Leave Me Behind
androidessence.com
androidessence.com
We need to remember the people, that we may never talk to, that are downstream of all of this software. Not necessarily “the users” as there are many pieces of software meant for other devs, but I think the users deserve consideration nonetheless.
Handing over software quality to the stochastic code extruder is causing a sharp drop in the quality of software put out into the world. This is on top of all of the problems that existed before LLMs, like human error and perverse financial incentives. Shipping poor quality and user hostile software actually hurts people. Real people. Harm is caused in both big and little ways to living, breathing actual people. This “inevitable” slide into generative AI harms every single person it comes into contact with. The devs, the users, the investors, everyone. Those harms may happen at different times and in different ways and the creeping nature of it all might make it easier to ignore, but it’s happening.
“AI” is a blight. You can leave me behind as well.
Well, first of all you and the author point to the same derisive comment of these models being, in your words "stochastic code extruder" or the one I have heard a lot "next-token predictors", and the connotation I read from these being that this makes them inherently dumb or unintelligent and I don't understand that. The fact that these "stochastic code extruders" can solve Erdos problems is sort of the proof in the pudding. Next token prediction is profound in that it is _a very simple objective_ yet it is _enough_ to take you to extraordinary heights.
Also I wonder how many folks honestly look in the mirror and think: how does the median programmer differ from an LLM. Do you really think humans are universally better and produce universally higher quality code? Not even universally, I would say _typically_. I would trust an LLM to not write a buffer overflow far more than a junior or a mediocre senior engineer. LLMs have built things in my domain that are non-trivial and impressive and correct.
Not to mention, these systems are following a predictable trend in performance improvement so these worries about quality just won't age well, and it seems to be a head-in-the-sand attitude that pretends like quality and reliability are not getting very very good _already_.
> Shipping poor quality and user hostile software actually hurts people.
Could not agree more. So why do you think humans are inherently better at this?
> This “inevitable” slide into generative AI harms every single person it comes into contact with.
I just don't quite understand this, is it that: (1) agentic code is inherently inferior to human code and thus (2) shipping agentic code is actively harmful?
Same with code: by using AI, one can produce passable software trinkets very cheaply, that would be uneconimical to produce by paying poor-quality human developers.
The floor has moved downwards, allowing to produce a flood of new, trash-quality, disposable code very cheaply. It does not mean that we'll have to use only that code. But unfortunately we'll have to live with it, too.
Actually, people had to be made to see it as cheap so they would throw away and re-buy more.
https://desis.osu.edu/seniorthesis/index.php/2024/09/15/crea...
> “It was a really difficult sell to the American public in the post-war period, to inculcate people into a throwaway living,” she says. “That is not what people were used to.”
> A solution companies came up with was emphasizing that plastic was a low-cost, abundant material.
> A 1960 marketing study for Scott Cup said the containers were “almost indestructible,” but that the manufacturer could still convince people to discard them after a few uses. To counter any “pangs of conscience” consumers might feel about throwing them away, the researchers suggested a “direct attack”: Tell people the cups are cheap, they said, and that “there are more where these came from.”
> A few years later, Scott ran an advertisement saying its plastic cups were available at “‘toss-away prices.”
It wasn't plastic itself, and likewise it's not "AI" itself.
We do have an abysmal track record as industrialized nations however, and more recently, in many parts of the software industry.
But we can change it. With so many things, tech people spent so much time and energy debating... like cookies or HTTPS or whatever... we often heard/said that while we care so much about doing the right thing, we can't achieve anything at work because consumers don't care. Well, this time, pretty much all of the world cares a lot. I mean, the Vatican just blogged about it!
Maybe we just "have to live with it", but in that case, there is also no utility in pointing that out, since we literally have to live with it. And of course, it's really about the shape of the "it", and how it's used, not that there is one that will never go away. That is also true about most things: stuff we don't currently use is in the museum or text books. Nothing goes away away, but we no longer drink out of lead cups, even though we still use lead. We don't have x-ray machines in shoe shops, even though we still use x-rays.
But it's worth noting that there's a new substance enabled by AI, the "slop", that is flushing violently into our world right now. Pretending it's not there, and that we won't see more of it, is perilous.
> Could not agree more. So why do you think humans are inherently better at this?
Because humans are capable of empathy
So many excited and insulted LLM adopters on this thread. There is nothing derisive in that comment, it is simply the purest possible definition of how they work. Stochastics is a branch of maths you know.
> can solve Erdos problems is sort of the proof in the pudding
For the non-engineer, non-mathematician it may sound authoritative, but you'd probably be surprised to learn that most of Erdos problems are not at all complex - they are just not very interesting or relevant. So it is a proof in the pudding, provided the pudding is made of shit - the kind of stuff LLMs produce most of the time.
> I just don't quite understand this, is it that: (1) agentic code is inherently inferior to human code and thus (2) shipping agentic code is actively harmful?
Yes and yes - have you not heard of that AWS incident with Kiro when the "agentic" shit deleted an entire infrastructure environment, complete with data, config, etc.?
> Also I wonder how many folks honestly look in the mirror and think: how does the median programmer differ from an LLM
Apart from the obvious absurdity of this statement - I know a lot of you non-engineer types feel "empowered" by the LLMs, in the sense of how they immediately seem a genius when you ask them on a topic you are not expert in, but you may want to read a book on programming first - maybe you'll get a clue then.
neither excited nor insulted.
> There is nothing derisive in that comment, it is simply the purest possible definition of how they work. Stochastics is a branch of maths you know.
Not sure what you mean by stochastics but this is more statistics. They are trained with a next token loss, that doesn't belie "how they work".
> For the non-engineer, non-mathematician it may sound authoritative, but you'd probably be surprised to learn that most of Erdos problems are not at all complex - they are just not very interesting or relevant.
It sounds like you are both an engineer and a mathematician? Can you confirm? These are problems unsolved for many years. You think no good mathematicians have taken a stab at them, even if just to say they have resolved an unsolved Erdos problem? They are "not at all complex" is quite an extraordinary thing to say I'm wondering if you actually do have the pedigree you are trying to make it sound like you have, or if you are just regurgitating the same HN talking points you've heard.
> Yes and yes - have you not heard of that AWS incident with Kiro when the "agentic" shit deleted an entire infrastructure environment, complete with data, config, etc.?
And this means agentic code is inherently inferior to human code? Howso?
> Apart from the obvious absurdity of this statement - I know a lot of you non-engineer types feel "empowered" by the LLMs, in the sense of how they immediately seem a genius when you ask them on a topic you are not expert in, but you may want to read a book on programming first - maybe you'll get a clue then.
in the beginning you mentioned there were a lot of "excited and insulted LLM adopters" and yet...this sounds quite excited and defensive. Believe it or not, I am not a "non-engineer type" and its telling you assume that people who don't seem to share the same opinion as you are somehow less qualified than I assume you think you are? Why is this statement obviously absurd. Maybe you work in a really talented engineering team, which kudos to you I also have worked in teams like this, and I have also seen what is the p50 engineer and they are just as error prone or more than Claude. Thank you for the advice to read a book on programming as if that somehow would have any bearing on this at all?
An engineer with an engineering degree, which as it may still be known to some, requires a fairly stringent mathematical underpinning. So yes, I know a thing or two - read up on Erdos and his problems, I am not here to enlighten every vibecoding PM that shows up.
> And this means agentic code is inherently inferior to human code? Howso?
Again, I am not here to explain the world to some clueless PM. You have your LLMs for that :) But for the sake of bringing you closer, the "agentic" code is often very inferior, implementing happy paths or just bluntly exposing secrets in clear texts, etc. Probably a consequence of it being trained on, as you put it "p50 engineering code".
> Maybe you work in a really talented engineering team,
Running my own company and been paying the LLM-Shit-Generators for my whole team for a long time, in the hope they would bring the advertised benefits. Guess what - for serious use-cases, they bring shit and more shit.
> Thank you for the advice to read a book on programming as if that somehow would have any bearing on this at all?
Oh yeah obviously not, I mean, its not like understanding software development would help you understand how LLMs are not similar to a "p50 engineer" at all:). I'd take the latter over the former every time.
> Why is this statement obviously absurd
Well for one, LLMs are not humans, but it should be obvious to even to most cretinous of the e/acc crowd. It's not like they can think in abstract terms or come up with completely new concepts. But then again, don't mind me - if you can live with below average AI slop - go for it.
I don't see really any hard source at all here from you except anecdotes that you seem to hold in very high regard. I do see an incredible amount of condescension and chest pounding about what is ultimately a very technical and...ahem...mathematical topic. I don't know about you but I don't really see many conference paper reviews that start with "I am not here to enlighten every vibe coding PM that shows up". I am sure you would agree with me.
> But for the sake of bringing you closer, the "agentic" code is often very inferior, implementing happy paths or just bluntly exposing secrets in clear texts, etc. Probably a consequence of it being trained on, as you put it "p50 engineering code".
I do appreciate this tiny delicious gift of "bringing me closer" because it (1) answers my question about "inherent" properties of agentic systems by giving anecdotal examples of existing systems, (2) completely misunderstands how agentic coding models are trained. Human code traces are a bootstrap to an RL with verifiable rewards stage. Not having the same "you are too beneath me to explain my wrong opinions" attitude, I will genuinely explain a bit because this isn't as trivial and obvious as you make it sound, nor is it a giant pissing contest. Likely the most important property of coding agents that has resulted in their existing and future success is that they are not limited by the quality of human training data. Seems to be a very common misconception, but this is, like you say, just math:
- Agentic coding models like Claude go through several complex training stages
- Pretraining which is kind of a compression step and gives them semantic understanding and a bit more
- Supervised fine tuning which gives them some task specific performance (this is where human traces and verified synthetic traces are used)
- Alignment to make them not give you meth recipes and to behave in the way you want agents to behave
- Reinforcement learning with verifiable rewards (RLVR): then they go forth and solve open ended questions. RLVR is not new mathematics, we know what happens when you take RL with good rewards and throw a bunch of compute at it, we've known that for decades now. This is where the "superhuman" performance comes in, it's not some "vibe coding PM" that's giving you an empty promise, it is the math that you and I so highly revere that promises you this.
> Running my own company and been paying the LLM-Shit-Generators for my whole team for a long time, in the hope they would bring the advertised benefits. Guess what - for serious use-cases, they bring shit and more shit.
This sounds like the experience I would mostly expect from a small company adopting Claude, it is not magic nor is it at the point where you can blindly trust it to not mess something up. It will waste your time. I find it kind of doubtful it has not given you any benefits, but I'm not sitting where you're sitting so I can't refute your experience. People talk resentfully about "advertised benefits" but then never cite what advertised benefits they interpreted these systems as having. Do you have like a quote or something that you can point at?
> Oh yeah obviously not, I mean, its not like understanding software development would help you understand how LLMs are not similar to a "p50 engineer" at all:). I'd take the latter over the former every time.
Maybe I misinterpreted you: I found you telling me to "read a book" to be more of a dismissive condescending comment but maybe you mean it sincerely in which case, sure I will continue to read programming books and following the published work in the field as I have done for years now.
> Well for one, LLMs are not humans, but it should be obvious to even to most cretinous of the e/acc crowd. It's not like they can think in abstract terms or come up with completely new concepts. But then again, don't mind me - if you can live with below average AI slop - go for it.
I do agree with you that LLMs are not humans but when you say this is obvious and then don't back it up, that is really not convincing. I think you overestimate the capability of human beings and underestimate the asymptotic capabilities of these systems. Their performance improvements are predictable and these predictions continue to hold. It seems the burden of proof is on you to explain why we should expect some sort of fundamental limit to these capabilities and where those fundamental limits would arise. I'm not aware of very many.
Good for you, I suppose, but all it tells me is that you have probably not developed software professionally - after all, PhDs in astrophysics "from a strong department" rarely end up in commercial software development...
> This sounds like the experience I would mostly expect from a small company adopting Claude
Who said it was a small company? You're making too many assumptions buddy :)
> will genuinely explain a bit because this isn't as trivial and obvious as you make it sound
It is literally the same technology developed in the 1940s mate, adding more GPUs will not magically make it become a god-in-the-box. How fucking innovative can you still claim it to be?
> I think you overestimate the capability of human beings and underestimate the asymptotic capabilities of these systems
Right, remember when LLMs constructed the rockets and modules for landing on the moon, using practically just the logarithmic tables? Or when they invented the vaccine? How about X-rays? Cars? Aeroplanes? You don't? Oh right, me neither! We must be downplaying their nonexistent "capabilities". And the use of word "asymptotic" - is absolutely not conveying the meaning you think it does.
> Do you have like a quote or something that you can point at?
Well, how about the CEOs of companies claiming to be worth 1T and upwards, stating that their products have almost superhuman intelligence? PhDs in the pocket etc?
Once you step out of pure-software orgs, it becomes clear that most would benefit from having AI write code. There's a huge moat between most people and the point where they can afford/find the effort of someone that can write software.
These people, that only care about practical results rather than somewhat tangential things like "elegance" and "maintainability", are going to benefit tremendously.
This claim is very misleading and not really true. It reflects the kind of exaggeration and spin made by corporate marketing. I would not call this a fact at all. Like many claims made by for-profit marketing, if one looks into the details and think critically about what is being claimed, one can see that consumers are jumping to false conclusions.
That said, it is very cool how an LLM helped human mathematicians in the recent specific Erdos problem solution announced by OpenAI. Just don't jump to the conclusion that anybody can input any Erdos problem into an LLM and a solution will come out the other end.
corporate marketing spins and hypes, but this is an ultimately pretty academic and mathematical field. The loud LinkedIn promoters are not building these systems.
"if one looks into the details and think critically about what is being claimed, one can see that consumers are jumping to false conclusions."
well then help us out here: can you be specific? To me it sounds a lot like goalpost moving. You're telling me that in 2020 if I showed you a system that can solve an Erdos problem or disprove a conjecture (just recently showed up) you wouldn't be blown away?
> That said, it is very cool how an LLM helped human mathematicians in the recent specific Erdos problem solution announced by OpenAI. Just don't jump to the conclusion that anybody can input any Erdos problem into an LLM and a solution will come out the other end.
Woah woah, that's not the conclusion I'm jumping to. That's not at all how these headlines happen. Solving problems like this is almost prohibitively expensive today, and they more often than not lead nowhere. The point I'm making is, today, 4 years since ChatGPT, we have systems that can and have solved them. First we had things like AIME and IMO benchmarks, then people said "well those are just cheats in the training data, wait for it to solve a real math problem" -- ok but now we're solving real math problems.
In "Remarks on the disproof of the unit distance conjecture" (https://arxiv.org/abs/2605.20695) I think Melanie Matchett Wood's remark is the most informative: "It is easy to jump to hasty conclusions, but what we can learn about humans, AI, and mathematics from this development is somewhat subtle. I believe if the level and type of human expertise that is represented on this note had been assembled to find a counterexample to this conjecture a month ago, and those people put in similar amounts of time working on it than they did to reading and thinking about ChatGPT’s solution, the mathematicians would have found a counterexample. However, without the claimed proof by ChatGPT, there is no particular reason anyone would have tried to look for a counterexample, assembled a group of experts with the appropriate expertise, or that the experts would have agreed to turn their attention to this problem."
Some readers might find some of the other remarks more appealing or more informative. I encourage folks to read these remarks rather than the OpenAI marketing video and spin.
> To me it sounds a lot like goalpost moving. You're telling me that in 2020 if I showed you a system that can solve an Erdos problem or disprove a conjecture (just recently showed up) you wouldn't be blown away?
I'm not sure what goalpost you are talking about. Regarding 2020: it depends on the framing, how much I know about the conjecture, the details of the computer system. I an easily imagine not being blown away. But I don't really see the relevance of our emotional reactions to computers doing new things we've never seen computers do before. If the goalpost is being "blown away" by what computers can do, then that happened I think around 1990 when I heard a computer program generate a vaguely human sounding voice. In math, I think it happened when I saw Mathematica simplified a huge nasty complicated algebra expression around 2000. I've been "blown away" by new things computers can do many times over the past 36 years.
> Woah woah, that's not the conclusion I'm jumping to. ... ok but now we're solving real math problems.
Sounds like we are in agreement then that (1) LLMs can not solve any given Erdos problem and (2) computers are solving more real math problems than they were before.
I honestly do think what the OpenAI group did with an LLM recently is a new milestone worthy of attention if one is interested in computer assisted mathematics. I don't mean to diminish the LLM feat. I just mean to throw shade on the corporate marketing, language, slick video, and spin.
I'd review the code written by both.
The best use case i've seen for AI is people generating random one shot projects for themselves, which is honestly so cool. You can make some basic app that does something very specific, that would have taken objectively a lot longer to make by hand. This is when 'crappy' software is more than good enough for a specific problem
And I say this as somebody who generally has a low opinion of AI generated code and rarely uses it. But it has its place.
I think this is a wonderful use case for LLMs. Who knows if regular people will try it out. We've spent decades making people feel like software is some special techy thing that regular mortals should stay away from. LLMs make it easy for anyone to turn an idea into a prototype.
I genuinely don't know if that's true and I doubt you do, either. It's all feels right now.
What I do know is I run a couple of personal projects and I can say they are of objectively higher quality now that I'm using AI to build out proper CI pipelines, expand test coverage, produce higher quality architectures, etc.
Why?
Because in the past I didn't have the capacity to invest in that kind of hardening, but with AI, now I do.
Of course you'll probably make the claim that my code is probably crap, the tests suck, etc, because you've already made up your mind. But having been in the industry for 25 years, I can tell you definitively that you'd be wrong about that.
Now, what'll happen to the median codebase? God only knows. Maybe I'm especially diligent.
But given we're really only 6-12 months into the agentic coding era, I think the only conclusion you can make is that the jury is still out.
I absolutely have deeply mixed feelings about these tools, the ethics associated with them, the impact on the industry, on the talent pipeline, etc.
But I also can't deny that they are incredibly powerful tools that are here to stay in one form or another.
And I say that as someone who, a year ago, was absolutely convinced that they were incremental at best and scoffed at everyone who said something like "yeah but they're so much better now!" or "they're only going to get better!"
Well, they were right, they did, and the world has changed. AI generated code is landing in the Linux kernel. 250+ security holes were found and fixed in Firefox. The impact is here and now, and it's mixed and ugly and complicated.
The amount of slop produced even in company setting is staggering and I don't like it one bit that neither the submitter nor the reviewer of the PR paid due dilligence. And I am only complaining because it then becomes my problem. So, then I have to start nagging people to clean that up. I can say with 100% certainty that the problems I face now would not have happened without LLMs.
That said, used with care, with proper supervision, with dilligence to review what LLMs did, I still think they can be and are beneficial.
I think that we are just not used to getting results of questionable quality from the tools we use. So, I am hopeful that we will learn and it will improve with time but still find myself dreading the age of the vibe coder.
I also think reading and reviewing code is a skill that connected to but very much independent of the writing of code, and the use of coding agents requires us to be far more skilled and diligent at it.
So put another way, people who were good at coding without agents may in fact be a poor fit with them, which means the entire industry is experiencing a dislocation between skills we have and skills we need, leading to extremely bimodal outcomes.
In fact, from my personal experience, going from junior to mid to senior, that was the hardest thing. Reading the code and thinking if what they did was really correct and will not have additional undesired side-effects was hard to become efficient at (it didn't help that we were working in C back then).
So, really, I think that for juniors it's actually much harder because if they want to do due dilligence they have to do the same evaluation but without the years of experience working with that code base. I can understand, even if I don't like it, that they just submit the output of the LLM for the senior to review.
Yeah, but the frequency, volume, and complexity of that activity, and its ratio versus all the other work that a developer was previously expected to do, has shifted dramatically, not least because now we're having to review the output of our own coding agents as well as that of other developers on our teams.
As a consequence, folks who were marginal but capable at that skill now likely find themselves working beyond their ability.
> So, really, I think that for juniors it's actually much harder because if they want to do due dilligence they have to do the same evaluation but without the years of experience working with that code base. I can understand, even if I don't like it, that they just submit the output of the LLM for the senior to review.
Yup, couldn't agree more.
But I agree fully with your last paragraph, and said something similar in a comment elsewhere where I stated my tangible bar as being a Ladybird like browser built from scratch achieving Chrome parity in six months while doing continuous stable releases with coding agents in tow. Otherwise, as you said, the jury is still out.
Yes! It absolutely was (and in many cases still is)! Both projects I've originated and projects I've inherited. I'm not ashamed to admit that. I build in my spare time to create things people (in particular myself) want to use, not to construct ivory towers of architectural purity.
Hell, I inherited maintainership of one gem that barely had a functioning test suite at all and is now at north of 85% coverage and is something I can now change with far greater confidence, and I recently forked a repo to work on another project that was a damned disaster that I massively refactored to make it clean and maintainable.
The world is absolutely full of janky side projects. Is that surprising to you? They're side projects, ffs, not five-nines planetary scale platforms.
I couldn't agree more. If you're using AI tools to produce worse software, faster, you should rethink how you are using them.
If we're not delivering better software with this stuff then what are we using it for?
Moreover, in my experience helping businesses on this topic; They never defined or made measurable what quality meant in the first place. Then when they finally do figure it out, it turns out that the average repository is a total disappointment in terms of absolute quality.
Also, while some code review comments are just plain wrong, LLMs did produce some damn good comments, on the same level as a different senior engineer might note had they taken the time to study the code carefully.
Its easy to lean too hard into vibe coding. I've spent the last week unpicking a bunch of poor decisions claude made while coding something up. Its my mistake - I trusted it too much. It made a lot of sloppy, poorly thought through abstractions and then built a bunch of bad code on top of them. I should have been more careful reviewing its abstractions first.
But as others in this thread have said, LLMs are also great at doing all the tedious quality-improving tasks that I sometimes don't have time to do. You can prompt LLMs to write tests, to do fuzz testing and to set up CI pipelines. You can get them to formalise the constraints in the system (and then try to find violations of those constraints). I'm doing some work at the moment with Cocoa (apple's old UI library). I downloaded all the cocoa docs I could find. Then asked claude to read it all and review our code. It found lots of places where we're not using the API effectively. Fabulous.
LLMs are an accelerator. If you already know how to write performant, reliable software, LLMs can help you get there faster. But if you sit back and let the LLM guide itself, who knows where you'll end up. Probably nowhere good. Over the next few years we're going to see every possible use of LLMs. Probably more vibe coding disasters than anything else. But as a senior engineer, I think its way too reductive to write the tools off entirely. They're useful. But using them well? That's the trick.
As for tests, I've seen LLMs produce good ones as well as useless ones. I guess it's all about instructions... sorry, prompts... no, sorry, prompt engineering to get it right and done properly.
That said, I am also very concerned with brain rot. Engineers nowadays can commit code they don't understand without a blink of the eye. Slowly, the knowledge may get sparse if we are not careful about it.
I wonder if we reverse that. When engineers write shitty, slow, buggy code or when they're forced to do that by the company, they are effectively killing people.
After all, if I spend an hour dealing with a preventable bug, that's an hour of my live I am never getting back. Multiple that by the userbase and you get entire lifetimes.
Structured programming is a blight Compilers are a blight Object oriented programming is a blight Code generation is a blight Agentic engineering is a blight
All of these blights have one thing in common, they are tools that the lazy person can use as a crutch to put out passible but problematic code. Laziness is a choice, and choices are made by humans with agency and free will.
For me, AI is the first time I have ever been able to get something resembling an opinion on specific problems/situations that I encounter. I can ask it a very specific question about what the best approach is for what I am working on and it can give me an answer that I read over and consider before deciding on what approach to take. I still frequently get answers that are nonsense, but even then it helps me think deeper on how I should approach the problem because I can ask myself "Are the statements made by the AI true?".
It’s an absolutely fantastic educational tool if you choose to use it that way.
... Put down the duckie[0]?
Unfortunately Twitter pre-acquisition was probably the focal point and since then, I don't think the community has been the same.
I can't wait for the day until all these people collectively snap out of it, and go "what the hell were we thinking with these chatbots".
It is just a tool and a useful one. Your loss for not making use of it, but .. you seem to live on a different timeline. I am reminded of old Japan, where the Samurais banned gun powder, so they could continue to fight with swords and not change their ways of living.
Have it do research (HN, Stack Overflow, technical blogs) and summarize.
Why do you think this wouldn't work?
You'll still need to be capable of understanding and judging the output, and it might miss some options, but it is certainly no worse than pre-AI.
There's a time when the operation of the machine becomes so odious Makes you so sick at heart that you can't take part, you can't even passively take part And you've got to put your bodies upon the gears and upon the wheels Upon the levers, upon all the apparatus and you've got to make it stop And you've got to indicate to the people who run it, to the people who own it That unless you're free, the machine will be prevented from working at all
Even then we have machines doing it all and yet we all function well. I think eventually this would be a tool usage which will take human intelligence to another pinnacle.
Machinists and lathe operators became CNC operators, they didn't lose their jobs, just that instead of turning the inputs by hand, they punch numbers in a machine, but the advent of CNC didn't mean anyone off the street can now punch numbers in the machine and replace the machinists since you still need the years of training and experience.
SW devs will be the new CNC operators, about knowing what data to input and how to wrangle the slop machine to get the desired output faster and better than your competition.
For now. Will that be true in 12 months? 4 years?
If you're a programmer, your skills have been devalued significantly in the last 12 months. What makes you think the remaining value you offer will be required 12 months from now?
I don't have a crystal ball. In 4 years whoever is US president might start another war and fuck the whole planet back to the stone age. Nobody will know what will happen in 4 years so why worry about it?
I don't have any - thinking is the only non-cattle job left for humans to do; if we outsource thinking for all of humanity, we're an evolutionary dead-end.
No it isn't. I still do the thinking on how to solve my problems, I only outsource the tedious part, which is typing the code and fixing the syntax errors till it all compiles and does what I want.
If you also outsource thinking to it, that's your choice though. Or the company's choice. But ultimately the free market will deiced with products made using LLMs outcompete those made without.
I've said this elsewhere before but I single-handedly produced more actual tangible business value with Microsoft Access than anything else since. What was an hour's work is now a procurement process and thousands of lines of tedious configuration and boilerplate that involves pipelines and tens of services all coordinated and hosted by someone who has created a moat to extract money out of me.
All I want is a fucking report.
The LLM makes us blind to the gigantic fucking shit show we built.
Do you really though?
Here's a question: how many times do you visit claude.ai or open Claude Code (or whatever harness you use) (or use whatever model you prefer) to help you solve a problem, ask it a question, etc.?
One thing I've noticed, which seems to go completely unmentioned, is how these LLM tools are like drugs. It can pump out thousands of lines of code ---> it can write my entire program for me ---> I've written quite a few programs with it and I don't write a single line anymore ---> I go to it for what would normally be things I could do on my own. The problem is that this isn't some immediate thing: like a drug, it sneaks up on you and you don't even realize it until it's far, far too late. I've been programming since I was 13 and I've just now started to notice the deskilling that's been happening to me. I've just now started noticing how often I'll visit Claude Web for something I should be able to do on my own. Nobody really seems to mention this, or it's repharsed as a good thing. And I don't get it: how is undergoing cognitive surrender a "good" thing by any metric other than the metric a beancounter would use? What worries me is that I fear this is happening to way, way more people than those who actually bring it up, potentially yourself included, and you just haven't yet realized it because you haven't really thought about it. That is ultimately what this "AI revolution" is going to bring. It's what the billionaires want, and what they want is usually what they get because the systems we've built are set up to not constrain them.
Congrats, you're in a small minority of people.
>What worries me is that I fear this is happening to way, way more people than those who actually bring it up, potentially yourself included, and you just haven't yet realized it because you haven't really thought about it.
People probably said the same thing about using pocket calculators when they started getting cheap and mainstream in the 1970s.
Do you still use your brain to calculate sales tax or the square root of pi just to keep your brain sharp on knowing calculus, or do you always use a calculator because it's a mainstream commodity that will never go away?
> Congrats, you're in a small minority of people.
Why did you respond to this? I fail to see how it relates to your point whatsoever.
Because it's this thing us humans call context. Or background. There are a number of terms for it. I was providing that so the rest of my post would make a bit more sense.
Except a pocket calculator is very different from an LLM. It isn't a tool you can go to and ask it literally anything under the sun and get an answer (even if incorrect). It's cold, logical and uninterested in your life or what happens. These LLMs are trained to be interested and empathetic. They are, IMO, way more like drugs than most other things.
The entire universe is cold, logical and uninterested in your life or what happens to you. But that's exactly what I want from a machine that's supposed to be my personal slave. I don't want a lecture from it on current day political correctness and sensibilities, I just want it to do exactly what I ask it no matter how rude or machiavellian it may be. Whoever will deliver this type of unhinged obedient slave to the masses(most likely China), will win the AI race.
> These LLMs are trained to be interested and empathetic.
Luckily there's already like a dozen LLMs at this point, many free and open source, especially from China, and I and many others will just pick the one that isn't trained to be interested and empathetic, but the one that's my personal slave and does exactly what I want from it, because of the competition between them will allow this freedom of choice.
In 5-15 years when semiconductor shortages ease up and hardware advances a bit more, there's a chance offline open LLMs in your pocket will be as obliquus as pocket calculators were in the 80s.
>They are, IMO, way more like drugs than most other things.
So was music, TV, video games, porn, etc. when they became mainstream. We adapted and survived.
I'm more scared about the state of world peace, economy, CoL, jobs and housing market than of LLMs being addicted.
There's millions of people on the planet who are a couple of meals away from hunger and rioting and migrating to other countries, and a climate catastrophe or one more wars could destabilize the whole planet for good, than LLMs being addictive. If LLMs are your biggest concern, you're in a very privileged position.
AI is automated irresponsibility, and it is nothing like any earlier transition.
When a technology trend means people literally won’t be able to tell if you are living or dead, and also stop caring about the difference— that’s unprecedented in the history of humans.
How can people say false things like this with a straight face?
Thanks to industrialisation, automation and mass production, the poor of today have access to things that even kings from hundreds of years ago couldn't even fathom, let alone poor people back then: abundant cheap food that poor people can now be fat instead of starve to death, cars, planes, MRI machines, helicopter ambulances, vaccines, personal heating and air conditioning, OZEMPIC, etc
Kings back then would eat hard bread, shit down a vertical shaft that emitted the scent through the whole castle, and their sleeping chambers had ice on the walls in winter and lice in the clothes and bet sheets, plus they had parasites in their gut and any small disease could kill you.
Meanwhile the cool homeless guy outside my building has 3 hot meals a day and a daily shower in the homeless shelter nearby, warm clean sleeping bag for winter, shades for summer, a bicycle for moving around town, a smartphone which he uses to watch youtube all day in his sleeping bag, plus access to medical care that kings of kings never had. All this with no job, and no care in the world.
Have you ever been without a job and/or homeless to say shit like this?
Also, how nice of you to ignore my entire argument on how the poor today are NOT as poor as they were hundreds of years ago, and instead sidetrack the conversation one offtopic tangent for a cheap jab in the name of scoring some emotional virtue signaling brownie points.
Thanks to colonialism, also in more modern form called globalism.
> Kings back then would eat hard bread, shit down a vertical shaft that emitted the scent through the whole castle, and their sleeping chambers had ice on the walls in winter and lice in the clothes and bet sheets, plus they had parasites in their gut and any small disease could kill you.
Wealth has no intrinsic value, only relative one. You are only wealthy relatively to other members of the society. Doesn't matter if pharaohs had less comfortable lives than me. What matters is that a gap between pharaoh and a worker working on pyramids was way smaller than between Jeff Bezos and person working at the Amazon distribution center.
Also, most sweet fruits of progress that its prophets like to enumerate are not direct consequences of technological changes, but they came only after political struggle that has arisen exactly because the direct consequences were very dire for most people. If people pushing today for AI could decide on these things, they would be very happy to take away these hot meals from homeless people and let them starve to death.
You don't have to participate in globalism, you can be a hermit state that doesn't trade with the evil imperial capitalists, like Cuba and North Korea. Everyone wants to live there because the QoL is so good.
QoL is so good here precisely because it is so poor out there.
If those "around the world places" had all the valuable resources that enabled modern civilisation when the west came and took them, why weren't they the ones to first make use of them to build computers, vaccines, MR machines and moon rockets, or at least clean drinking water and indoor sanitation facilities for their people?
Or why didn't those "around the world places" then go and extract wealth and materials from the west instead? What stopped them? Surely it wasn't their moral values, since war, slavery and genocide even cannibalism, of their neighboring tribes and nations, was the norm of the day over there.
>and still extracts through globalisation
Correction: "pays for it through globalisation". That's how countries like India and China got so many people out of poverty after globalisation.
For example, I want to make:
- A mini OS on top of SeL4
- A UI framework based on SolidJS, for native apps, in rust.
- My own photo manager (which can do backups & sync across all my devices). And a gallery to share photos with friends
- A local first data store, built on top of CRDTs
- My own programming language
And lots more.
Each of these projects on their own would take months of time. If LLMs can speed up development, that's great! I don't care if nobody else uses what I make. I want a personal computer full of my own software.
For example, the photo backup system I want to make will let me manage my ~400gb photo library. I want my library backed up on a couple devices, running linux and freebsd. I want my mac and iphone to have a local mirror of all the favorited photos, and when I'm at home, I want to be able to browse all photos from those devices by streaming them over the local network. I want native macos & ios app interfaces to view and manage all that.
I don't know any existing software that meets my requirements. I don't think any such software exists. Apple, Dropbox and Google will solve this problem for me if I store all my photos in their cloud and pay them an ongoing subscription for the privilege. I'd much rather make something myself, and back up my photos on my own hard drives.
Making something like this is simple enough, but very time consuming. If claude can take the drudgery out of it, well, I think that's just delightful.
You can accomplish most of that by installing Syncthing.
But the objection is that you’re not really building anything new even if you think it’s a new idea. By your definition you’re building for yourself and not sharing…so what good are your little projects. Reading your original list it just seems like you want to build and run software without having to do any research, even if a solution already exists.
At some point developers seem to forget that making stuff is fun. The fact SQLite and GCC exist doesn't mean you're banned from making a database or a C compiler.
You're allowed to make stuff just because.
What an incredibly blinkered view. You think there's no value at all in making software unless your program solves a truly novel problem, and does so for millions of users?
There's so much joy that can be found in simple, disposable acts of creativity. I don't need the world to applaud when I cook an omelette, make a table out of wood or play a piece on the piano. Why would I need that from the software I make? Just because you can buy a table from IKEA doesn't mean its not fun to make one yourself. Just because someone on spotify played it better doesn't mean there's no joy in playing music.
Why would I hesitate to make my ideal photo library program, just because other programs exist? What a sad, self-limiting rule.
Does that make you mad? Why does other people enjoying themselves with LLMs make you mad?
AI is seen as a way out of drudgery but you're just trading one problem for another.
It's actually quite similar to buying the services of a programmer off Upwork to build something for me, only with LLMs it's way cheaper and faster, with a shorter feedback loop.
If I spend a weekend standing up a self-hosted media system or something, I doubt anyone would dispute that's a fun building exercise.
If I do the same thing but use an LLM to build out instead, somehow it's not.
Yeah it's not the same kind of building as what we might have done pre-LLM but it need not be any less satisfying or rewarding.
The real disconnect seems to be the classic dichotomy: people who see coding as the point and the purpose, vs people who just want an outcome.
And that's fine!
I'm just don't understand why the one camp feels the need to deride the other.
I don't enjoy CSS, so I delegate that to an LLM. Just like I don't like writing assembly by hand, so I delegate that to a compiler. So what?
Take woodworking for example. When I build a kitchen cabinet, I can get lumber that's already smooth and treated, I can buy drawer tracks, I can use power tools instead of a handsaw and a screwdriver, I can use a pocket hole jig to make joints easier. I still have to do more planning and assembling than with the Ikea cabinet, which also takes more work than having a contractor do everything for me.
I'm doing it my way because it's fun for me. Other people might enjoy other parts of the process - or different things altogether.
There's a whole spectrum between doing everything from scratch and paying someone to have it done for you.
What do you think architects do? Or interior designers? Or civil engineers?
Architecture certainly isn't building, and neither is interior design. Civil engineers calculate and specify the loads in excruciating detail, because if they didn't, people would die.
The biggest value proposition of an LLM is being able to focus on the truly high-value activities while allowing the machine take care of much of that reification.
That you think architecture or interior design isn't building tells you prefer to downplay or devalue any work that isn't hands on construction. It's an interesting perspective, but it's one I'll never be able to understand or agree with.
For someone who's been around the block a few times, there's little new under the sun. Coming up with good architecture for a problem space and having the LLM fill in the details is incredibly effective as a pattern.
That said, tbh I'm not sure why we keep exchanging messages. Your other comment about coding with AI assistance as being akin to buying a statue tells me you have either never coded with an agent, and so think it's a simple matter of saying "build me a to-do app" and then calling it a day, or you're experienced with these tools and deliberately misrepresenting the process.
If it's the former, then we don't share enough understanding to have a healthy conversation, and if it's the latter then there's little point in arguing with someone who insists in doing so in bad faith.
- I want native desktop apps
- I want P2P device sync. My laptop and my home server should be peers. Just with different rules for which photos are stored locally on each device.
- I want per-device storage rules. (Eg, "I want all my favorited photos, and recent photos up to 50gb on my laptop").
- Backups. I want to be able to plug in an external hard drive and backup my photos (with some rules). Then disconnect the drive and see in the metadata for those photos that I have a copy of the full res RAW on that drive.
But even if other programs exist, its still fun to make something myself. Authoring your own software gives you a different relationship with the computer. You're less of a consumer. You can change or tweak features on a whim. Psychologically, its kind of like being in your own home vs being in a hotel room. If you hate the furniture at home, you can move it or change it. You can decorate however you like. In a hotel, you're affected by your environment but you have no agency over it. I think its much more healthy to create.
But I’m optimising for the feeling of joy and agency over my tools. I also want to experiment with p2p / local first data structures in a real app. I could fork immich and rewrite the data model. But that all seems much less fun.
At the end of the day, I’m not doing this for other people. I’m doing this for me. Learning dart so I can hack on someone else’s code sounds like a pain in the butt. Not like something I’d want to do for fun on the weekend.
Working on someone else’s code seems like a strange default. You and others have jumped on me for wanting to make my own thing instead of using - or forking - someone else’s software. Can you help me understand why? Why would that be the default for a fun hobby project? I’m confused.
Back to 1988.
Think I have a fair bit of experience.
Is it not engineering anymore even if you micromanage and relegate the machine to a better typist, following patterns and doing research around?
Oh get off your high horse.
You may as well say "If the compiler compiles your code, you achieved nothing". Its the same argument.
I love meaty algorithmic challenges. And I enjoy the design challenge of making a piece of software. But I don't write CSS for fun. Debugging styling issues on mobile? I hate it.
LLMs let me work on the parts that I find enjoyable and satisfying. Like UI design, or figuring out the data model. And then I can delegate the parts I don't like to Claude. It seems very happy to bang away at CSS.
That's a great deal. If you want to personally write every line of code, good for you. But don't police my leisure projects.
There are a ton of opportunities available right now to make new things. And make them better, more customizable, and more sovereign.
To the replies: be the change you want to see in the world, guys. That may be trite but focusing only on the negative will just make your own life shitty.
Is that the same as buying a computer program from anthropic?
I have a friend who's an architect. Sometimes we're in the city and he says "Oh see that building? Its one of mine." He didn't lay the concrete, or do the wiring or anything. Should he not be proud, because a construction company built each floor? Does it not count? Does his contribution not matter?
Since you said you want to make those things that you list, I assume none of these things have been built yet. If so, I would encourage you to consider how excited you will be to constantly maintain those things you build. But even if the maintainence cycle won't be as exciting, since you are the sole user you have the advantage of being able to proceed at a leisurely pace even while doing maintainence work.
In a professional setting, the dopamine hit of being able to build something quickly that works in an area that you have little to no knowledge in makes you more dependent on the AI in the maintaince cycle as you want to chase that dopamine high by maintaining the same development speed. This in turn leads to a bigger burnout crash after that peak dopamine hit. Maintainence is a phase of diminishing returns even without AI, but when your coding agents are introducing new bugs at record pace with their bugfixes with no new features to write home about you are in a special place in Hell.
I'm all for using AI to build ambitious projects. I have yet to see a person/company/organization continuously release huge software endeavours in a stable professional manner day in and day out with a coding agent harem in tow.
If something like the Ladybird browser, or any browser that is "built by scratch", achieved Chrome parity in six months and consistently maintained the same level of stability with continuous releases then I would see that as proof that this approach has become professionaly sustainable.
The reason people are getting away with so much using AI is because of the open secret in most enterprise engineering practices: the customer cares more about the response time for fixes they report than they do about overall or longterm product quality.
Why should I consider that?
Its funny how the default with programming is that the piece of software exists forever. I've been learning to play the piano lately. The default with piano is that every piece is ephemeral. If I don't go out of my way to record something I play, after the notes have run out, the piece is gone forever. The same is true of cooking - except you can't record a meal at all. Once you eat it, its gone. Lots of art forms are like this - theatre. Dance. The circus. They're no less beautiful for being ephemeral.
Why do we assume software has to be maintained indefinitely? Why even think about that right now? Maybe I'll work on these projects for awhile, maintain them as long as I want, and then in a few years someone will make something way better and I'll use that instead? Would that make the effort I put in pointless? I don't think so. I think it would make programming more like playing the piano. How lovely.
> I'm all for using AI to build ambitious projects. I have yet to see a person/company/organization continuously release huge software endeavours in a stable professional manner day in and day out with a coding agent harem in tow.
Yes, I've burned through enough claude tokens now that I find myself agreeing with you. I wouldn't use an LLM to make and maintain google chrome. But I wasn't planning on doing that anyway. There are also a lot more options than (1) write everything yourself and (2) vibe code the whole thing.
LLMs are good at small-to-medium scope tasks right now. Fine. I'll use them - or not use them - with their limitations in mind.
By all means don't consider it if you don't plan on using them for a considerable amount of time, but there's a lot of of distance between a decent amount of time and "forever". You listed a mini OS and a UI toolkit among your projects, I hope you can forgive me for assuming you were planning to use those things to build more things, which would in turn often entail improving and maintaining these building blocks while they are actively used.
Adding the features I have a need for over time is the fun part as far as I’m concerned.
I actually want less software for myself. Less things to maintain. I've become a "digital minimalist" in that I use very few software, only ones maintained by others who can afford and are willing to keep them working.
Some people find joy in home cooked meals. To others, wealth is never needing to cook again.
To each their own.
As for maintenance, I think LLMs are actually very well suited for small maintenance tasks. The expensive part of programming is loading and unloading context into my head. "Oh dear, what is all this code doing? Oh no, coffeescript? And some ancient version of express JS? How does any of this work?". LLMs can bomb in, fix a compilation error and bomb out. "This program isn't compiling any more on macos X.Y.Z. Read the patch notes. Figure out why and fix the problem." Perfect.
Think about it: everyone has characterized AI slop, as slop. Which means that we negatively value it in terms of originality. Combine that with the fact that there will be a lot of it, this means that original work will 1. stand out or be very distinct from slop, and 2. have its value amplified as a result of this polarization.
basically, we value originality more AND are able to identify it more readily.
related is also the fact that originality will literally be valuable as training data for future models
Also there have been dozens of HN submissions and comments where the poster didn't even bother to remove the em dashes. Most people just don't care. The people who continue to post like this wouldn't have been as visible had they not discovered AI and pounced on it, but they were always there. The idea of posting with an AI voice, em-dashes and all, would likely have still appealed to them if you'd asked 5-10 years ago. Nowadays it takes hardly any energy for them to have a persistent voice.
Slop has always been around. AI has cheapened its creation.
If something is able to generate things with soul and true originality... we're talking about something incredible, a new intelligent species potentially
think about how in music, when an artist comes out with something original and awesome, and then everyone starts copying it and creating their own derivative works, like Jimi Hendrix or something.
Did Hendrix become useless? Did everyone end up thinking he sucks or something? No, he is even more revered, as the originator of a new type of sound that probably created multiple genres
The same thing applies here. Originality will be valued and even empowered as extrapolation and development off of it can increase in speed and quality in the case you mention
True, but some nuance is that a LOT of artist/creative types lean exclusively on the mechanical skill needed to create, without anything really much to say. They also very frequently copy other's styles, etc.
I'm not defending AI pumping out crap, but this also shows a lot of folks don't have much to offer beyond the mechanical aspects and we shouldn't glorify churning out stuff by hand as high art either.
How? It's undermining what the human intellicence is made from, learning.
I see no indication that current human intelligence is at anything close to a historical pinnacle. Human knowledge, yes, but intelligence? No. Collectively, we're dumb and trending dumber, and the tendency towards lazy thoughtlessness which AI engenders will accelerate that trend.
> Human knowledge, yes, but intelligence?
of the post you responded to?
Intelligence is the same but for mental faculties. A human is intelligent when they develop their critical thinking, memory, focus, logical reasoning, etc. A human is unintelligent when they fail to develop these things to their personal potential. And when I look around me I see a culture of inustrial-strength distraction that has robbed people of their ability to focus, I see encyclopedias in everyone's pockets that have robbed them of any incentive to remember, I see a society of comfortable complacency that has shielded them from any consequence of poor logical reasoning, and with LLMs I see a mass surrender of the need to exercise critical thinking in exchange for the warm embrace of thoughtlessness.
There's no reason that things need to be this way. The human hardware hasn't fundamentally changed in 100,000 years, and we have so many more resources today that it's easy to imagine that we could all be, collectively, more intelligent than ever if we could somehow inspire people to care. Sadly, we don't seem to be able to.
> I see no indication that current human intelligence is at anything close to a historical pinnacle. Human knowledge, yes, but intelligence? No. Collectively, we're dumb and trending dumber
Also, I think we (non-scientists) might be overestimating the average historical intelligence - see Flynn Effect [1] - perhaps because of a bias in our perception of the past levels based on who published books and thoughts - basically more intelligent members of our species.
> and the tendency towards lazy thoughtlessness which AI engenders
----------------
[1] https://en.wikipedia.org/wiki/Flynn_effect
[2] Thucydides, History of the Peloponnesian War, 1.20 — 5th c. BCE Greece. “So little pains do the vulgar take in the investigation of truth, accepting readily the first story that comes to hand.”
[3] Aristotle, Rhetoric, 1.2 — 4th c. BCE Greece. https://kairos.technorhetoric.net/stasis/2017/honeycutt/aris.... Aristotle treats public persuasion as necessary partly because ordinary audiences cannot easily follow complex chains of reasoning. He says rhetoric addresses deliberative matters before people “who cannot take in at a glance a complicated argument, or follow a long chain of reasoning.”
[4] Plato, Republic, Book V — 4th c. BCE Greece. https://topostext.org/work/768. Plato distinguishes philosophers from the many “lovers of sights and sounds,” who enjoy appearances but do not apprehend deeper truth. The text says their thought is “incapable” of grasping the underlying form or nature of beauty, and that few attain that deeper vision.
[5] Confucius, Analects, Book 2 — 5th c. BCE China. https://www.chinastory.cn/ywdbk/english/v1/detail/20190722/1.... "Learning without thought is pointless, Thought without learning is dangerous".
[6] Buddhist tradition, Dhammapada, Appamāda-vagga. https://suttacentral.net/dhp21-32/en/sujato. Dhammapada contrasts heedfulness with heedlessness, treating heedlessness as a central human failing. In one translation: “Heedfulness is the state free of death; heedlessness is the state of death. The heedful do not die, while the heedless are like the dead.”
[7] Francis Bacon, Novum Organum, “Idols of the Mind” — 1620. https://history.hanover.edu/texts/bacon/novorg.html. Bacon argues that the “Idols of the Tribe” are rooted in human nature itself, and that human understanding distorts reality like a false mirror.
Excellent question. Global GDP estimates going back to 1 CE are here [1][2]. I would argue that GDP is a good proxy for an estimate of the summation of per-capita intelligence being used for "productive" ends, by definition.
But, the global average GDP per capita is [3] with a similar hockey-stick curve, and it is unclear whether the per-capita intelligence used for "productive" ends has also been increasing similarly, because the confounding factor is the effect of tools as force-multipliers (impact on productivity). The Flynn Effect is the strongest indicator that yes, average intelligence has been rising as measured in certain populations where cultural differences do not wreck the applicability of IQ tests.
[1] https://ourworldindata.org/grapher/global-gdp-over-the-long-...
[2] https://web.archive.org/web/20260525015042/https://ourworldi...
[3] https://ourworldindata.org/grapher/global-average-gdp-per-ca...
Like renewables were on track to bring us peak oil, maybe AI will bring us peak human intelligence.
While all those are true they are not reflecting the level of intelligence of people: intelligent people take personal stupid decisions because while intelligence is a function of let's say the more "abstract brain", decisions are emotionally driven and influenced by the "ancient, threat focused, pleasure driven brain".
Here is a quick way to think about this: some intelligent people are obease, some others don't exercise, and others don't take their health seriously while also working on the most amazing problems we ever solved. You know what's the biggest paradox here: they all have the capacity to understand fully the impact of their lifestyle on their health but still making a life style change is hard due to not being driven by knowledge and logic.
We are at a peak in absolute terms, though the decline is coming quickly:
https://x.com/jonatanpallesen/status/2034755779501105321
But in relative terms intelligence has indeed been declining for a long time:
Also, science used to directly correlate with improvements in life standards. Nowadays we see advancements in science (AI, psychology) used to actively reduce the standard of life.
It really is a remarkable picture, but I'd like to note that it's all physicists, not scientists in general. It was the golden era of physics.
But that shouldn't be any surprise. We're already at more than a hundred years of deliberate dumbing down of the population through schooling and mass media. These effects are exponential through generations.
Mechanical replacement cannot be compared with thought replacement anyway, but the most thoughtless pro-AI comment tends to be at the top.
Introducing a machine to a manufacturing role obviously makes the manufacturer less fit, but it enables society to break through fitness barriers in general
If your point is that it's not orders of magnitude fitter, that's a good one. I don't think people will be much more intelligent in the future than they are today but they'll probably just be more specialized and have deeper knowledge
We also know that testosterone levels (which are connected with muscle mass, recovery etc.) are lower today than even few decades back.
I’m not sure what it means if we want the average person to be fitter, it seems like we’d have to reinstitute manual labour if that was the goal.
Which is maybe a good analogy for not using AI in the classroom and forcing kids to use their minds
The other factor is that we all get significantly less physical activity than before, and obesity is a increasing problem.
And while the first effect is a effect of the machines, it is the latter effect I think most easily maps onto today's situation.
Personally I am quite certain that if you could teleport 100 random 20-year old from 1820 they would be better than 100 random 20-year old today at most physical tests, especially if you gave them food first.
Sure...
https://news.gallup.com/poll/694199/u.s.-depression-rate-rem...
>this would be a tool usage which will take human intelligence to another pinnacle.
Between the endless slop, loneliness and depression epidemics, record low reading comprehension, attention shortage, we're not in any pinnacle today. We're in a regression from a few decades ago, getting worse.
For me, building software has often been a solitary process in which I was far more obsessed with it than those around me. I'm not in a tech-heavy area and I don't have a ton of well-informed people to talk to about programming, software engineering, or AI. I have had experiences like the author in which I needed to learn a new technology or a new language but ended up doing so on my own at home, not with the assistance of a much more knowledgeable developer with significantly more experience.
To me LLMs have left us in a situation where the following things are true and moving forward lies somewhere in figuring out how to reconcile / resolve these things:
- You can use LLMs and learn things or not learn things; this is a result of the approach, desire, and willpower of the user.
- There is a level of skill associated with using LLMs much like nearly everything else in existence. The user's skill level impacts their perception of the technology and also affects the way those around them view the technology. Unskilled users will generate more negative sentiment.
- Some people love to do the things the machine is good at and do not want the machine to do them, while others hate to do the things the machine is good at and want the machine to do them. I realized at some point this year that I don't love programming anywhere near as much as I love building and designing systems and solving problems.
- Software development is many things wrapped up in one and talking about it as a single thing makes it more confusing. Some people like to think through the logic of the application and have an LLM write the code while others want the LLM to think up the solution, implement it, and test it. These are two very different people with likely different goals and different desires.
- When someone else looks at Claude or ChatGPT they might see something completely different than what you see.
I hope some of this resonates with others.
And this has held me up in good stead.
Now with AI, I've found a tool from which I can learn, show me the right way to do things, and explain in detail what has been done. I can ask questions, point out mistakes, go back and forth on different implementations and at the end of it, come out a better programmer.
As many commentators have mentioned, AI means different things to different people. For me, it has been empowering, enlightening, and humbling.
There have always been so many things to learn, but never enough time. Now, it doesn't quite feel that way.
I'm all for open models, open source agents, etc. I don't want to give more power to the big corps, though. Imagine what software engineering could become in 5 years if all thse big corporations gain even more power over us. It's a terrifying scenario (e.g., pay more so we don't show you ads in between claude code prompts; pay more so that the produced code doesn't incrust ads in your app...). Do you really want the same shitty experience we have now in the global internet, but deeply ingrained in your software engineering workflows?
Even in regards to programming competency and the culture of vibecoding, it's similar to the early stages of electric cars where they fill some roles better than ICE engines, but it's still a decade away from doing the same job. And the whole time there was people trashing electric cars as a novelty/impractical/expensive/dangerous/etc because the infrastructure wasn't there yet and the technology was immature.
The only real moat pattern we see is datacenters being hotly in demand but even those will scale up and commoditize, RAM manufacturing will catch up etc.
So wouldn't it be wiser to err on the side of caution?
What's it about is once you remove the paycheck that all proletariats need when things get "interesting".
You get some AI-slop like this:
> Your AI email deliverability specialist that tests and fixes your emails — backed by a Top-Rated expert.
However, missing the joy of the old-school way of growing as a developer is not only the wrong reason, but also very dangerous according to Darwin.
Our customers don't care about how it is made after all, but they do care about long-term support, costs, and predictability, etc.
But I'm not sure whether we can say we made a real net positive progress in the industry. The whole thing is a big mess. In many cases, AI moves us in the same direction in turbo mode, making it not only messier and more expensive but also dangerous.
I tell them, "Leave me alone", as I see this mess as an opportunity if you think the right way, starting from the first principles.
Per the human element, the author is in part relaying about formative experiences from youth that you won't easily repeat, and also experiences that are not decoupled from the work as it still exists, unless you are entirely remote, which is not a LLM-specific problem.
All of which to say, the emotional element behind it is valid, but the diagnosis is off the mark. I think the human element, should it be jeopardized, is in part through the complacent convenience of remote work and disinterest in community participation. But, communities still exist, and tech communities historically were always niche. As it stands they're probably bigger now than they ever were.
There are still new frontiers with software where LLMs will be less effective. Yes, there is less friction than before for learning technologies, but all this does is move the goalpost as we can accomplish more with our time.
Instead of hacking things out through trial and error on mature stacks (with or without others), you'll be closer to the cutting edge and have different problems. Many of which will still be technological in nature.
"If you let AI do your writing, I will come to your house and kill you"
Thank you for posting.
And what "tells" are in this article, anyway? It reads very straightforwardly; it does not, in fact, have any weird metaphors. Em dashes are not actually a tell, but using the wrong characters for em dashes (the article uses " - ", e.g., space hyphen space) seems pretty human. Last but, I would argue not least, I am willing to give the author of an article about not wanting to program with LLMs the benefit of the doubt here. "The orphan-punching machine fills me with existential angst and dread, and the only way for me to communicate this to the audience is to let it punch a few orphans for me."
Now, if you'll excuse me, all this is making me need to be literally the size of a high-proof daiquiri.
> Throughout all of these experiences, and so many more than I have time to share, it was the human connection that made it special. The laughter that helped me get through a hard problem. The sleepless nights that reminded me I was not alone. The selflessness of others to get on stage or behind a camera and teach people they didn’t even know, often for little or no cost, just so that others would be enabled to build and have an influence.
> They don’t laugh with me when my code fails to compile after I swear “this is the one.” They don’t help me develop an understanding of my software, so that when someone says “how does this work” I can pour my heart out with passionate explanations. Most importantly, they don’t turn their head and smile and participate in the inexplicable elation of saying “we built this!”
> I desire to connect with people. I long for the days where I was vulnerable and shared my struggles with engineers who charitably stepped up to support me. I miss taking what I learned from those struggles and sharing them back out as a blog post or presentation, encouraging the next person to overcome the same challenge.
>>It (software development) is sincerely an art form; A craft that takes >>dedication, perseverance and especially, a strong community to endure. >>Software was built by humans, for humans.
100 years ago, you could only buy furniture made by crafts people. Real artisans. Now you have a choice: IKEA or hand made Most people choose IKEA because they don't care how it was made, as long as it does the job There are still those who prefer hand made furniture, and they pay a pretty penny for it. I think that's where this is heading, and I agree it's unfortunate. Software development will become a hobby (many people do wood working in their spare time). There will be a few real experts left, who largely do consulting. Maybe they create training data? Maybe they design frameworks for the AI to master. I don't know. But things sure are going to be different from here on out, and not entirely for the better.
>>If it’s not built by humans, then who is it being built for?
Right now it's built by AI for humans (and sometimes other AI). Very soon it'll be built by AI for other AI (and sometimes humans). Later it'll be built by AI primarily for AI (rarely humans).
Perhaps check in your country if that isn’t true as well, people automatically assume IKEA killed all local shops, but there are many out there if you search for them.
The public debate around AI makes me think of the cognitive distortions discussed in cognitive behavioral therapy. In particular, all-or-nothing thinking and catastrophizing. These are often symptoms of anxiety or psychosis. I sometimes wonder whether entire societies can experience these symptoms.
https://en.wikipedia.org/wiki/Splitting_(psychology)
https://en.wikipedia.org/wiki/Cognitive_distortion#Decatastr...
Those are the minority. The majority will find the middle ground.
I agree, especially with my own projects. Except you don't always get to choose in the workplace. Now that teams are being measured by PR throughput and token usage, you will look "worse" next to the person who is totally vibecoding. My fear is that if I don't vibecode, I'll be passed up for promotion.
The indicators that vibecoding may be bad, are lagging. What I mean is that issues that come up from vibecoding whether its performance, service degradation, massive data migrations etc. will always show up later.
Long before vibe coding, it was well understood that output in terms of lines of code wasn't a good metric for quality or productivity. I'm predicting that the industry will come to the same understanding in terms of PRs and token usage. Eventually, those lagging indicators will catch up and teams will have to relearn the "No Silver Bullet" lesson.
The "powers that be" would prefer if you sideline yourself. Instead, pop a bird and say "thank you kind sir, but no."
https://news.ycombinator.com/item?id=47591796
More specifically to the submission, I’ll say I agree with the author. This “being left behind” fear mongering is an exhausting uncritical talking point. Life isn’t about rushing through the end and killing yourself to be “productive”. “Being left behind” is only bad if what’s “ahead” is an improvement to your situation, and that’s not a given. Humans aren’t built to be pushed to 11 without rest. Stopping to smell the roses is good. Immediatelly thinking “how can I kill these to package the smell to sell to others at a profit” is not.
You can experiment with different agent harnesses (pi, hermes, opencode, goose, ..) do experiemnets with gastown and paperclip ai.
I also get a lot of fun at experimenting with ideas i have and had were the focus is not the pure programming language but the application.
But i do think and we will see, as soon as there are clear winners of agentic platforms, we will finetune and adjust these agents.
I have a spreadsheet now of people I can't be fucked with.
What about 'machines' with hands and human-level cognition?
That said, this goal might itself be a non-goal. AI is going to be — or already is — more powerful than any individual human in many ways. But what my intuition points to is that humans will still have plenty of interesting work to do, like the author's example of handwriting code: it shifts from being scalable value creation into a form of craftsmanship.
But this post (and the many I see like it) feels like giving up. And now's not the time for empathetic people to give up.
Technology is how we expand human capability. We are well within our rights to pick and choose how we interact with that capability. But it's starting to terrify me how it seems that the worst people in the world are more than willing to wield this power, while good people opt out. Billionaires are doing a remarkable job at making their vision of the future seem inevitable. Don't fall for it.
If more people aren't willing to help us steer this capability towards a better future, then we all know how this ends.
Maybe it's just that the capability is bad. Adtech, for example, isn't something that anyone uses for good. They blow a lot of smoke about it--looking at you, Apple--but despite the "good-washing" it's all just the same extractive, invasive, dehumanizing business. Bad people will naturally concentrate around this capability. I know because I've worked with a few of them.
AI coding tools seem like they're engineered to undermine cautious, rigorous, and pragmatic engineering discipline. Of course the bosses want that, they see a short term path to massive output increases and nothing sounds better. They'll be cashed out by the time the mess needs to be cleaned up, that's someone else's problem. People who are predisposed to this kind of antisocial behavior are the ones who concentrate around AI tools. Rigorous, careful engineers who care about building maintainable systems that will outlast their tenure find less value in them.
I think it's more nuanced than
> Technology is how we expand human capability.
I think as a general statement about technology as a whole it's true. But do all technologies expand human capabilities? I don't think so.
Advertising tech and AI coding tools are applications of technology stacks that could have been used to create what you and I might agree to be "better" tools. I don't need to tell you why ad tech got created instead of something that is a net societal benefit.
At the same time, I would say that, yes, those applications do indeed expand human capabilities. The important nuance here is whose capability, and to what end.
All I am saying is that opting out of this, in whatever form that takes, hands your agency over to those who would use it to enrich themselves at the cost of others. I sincerely feel that decades of this type of capitulation is exactly how we got to where we are today.
Unlike weapons of mass destruction, though, there's nothing to be gained by working on adtech. Let me unpack that a bit--a "good guy" can justify working on nuclear weapons because if he doesn't his country will be unable to defend itself when the treaties break down. There's no similar existential threat with adtech. If I choose to not work on that poison, I'm not leaving any advantage on the table. My agency in this situation is worthless. At least that conclusion is why I don't do adtech anymore.
I broadly agree with your bigger point though. To the extent that AI coding tools are useful it's important to explore their use, and evaluate whether they're any good. But if I'm in a company that's forcing their use, or putting up token use leaderboards, or any of the other horror stories we've been hearing about, you can be damn sure I'd get myself fired or quit. At that point the inmates are running the asylum, and there's no reason to stay.
I don't subscribe to that even to begin with. Learning "how to use AI" in a a dev workflow takes less than a month, just by practice. Assuming "how to use AI" refers to using agents, rather than just a more advanced auto complete, once you've made the couple of mistakes you can make a couple of times (mainly, leave the coding agent to do something too big for too long), you're caught up. Maybe it take a little more time to get the habits into your workflow.
Leveraging a LLM for coding is orders of magnitude simpler than the code it's writing, and the skills you need to review said code.
These LLMs are prediction machines. They are text generators that are ultimately a bunch of fancy statistics,
Anyone still have some hope for humanity? Or yourselves as a person? Asking for a friend who thinks that the dark horizon has already swallowed all but our eyes, leaving us the brief observers of our oblivion
see! only the developers get the good matches.
Unfortunately, unfettered capitalism has few incentives to support that, and would rather we pushed out software faster with fewer people being paid less. Paying down tech debt has always been challenging.
So is the problem AI, or capitalism?
It also is a great example of why AI has such a PR problem among normal people.
I’m forever getting asked for help by people who suddenly value the human experience when their machine god fails them.
Sometimes fuck ‘em because they devalued me first.
I’ll be the devil’s advocate and suggest that it might not be AI usage, but the technology attracting vilest scum of the Earth. It’s just they were staying mostly silent before, or wasted someone else’s time in different circles.
Conflating a preference for manual creation with opposition to the existence of a tool should be the single biggest signal flare that they are someone who will not argue with you in good faith. They're the ones who barnstorm every single one of these posts to denigrate the author rather than even attempt to empathize with their plight or evaluate the validity of their arguments. Surely the current cohort of HN commenters have seen this repeatedly in just the past five years as technical circles have jumped from cryptocurrency to blockchain to NFTs to LLMs to GenAI; every single one is a "must have", every single one something we "must learn or be left behind forever", and every single one refused to be evaluated on its merits in favor of simply embracing something new for its novelty.
I have given up debating with these people, because they do not wish for debate, they wish for dominance. I have better things to do with my time - as do you, as do all of us - than to give a moment of consideration to a viewpoint that relies on pithy quotes out of context and a reductionist narrative of history to justify their own superiority over others, in lieu of nuanced discourse.
Remember that it is not the obligation of the status quo to defend itself, rather the obligation belongs to those advocating changes to justify and defend their position and its benefits. In that regard, the pro-AI camp continues to come up hollow and empty.
I’d also hope, though, that as humans, we can recognize that tools do not exist in a vacuum, and that their effect on ourselves and society at large can be net negative even if they have a net positive effect on our work (whether due to something fundamental to the tool or due to the way it is being applied at scale). We can’t responsibly leave these discussions out of our analysis of the tool itself and its fitness for purpose, because we are members of society, and our adoption/use of the tool helps to determine that societal impact.
Oops, wrong input field.
how about some true synergy instead of boring zero-sum people? smh. the true poetry here is that zero-sum thinking will become more of a thing of the past so there is some natural comedy with this title
Almost no one I know wants agent usage to be a zero-sum activity. There are a few oddballs who obviously only got into software for the money, so any means to that end is acceptable. That does not stop those with say-so over things like employment (and, if you're in the USA, the associated healthcare), from treating it as a zero-sum activity.
When engineers are being told to maximize token usage, are constantly being brought into meetings where they're expected to reveal their latest and greatest use of LLMs, and not using enough tokens in your role is seen as a negative, then the pressure starts to creep in. Yes, I know this is silly to most people who read this site, and I agree. It's bonkers. But there is certainly something to the idea of "AI psychosis" in upper management that is making agent use zero-sum company-wide.
That shit with upper management sounds stupid af and I've heard the same type of shit from people I know who are in other fields. I'm guessing it's happening from a combination of ignorance, FOMO, investment, etc etc. But that is more of a systemic issue than anything to do with these tools imo.
My biggest problem as an independent contractor is marketing and notoriety. Security has been a race to the bottom for over a decade now, but it's gotten exponentially worse. LLMs can't just do my job, but there are enough people with checkbooks who believe that it can and enough companies out there with an incentive to confirm that belief that it's getting harder for me to find work organically.
If that's indeed the case, then it sounds like an opportunity to get ahead of them since you know they will trip and fall at some point.
That's what I make of what you're describing, while sleep deprived and having given it some light thought LOL. So take that with a lot of salt. But it's kind of what I've been thinking these days anyways. Add to that that entrepreneurship is most likely getting empowered, and I think investing in yourself is the move these days. It will probably characterize the coming years strongly.
The 2008 housing crisis affected everyone. Bubbles that get too big pop across the population, whether they're complicit or not. As a little guy in a big world, with no expertise to truly know if there's a meaningful difference, I have a bit of anxiety about it all. I just don't want to catch collateral.
It might if the subscribers realize/admit that things are failing because they were wrong from the beginning. But that doesn't seem to be how things work.
It's specially hurtful to see open source developers and projects using and supporting the tools of companies which blatantly stole their work - and now profit from it - and that are actively against the open source ideals (Anthropic when trying to close github repos with their code)
I've stopped all my monetary support for open source projects, and moved all my code to a self hosted instance.
I'm not against AI tools I see where they can be useful, I'm just morally opposed to and disgusted of the ways these companies work.
Seems like I'm reading the pope encyclical
What I mean is that the thrilling buddy system coding starts to happen less frequently over a career, and the time for deep exploring and side projects is organically maximized early and during school.
While LLMs have forced that divide to be more stark, the human connection and sense of wonder has always required maintenance, and it's best to get into the habit of maintaining it before your 36th JIRA triage meeting in a week completely destroyed your love of the industry.
Well before LLMs I went through exactly what TFA describes when I had to adapt from grad school labs to industrial labs, then to project management or task leadership (even just filling in for my boss), and each new job has required me to say goodbye to great friends and colleagues and make new ones.
Its just inevitable to fall out of love of the craft, we all could probably write this post for our own reasons.
This is not some kind of universal truth. I can see how being stuck in an unfulfilling job could lead you to say this. But for the last 20+ years I love the craft of writing efficient, dependable, understandable code more with every new insight from every hard-won experience.
Yeah yeah back to Reddit
For real though: you can keep doing artisanal hand-written code as a hobby. Just like you can still write a web server in assembly if you really want to. But that’s just not how professional software development is done anymore. Just a new tool, I don’t think it’s as deep as the author is making it out to be.
That’s easy to say for someone in their 50s who built wealth under favorable conditions.
But it’s quite ignorant and inhumane to say that to someone in their 20s who is just starting their career.
Too bad to see these boomer antics continue to be perpetuated.
So this person's twelve years out of college. You may want to train your fire elsewhere.
But that's exactly my point. He got twelve good years out of it.
>continue to be perpetuated
Because we don’t agree on this premise
The factories that replaced shoemakers actually produced physical shoes that real humans wore. What is your AI producing, other than garbage bug-filled software that no one wants or needs? What real-life human problems is it solving?
Even by trying to reassure (the reader? Himself?) that LLMs are just a tool for humans, he asserts in the final paragraph that software is no longer made by humans. Something something linotype operators.
He believes that AI is making him redundant and soul-less, and at the same time tries to minimize it by reminding everyone that LLMs are a tool - yet he argues that software can now be created without humans so his perception of AI is inconsistent at best.
This is the writing of a person with an existential crisis, which will be obviously egocentric by nature, and I wish the author the best. His approach is not too different of Japanese woodcarvers, Linotype experts switching to computers... there is no real mention about "AI creep" so the extent of AI usage is inconsequential. He has realized he mastered a craft that, as all types of crafts, are subject to technological advancement. At a personal level, it's a hard pill to swallow
Another grief-post with people unable to cope with the fact that the whole structure of learning and work is going to change so they resort to pseudo nostalgia and romanticism. Not to mention that "They are text generators that are ultimately a bunch of fancy statistics" is basically incorrect and belongs in 2024.