can whip up some half-baked solution it sure looks like you can focus on the product and "get someone to code it up for you". But unless it's a well-understood and previously executed solution, you're going to run into actual technical problems and have to fix them. But your motivation to deal with the irritating pedantrics of the modern computing stack (which are the same as all technology ever, with orders of magnitude more parts) hasn't been built up. There's no beneficial flywheel, just a fleet of the Sorceror's Apprentice mindless brooms that you hope you can get work enough to ship.
Monitoring AI output on any task is high arousal, low satisfaction, unless you're constantly prompting for quick wins.
I disagree that this has anything to do with people needing a break. All code eventually has to be reviewed. Regardless of who or what wrote it, writing too much of it is the problem. It's also worth considering how much more code could be eliminated if the business more critically planned what they think they want.
These tensions have existed even before computers and in all professions.
The issue of senior-juniors has always been a problem; AI simply means they're losing their hiding spots.
On the flip side, there have been lots of times where I personally didn’t have a lot of time to deeply research a topic (read papers, build prototypes of different ideas, etc) due to lack of time and resources. If all of the boring stuff is gone, and building prototypes is also 3x faster maybe what will end up happening is we can now use all of this free time to try lots of different ideas because the cost of exploration has been lowered.
That's matches my experience. In my first job, every time a new webapp project has been starting it was fun. Not because of challenges or design, but simply because of the trivial stuff done for n-th time - user accounts, login, password reset, admin panel. Probably should have been automated at this point, but we got away with reinventing the wheel each time.
Part of the problem is that in big orgs, you need to show consistent progress in order to not get put on some PIP and kicked out of the company. There are performance review cycles and you have to show something continuously.
That ONLY works if you have boring, easy work. It's easy to deliver consistent progress on that.
Interesting and difficult work is nice only if you are trusted to try your best and given the freedom to fail. That's the nature of hard problems; progress in those domains is very sudden and Poissonian and not consistent by nature. If you're going to be judged on your ability to be sub-Poissonian and consistent, and get put on a PIP for not succeeding at it one review cycle (and possibly risking income that you use to put a roof over your head or feed your family) it's not worth the career risk to try difficult things.
Not saying this is the way I think, it's just the reality of how things often work in big orgs, and one of the reasons I dislike many big orgs.
My life quality (as a startup cofounder wearing many different hats across the whole stack) would drop significantly if Cursor-like tools [1] were taken away from me, because it takes me a lot of mental effort to push myself to do the boring task, which leads to procrastination, which leads to delays, which leads to frustration. Being able to offload such tasks to AI is incredibly valuable, and since I've been in this space from "day 1", I think I have a very good grasp on what type of task I can trust it to do correctly. Here are some examples:
- Add logging throughout some code
- Turn a set of function calls that have gotten too deep into a nice class with clean interfaces
- Build a Streamlit dashboard that shows some basic stats from some table in the database
- Rewrite this LLM prompt to fix any typos and inconsistencies - yeah, "compiling" English instructions into English code also works great!
- Write all the "create index" lines for this SQL table, so that <insert a bunch of search usecases> perform well.
[1] I'm actually currently back to Copilot Chat, but it doesn't really matter that much.
That's one of the thing that I wouldn't delegate to LLM. Logging is like a report of things that happens. And just like a report, I need relevant information and the most useful information.
...
A lot of these use cases actually describes the what. But the most important questions is always the why. Why is it important to you? Or to the user? That's when things have a purpose and not be just toys.
As to why, it's because I'm building an app with a growing userbase and need to accommodate to their requirements and build new features to stay ahead of the competition. Why you decided I'm describing a toy project is beyond me.
The reason senior engineers are being paid that well is not because they need to type a lot of code to get new features in. It's because they need to figure how to have less code while having more features.
The problem is that junior developers are what we make senior developers with— so in 15 years, this is going to be yet another thing that the US used to be really good at, but is no longer capable of doing, just like many important trades in manufacturing. The manufacturers were all only concerned with their own immediate profit and made the basic sustainability of their workforce, let alone the health of the trades that supported their industries, a problem for everyone else to take care of. Well, everyone else did the same thing.
In 15 years senior developers will not be needed as well. Anyway no company is obliged to worry about 15 years timescale
Most people don’t share your confidence that we will replace senior engineers and I’d gobsmacked if we could. Just like the magical ‘automation’ can’t replace the people that actually make the physical things that the machines use to do their jobs, or fix the machines, no matter how good it gets. But the quantitatively-minded MBAs just kept kicking the can down the road and assumed it was someone else’s problem that the end of the road was approaching. It wasn’t their problem that there would be a problem in 30 years, and then it wasn’t their problem when it was 10 years, and now that we’re standing at the edge of a cliff, they’re realizing that it’s everybody’s problem and it’s going to be a hell of a lot more painful than if they’d had an extremely modest amount of foresight.
Now, US manufacturers are realizing that all of their skilled laborers are retiring or dying, and there isn’t enough time to transfer the more complex knowledge sets, like Tool and Die making, to a new set of apprentices. Many of these jobs are critical not only to national security, but also our country’s GDP because the things we do actually make are very useful, very specialized, and very expensive. Outsourcing jobs like making parts for fighter jets is really something we don’t want shipped overseas unless we want to see those parts pop up on aliexpress. If nobody is responsible for it and nobody wants to fund the government to fix it, but it is a real problem, it doesn’t take a genius to see the disconnect there.
Why the insistence on anthropomorphizing what is just a tool? It has no agency, does not 'think' in any meaningful manner, it is just pattern matching on a vast corpus of training data. That's not to say it can't be very useful - as you seem to have found - but it is still just a tool.
Not really though can it, unless that codebase and Jira happens to pattern match correctly somewhere.
I'd say that the difference is just that LLMs have a natural language interface (for good or for ill).
It's what is, to me, so bizarre about the present moment: certainly investment is exceptionally high in AI (and of course use), but the dominant position in the media is precisely such a strange 'anti-AI hype' that positions itself as a brave minority position. Obviously, OpenAI/Altman have made some unfortunate statements in self-promotion, but otherwise I genuinely can't think of something I've read that expresses the position attacked by the anti-AI-ers -- even talk of 'AGI' etc comes from the AI-critical camp.
In a sense, the world seems divided into three: obvious self-promotion from AI companies that nobody takes seriously, ever-increasingly fervent 'AI critique', and the people who, mostly silent, have found modern AI with all its warts to be an incomparably useful tool across various dimensions of their life and work. I hope the third camp becomes more vocal so that open conversations about the ways people have found AI to be useful or not can be the norm not the exception.
[1] https://antirez.com/news/154
This persona driven workflow is so weird to me. Feels like stuck in old ways.
I'm in the same boat (granted, 10 years less) but can't really relate with this. By the time any part becomes boring, I start to automate/generalize it, which is very challenging to do well. That leaves me so little boring work that I speed run through it faster by typing it myself than I could prompt it.
The parts in the middle – non-trivial but not big picture – in my experience are the parts where writing the code myself constantly uncovers better ways to improve both the big picture and the automation/generalization. Because of that, there are almost no lines of code that I write that I feel I want to offload. Almost every line of code either improves the future of the software or my skills as a developer.
But perhaps I've been lucky enough to work in the same place for long. If I couldn't bring my code with me and had to constantly start from scratch, I might have a different opinion.
The two aren't mutually exclusive. You can use AI to build your tooling. (Unless it's of sufficient complexity or value that you need to do the work yourself)
I expect that in a year my relationship with AI will be more like a TL working mostly at the requirements and task definition layer managing the work of several agents across parallel workstreams. I expect new development toolchains to start reflecting this too with less emphasis on IDEs and more emphasis on efficient task and project management.
I think the "missed growth" of junior devs is overblown though. Did the widespread adoption of higher-level really hurt the careers of developers missing out on the days when we had to do explicit memory management? We're just shifting the skillset and removing the unnecessary overhead. We could argue endlessly about technical depth being important, but in my experience this hasn't ever been truly necessary to succeed in your career. We'll mitigate these issues the same way we do with higher-level languages - by first focusing on the properties and invariants of the solutions outside-in.
I actually think this is one skill LLMs _do_ train, albeit for an entirely different reason. Claude is fairly bad at considering edge cases in my experience, so I generally have to prompt for them specifically.
Even for entirely “vibe-coded” apps I could theoretically have created without knowing any programming syntax, I was successful only because I knew about possible edge cases.
However. There's also good news. AI is also an amazing tool for learning.
So what I see AI doing is simply separating people who want to put effort forth and those who don't.
If you aren't talented enough to write or record your own music, you aren't really a musician.
If you have a quick question about music theory and you want a quick answer, AI can be a benefit.
Absolutely. For example, I've been learning Autodesk Fusion, and after establishing a small foundation through traditional learning techniques, I've been able to turbocharge my learning by asking precise questions to AI.
I image this really sucks for those whose business model relied on gatekeeping knowledge. (like training companies)
AI has also been a really good brainstorming partner - especially if you prompt it to disable sycophancy. It will tell you straight up when you are over-engineering something.
It's also wonderful at debugging.
So I just talk to my computer, brainstorm architectures and approaches, create a spec, then let it implement it. If it was a bad idea, we iterate. The iteration loop is so fast that it doesn't matter.
Did you end up regretting a design choice, but normally you'd live with it because so much code would have to be changed? Not with Agentic coding tools - they are great at implementing changes throughout the entire codebase.
And its so easy to branch out to technologies you're not an expert in, and still be really effective as you gain that expertise.
I honestly couldn't be happier than I am right now. And the tools get better every week, sometimes a couple times a week.
Can you share more about how you are prompting to disable sycophancy?
This is where AI actually helps - you have a very precise vision of what you want, but perhaps you've forgotten about the specific names of certain API methods, etc. Maybe you don't want to implement all the cases by hand. Often validating the output can take just seconds when you know what it is you're looking for.
The other part of making the output do what you want is the ability to write a prompt that captures the most essential constraints of your vision. I've noticed the ability to write and articulate ideas well in natural language terms is the actual bottleneck for most developers. It takes just as much practice communicating your ideas as it does anything else to get good at it.
Analogies to humans don't work that well. AI is super-human in some respects while also lacking the ability to continually work toward a goal over long periods of time. AI can do very little on its own - just short / scoped / supervised tasks.
However, sometimes the situation is reversed, AI is the teacher who provides some examples on how to do things or provides hints on how to explore a new area and knows how others have approached similar things. Then, sometimes, AI is an astute code reviewer, typically providing valuable feedback.
Anyway, I've stopped trying anthropomorphize AI and simply try to reason about it based on working with it. That means combinations of direct ChatGPT usage with copy / paste / amend type workflows, async style / full PR style usage, one-shot "hail Mary" type throw away PRs just to establish an initial direction as well as PR reviews of my own code. I'm using AI all the time, but never anything like how I would work with another human.
I've struggled heavily trying to figure out how to get it to write the exactly correct 10 lines of code that I need for a particularly niche problem, and so I've kind of given up on that, but getting it to write the 100 lines of code around those magic 10 lines saves me so much trouble, and opens me up to so many more projects.
I think you’re the best case support for AI coding. You know clearly what you want, so you know clearly what you don’t want. So if you had decent verbal dexterity you could prompt the AI model and manage to accomplish what you intended.
A lot of programming problems / programmer contexts don’t match that situation. Which is the problem with universalizing the potency of AI / benefits of AI coding.
But the issue is some of that speedrunning sometimes takes so much time, it becomes inefficient. It's slowly improving (gpt5 is incredible), but sometimes it get stuck on really mundane issue, and regress endlessly unless I intervene. And I am talking about straightforwars functional code.
the laziness manifest itself into productivity as crazy as this sounds. how? lazy people find a way to automate repetitive tasks. what I have learned from these over the years is that anything you do twice has to find a way to be automated as third time is around the corner :)
what does this have to do with AI? the AI has taken automation to another level allowing us to automate so much of our work that was not previously possible. I found myriad of ways to use AI and several of my best (lazy) co-workers have as well. I cannot imagine doing my work anymore without it, not because of any “magic” but because my lazy ass will be able to do all the things that I have automated out
I love AI and use it for both personal and work tasks for two reasons:
1. It's a way to bounce around ideas without (as much) bias as a human. This is indispensable because it gives you a fast feedback mechanism and validates a path.
2. It saves me typing and time. I give it one-shot, "basic work" to do and it's able to do accomplish at least 80% of what I'd say is complete. Although it may not be 100% it's still a net positive given the amount of time it saves me.
It's not lost on me that I'm effectively being trained to always add guardrails, be very specific about the instructions, and always check the work of AI.
I find it best as a "personal assistant," that I can use to give me information -sometimes, highly focused- at a moment's notice.
> The potential for then stunting the growth of actual juniors into tomorrow's senior developers is a serious concern
I think it's a very real problem. I am watching young folks being frozen out of the industry, at the very beginning of their careers. It is pretty awful.
I suspect that the executives know that AI isn't yet ready to replace senior-levels, but they are confident that it will, soon, so they aren't concerned that there aren't any more seniors being crafted from youngsters.
Would suck, if they bet wrong, though…
> This isn’t to say “AI” technology won’t lead to some extremely good tools. But I argue this comes from increased amounts of search and optimization and patterns to crib from, not from any magic “the AI is doing the coding”
* I can tell claude code to crank out some basic crud api and it will crank it out in a minute saving me an hour or so.
* I need an implementation of an algorithm that has been coded a million times on github, I ask the AI to do it and it cranks out a correct working implementation.
If I only use the AI in its wheelhouse it works very well, otherwise it sucks.
My tests with full trace level logging enabled can get very verbose. It takes serious time for a human to parse where in the 100 lines of text the relevant part is.
Just telling an AI: "Run the tests and identify the root cause" works well enough, that nowadays it is always my first step.
Did geohot not found one of these?
In any case I don't fully understand what he's trying to say other than negating the hype (which i generally agree with), but not offering any alternative thoughts of his own other than- we have bad tools and programming language. (why? how are they bad? what needs to change for them to be good?)
He's confidently wrong a lot. (Even if I happen to agree with his new, more sober take on AI coding here.)
He is right, however AI is still darn useful. He hints at why: patterns.
Writing a test suite for a new class when an existing one is in place is a breeze. It even can come up with tests you wouldnt have thought of or would have been too time pressed to check.
It also applies to non-test code too. If you have the structure it can knock a new one out.
You could have some lisp contraption that DRYs all the WETs so there is zero boilerplate. But in reality we are not crafting these perfect cosebases, we make readable, low-magic and boilerplatey code on tbe whole in our jobs.
But I do closely review the code! It turns the usual drudge of writing tests into more of a code review. Last time I did it it had some mistakes I needed to fix for sure.
A lot of the tests have those things. Boilerplate becomes knowing which to use, as well as the it.each(...) etc. ceremonies
https://en.m.wikipedia.org/wiki/Conway%27s_law
> It’s why the world wasted $10B+ on self driving car companies that obviously made no sense. There’s a much bigger market for truths that pump bags vs truths that don’t.
This reeks of bias-dismissing massive investments as ‘obvious’ nonsense while hyping its own tinygrad as the ‘truth’ in AI coding.
Author is allowed to claim ‘most people do not care to find the truth’ but it’s hypocritical when the post ignores counterpoints, like PyTorch’s dominance in efficient coding benchmarks.
Author doesn’t seem to care about finding the full truth either, just the version that pumps its bag.
Obviously... in what way? I feel the anti-ai pattern is clear.
Self-driving cars don't work in my city so the whole concept is a hoax. LLMs don't code my proprietary language so it's a bubble.
> From this study (https://arxiv.org/abs/2507.09089)
I can tell this is going to be the most misquoted study in blogs and pop-sci books after the 10,000-hour mastery study. And it's just a preprint!
- Autocomplete in Cursor. People think of AI agents first when they talk about AI coding but LLM-powered autocomplete is a huge productivity boost. It merges seamlessly with your existing workflow, prompting is just writings comments, it can edit multiple lines at once or redirect you to the appropriate part of the codebase, and if the output isn’t what you need you don’t waste much time because you can just choose to ignore it and write code as you usually do.
- Generating coding examples from documentation. Hallucination is basically a non-problem with Gemini Pro 2.5 especially if you give it the right context. This gets me up to speed on a new library or framework very quickly. Basically a stack overflow replacement.
- Debugging. Not always guaranteed to work, but when I’m stuck at a problem for too long, it can provide a solution, or give me a fresh new perspective.
- Self contained scripts. It’s ideal for this, like making package installers, cmake configurations, data processing, serverless micro services, etc.
- Understanding and brainstorming new solutions.
- Vibe coding parts of the codebase that don’t need deep integration. E.g. create a web component with X and Y feature, a C++ function that does a well defined purpose, or a simple file browser. I do wonder if a functional programming paradigm would be better when working with LLMs since by avoiding side effects you can work around their weaknesses when it comes to large codebases.
124 more comments available on Hacker News