ES version is available. Content is displayed in original English for accuracy.
Advertisement
Advertisement
⚡ Community Insights
Discussion Sentiment
64% Positive
Analyzed from 4525 words in the discussion.
Trending Topics
#code#better#llms#more#llm#software#human#don#humans#still
Discussion (62 Comments)Read Original on HackerNews
> But the idea that AI has or will surpass humans any time soon in either capabilities or efficiency is simply not true
AI is already better than most developers. I'm not sure what alternative reality people are remembering, but human coders for the most part have been really awful at writing code. I think the average PR from an LLM is head and shoulders above the average PR from a human. Does it write code in the preferred style and architecture of the project maintainer 100% of the time? No, and neither did humans.
I think there are many legitimate criticisms of AI, but "they suck at coding" isn't one of them. The progress we've seen in the last couple of years alone suggest that very soon they will be better at coding than any person. As a coder of over 30 years, I've embraced this fact and come to terms with it. Leverage your knowledge of systems, software engineering and product design and you can be living in a golden era for software development. That's how it feels to me at least.
And even if the AI is better than most humans, the speed means that you get more defects and issues! If a human developer has a change failure rate of say 10%, (1 in 10 changes causes a defect or issue), and AI is twice as good and only introduces bugs 5% of the time, but submits 10x as many changes, then you go from 1 bug per unit of time to 5 bugs per unit of time, so your velocity is up 10x but your defect rate is up 5x...
You forget to account for the rate of error correction.
You can just choose how many bugs you want now:
https://nolanlawson.com/2026/08/16/you-can-just-choose-how-m...
You mean “a codebase of high technical debt” I think
I often find myself throwing away large amounts of LLM driven code not because it’s bad, but because it doesn’t fit within my attention span. The code itself looks very reasonable, passes tests, benchmarks well, etc. But I throw it away because the models don’t yet “explain” their decisions in ways that elicit psychological safety. Humans are still very good at that, even when their engineering is worse.
If the code passes the (extensive) tests, I don't need to read or understand it. That said, I retired before LLMs became popular, so my experience is limited to vibe coding at home.
(My experience has been the polar opposite: the people I know who are the most “AI pilled” are also the ones who have the shortest technical horizons in terms of how transferable they expect their LLM skills to be.)
Don't use Astral, they want to make you unemployed!
It’s even more powerful with large data and knowledge sources connected.
Takes a lot of work to set up effectively, but when connected to Slack _properly_ (not their MCP; but API which is more powerful), a database of your repo’s PRs/comments, data warehouses including analytics/telemetry and logs; and in a strong harness (including using multiple models simultaneously; like the OMP advisor pattern), what AI can achieve combined your domain expertise and human intelligence is just mind bogglingly crazy.
The larger your codebase / product / volume is; the more powerful it gets. AI has found many needles in haystacks that’s just impossible for a single person or team in large companies; because nobody has all the context.
I’ve embraced it too now. Initially I felt a bit disempowered and just somewhat uncomfortable.
Over time, I realised that I’m still doing serious and interesting engineering: just at a higher level of abstraction.
And for the craft and passion of software engineering, I have a couple of pet projects where I use ‘limited AI’. Good to still keep your wits sharp.
I really can’t imagine what would happen if I didn’t manually intervene sometimes and just kept prompting it for the new behavior I wanted.
On one hand I think, I want to be proud of it, I want to be able to explain it, if it breaks I want to be able to figure out why.
On the other hand, AI can do all of that with badly written code, so who cares.
Edit: however, it still feels like an amazing power tool, but it has taken me months to figure out how to use it.
I have the opposite experience of everyone else I follow online, I find it terrible at green field and great at brownfield. Green field it makes horrible choices as it has nothing to follow.
I'm in no way a very good programmer, or very smart, but the code I saw most of my co-workers writing was about the same quality as AI, not very good.
For a short while, maybe. If an LLM can keep track of a 500KSLoC codebase, it's gonna easily replace systems knowledge workers, software designers and product designers.
None of systems knowledge, software engineering and product design is a moat against this.
Now the thing to claim “a better coder than humans”, is you can’t just stop at making a production grade C Compiler, you then also need to make the leap to make something new that is a definite improvement over everything that existed before it. This is an also a question of taste not just implementation chops. Think Zigs cross platform C Compiler, Rusts memory safety opinionated compiler and more.
The day AI can do both, implement a complex production grade project, and make a conceptual actual improvement upon SOTA is the day I’ll agree AI has become better than humans at coding. I’ve underestimated AI in the past, maybe with 10T of compute they’ll get there, maybe they won’t , we’ll know in the coming years
My experience, no its not. It just doesn't fight back as much when you tell it that its wrong.
My sister team is vibecoding the shit out of a couple of product PoCs. There is only one person on that team that appears to understand how to vibe code properly. the rest are just producing shite and breaking the service everytime they deploy. However, the code it creates is fine enough, just the architecture is bad, or the prompter is bad.
_however_ the problem with the post is that its using tangential metrics to prove the point. The opensource maintainer bit doesn't always mean that the output is bad, it means that either:
1) the maintainers hate AI
2) the shit they are getting is huge and takes too long to review
3) The shite they are getting solves a specific problem for one user at the expense of everyone else
4) the PR is nonsense.
only one of those options area signal for code quality from LLMs. the rest are about the skill of the creator, or attitude/time budget of the maintainer.
They are better when copying and pasting expert code, even when they don't understand it.
A tool can only be as good as the person who use it.
In my experience, the top models still generate tons of useless slop on any non-trivial implementation request that I do not essentially solve in the prompt beforehand (change this class, this function, etc). They also still make trivial errors that no human would make (although the inverse is also true). In this sense, they do suck at coding.
On the other hand, even weaker models can understand large sections of code, come up with correct implementations of changes, and catch non-trivial edge cases in many situations that is obviously better than most devs. In this sense, they are better than almost all human devs, especially when considering the time and cost.
Perhaps in the long term AI will help us distinguish better between different types of coding tasks and programming disciplines
By what measure? How do you even compare developer skill?
From where I snipe, I see a big divide between those...
1) that try to surrender to AI, aiming to fully replace value-added intellectual effort and often also to augment enterprise value-mask slop busywork...
...they fail, and succeed, and the collective suffers.
2) those that ride AI to get more challenged, more feedback if any kind, to tread further but with attention to the right details
...they succeed
Like I see people say this, and yet the teams that are AI maxing produce worse code than ever. Software has rapidly gotten more unstable and unsustainable over the past three or so years than I've experienced in the past 20.
Let's assume for a moment that you're correct. That AI is already better than most developers, for whatever definition of "better" you like, and they will very soon be better than any person. (I think this is a total fantasy and you've failed to recognise the limitations as the article points out, but I digress.)
In that case the end goal of these companies is to replace all software engineers. Do you not see that? They've not been hiding this fact. It's a good thing for you and I that these models don't work, because if they did the "golden age" is not coming for us, it's coming for people who own compute capacity and the rest of us will become labourers.
Fixed weights don't preclude in-context learning and out-of-the-loop weight updates. And that's the only principled limitation mentioned in the post.
Humans have value judgement. AI doesn't. AI doesn't "know" what good code even is. For something to be good, there has to be a purpose. Good is the measure of how well that purpose is fulfilled.
Humans can write better or worse code, but AI is not even in the category of things that can write better or worse code. A human needs to be there to tell bad code from good code.
* * *
Also, in my personal experience with AI code: I'm yet to see good code (but I haven't worked with people who are good at directing AI towards their goals). All code I've seen generated by AI so far ranged from "absolute garbage" to "passable". Which, most likely, reflects the ability of those who managed the tool: before they did that, their code was also atrocious. It was easier to deal with, because the velocity at which these people produced garbage didn't cause a deluge in the same way they do it with AI help.
* * *
A note on what I believe to be good code and its distribution. First of all, I agree with you on that the vast majority of code produced to date is very bad. There are many reasons for it: until few years ago the demand for programmers was smaller than supply and the industry was on course to create conditions for very bad programmers to succeed anyhow (help the losers lose less, tee-hee!). It still didn't recover from all the "paradigms" it created to support bad programmers.
Unlike in well-established fields, where you'd expect normal distribution in terms of how skillful the workers are (i.e. you'd expect very few to be very bad and very few to be very good, but most would be good enough), the distribution in programming is exponential: overwhelming majority are at the proverbial bottom of the barrel, only a few are OK, and you probably will never meet a truly good one. This defies intuition and leads us to assume that the barely palatable is the best it can possibly be. And that's, roughly, where AI is at at the moment.
Think about it: code is cheap now. You'll have accountants realizing they can create scripts to automate their work flow, so they hack together something. These scripts will become the backbone of the accounting pipeline of a company. Now someone needs to productionalize and maintain these scripts, but the accountants don't even have the vocabulary required for that, so they hire a software engineer to do so.
A simplified example, but I can see that happening. The only (personal) issue I have with this is that the LLMs get to take the fun part of the job.
https://arxiv.org/pdf/2510.04950
It was published a while ago. But I wonder if it still holds true today.
>Also stop saying “please” to an LLM. It does not have any feelings.
It emulates having feelings and it's a next token predictor. The next token in a dataset where you respond to an engineer rudely and call them stupid is rarely said engineer locking in and delivering incredible code. You have to play along to get the output you want.
> Again, I recommend people try running small LLMs locally where temperature and other settings are fully exposed and configurable to see this themselves.
This is like saying you should experiment with a paper airplane to see why fighter jets are overrated. Also the general understanding of temperature is not super solid here - it's not just that its "too boring" without it - random sampling is required for the models to work.
>Why are benchmarks showing they're still improving?
Ironically this section is far too generous to LLMs and benchmarks. lLMs cheat and companies benchmax. Don't trust benchmarks. They lie
Also in the what I do section - are these using local LLMs too? "Sometimes looping llms on itself can make it fix its own errors" feels very 2025 - the modern state of things is more "we've given up on one shots and getting it to produce the right answer immediately, set it up with a test harness so it can fix its own mistakes and let it loop otherwise it won't work." Also "asked fellow developers to make sure their code is well structured, easy to follow and documented. LLMs unfortunately make it easier for people to cheat in this regard" - really? Easy to follow? Maybe gpt models with good steering but trying to get an anthropic model to speak coherently and clearly and write documentation that isn't incomprehensible slop is a Herculean effort
They look smarter to people they've made dumber.
They’re getting smarter. Next question.
I was intrigued by the hype and gave a chance this week to codex+sol 5.6 and cc+opus 5. They cheated, lied, disobeyed, and shipped subtle bugs so often, it wasted more time that if I did it myself.
Is half of the industry under AI psychosis right now ?
A bad coder sees bot code as an inprovement. A good coder sees bot code as deterioration.
They are both correct.
> Also, it seems that many don’t want to learn but instead expect to have all understanding outsourced to LLMs. Many seniors have noticed this and have stopped teaching juniors as the seniors don’t like the feeling of having their time wasted by teaching people who don’t want to learn.
Or may be it is because now a junior dev is expected to deliver to output of a senior?
> Also stop saying “please” to an LLM. It does not have any feelings.
Yes, but I still prefer a nice tone. Like why should I change my manners just because it has no feelings? If anything the statistic predicts a friendlier answer when I say "please".
> Again, I recommend people try running small LLMs locally where temperature and other settings are fully exposed and configurable to see this themselves. It is a good antidote to falling for the illusion that LLMs would actually be intelligent.
It's like recommending someone to buy the cheapest Lenovo Thinkpad to prove that Lenovo sucks.
> However, the best models still have a pass rate of only about 50% on the Humanity’s Last Exam.
Haha, as if he (or really most people) even would understand 50% of those questions. I find it rather mind blowing that it's possible to put such diverse knowledge into a couple of TB. Or may be I'm just an idiot and it's common knowledge, "how many paired tendons are supported by the sesamoid bone of hummingbirds within Apodiformes".
> LLMs are not a scam, but a useful tool and technology that has its uses. But the idea that AI has or will surpass humans any time soon in either capabilities or efficiency is simply not true
He's not wrong that LLMs are just useful tools but isn't it the purpose of a tool to surpass human capabilities and efficiency? Like even a bicycle makes traveling more efficient than just walking and a car has the capability to transport more items than any human. And its know more than two decades since computer surpassed human capabilities in chess. If a tool is neither more capable nor efficient, it's just a useless tool.
I mean I get his point that GenAI is to some degree over hyped but his arguments just dont hold in my opinion.
> Yes, but I still prefer a nice tone. Like why should I change my manners just because it has no feelings? If anything the statistic predicts a friendlier answer when I say "please".
It's a probabilistic generator. If you set the tone of the conversation it will follow it.
Try inserting a few jokes, puns etc into a conversation and you'll see that it responds kn kind.
Just like if you ask highly specific technical questions, it responds using highly technical language. If you ask questions in legalese it responds using precise legal terms.
That's what happens when a token predictor has a conversation: it can't have any tone other than what you give it.
So continue saying please, and you'll have more pleasant conversations.
Like you I generally prefer polite phrasings over rude ones, and I'm certainly not going to tell you how you should communicate to LLM models, however it's interesting to note that being somewhat rude and to the point seems to have some objective advantages, see for example:
Mind Your Tone: Investigating How Prompt Politeness Affects LLM Accuracy (short paper)
https://arxiv.org/abs/2510.04950
Understanding Tone-Dependent Inference Cost in Large Language Models
https://arxiv.org/abs/2607.23915
I would hate for this style of communication to leak through to daily communications with humans (e.g. colleagues) though, so even just for that it can make sense to hold on to a polite form of communication :-).
Second, I agree with the distinction evals are not enough, and this is dangerously making people reckless, I think the conclusion is still quite incorrect.
There is essentially nothing magical about being human. Our primitive brains were trained on logic, reasoning and processes. An LLM is simply this on less efficient hardware, but, improving at a speed faster than human evolution.
It's widely reported on and discussed. You'll find plenty of people saying it isn't one, but I'm sure you can find an economist willing to take any position you like if you look hard enough. Not to mention everyone who's financially incentivised to continue to claim that there isn't a bubble.
> Even the most skeptical investors to bubbles like Buffet and Burry, have large stakes in Google and Microsoft, front line bets on AI.
Burry has been warning about an AI bubble for years.
> There is essentially nothing magical about being human. Our primitive brains were trained on logic, reasoning and processes. An LLM is simply this on less efficient hardware, but, improving at a speed faster than human evolution.
There is no basis for this claim. There is no evidence that human brains and the process of evolution that led to modern humans is at all similar to how LLMs function. If you're going to evaluate an LLM then evaluate an LLM, you can't say "Humans can do X, LLMs are like humans, therefore LLMs can/will do X" because there is no evidence to connect them.
Our media is a weaponized crap factory. There is literally nothing but propaganda and insider knowledge. You're network is how you survive today, mass media and social media and anyone that uses these medias as guidance is weaponized against their own interests. Your own critical analysis, secondary considerations, knowledge of real factual history, and your effective communications skills are how you survive. Not by listening to others and their parrot echo of some media propaganda trying to control the narrative.
It doesn't matter that AI is useful (but so was internet).
It doesn't matter that revenues exist (mostly fake and circular).
It doesn't matter that real unit costs go down (models are getting commoditised).
It doesn't matter that the margins are high (companies are secretly hiding the real costs because that would pop the bubble).
In their world view, the bubble is predestined. There's nothing that could prevent it - even with higher demand, even with higher revenues, even with higher margins. Nothing.
They go nowhere because people are using wildly different definitions and contexts. There's one already in here about how ai is better than humans at coding.
- Yes llms are better at the mechanics of coding
- no they're not good enough for overall software dev.
- yes, this stuff should've been automated years ago in frameworks and in libs, or through sane programming langs that dealt with memory and logic flow better.
- yes, they're much more useful for documentation, search, etc. than they are at actual coding
- no, your stats aren't useful - 90% of coding is meaningless if you've also massively increased the amount of slop produced
- yes, llms do make you more productive overall, whether it's 10-30% or 1000% is context dependent
- yes and no that AI will change everything; no it doens't make sense to keep comparing pre-ai and post-ai worlds, it is very likely that the gains cancel each other out and we all just move up a layer of abstraction, and end up in a simlar situation to now.
- no we don't know what will actually happen to the job market, things can remain irrational longer than you can remain solvent. We've had the technology to be where are 20-30 years ago if we were focused as a society. Things take time, and real world is complicated.
My devs keep throwing Claude-generated writeups at me that look okay at first glance but fall apart when you actually try to follow the instructions they lay out (which makes them useless for anyone who's not already familiar with the product or feature they're describing), and/or have major structural and logical gaps, and/or gloss over messy details in a way that makes the product or feature sound better but doesn't reflect the reality of what we actually built.
Which turns into an ironic ouroboros where I then have to punch my own queries into an LLM and ask it to read our codebase, compare those details against the original writeup, try to decipher what the hell it was trying to say, etc. And even that turns into an extended back-and-forth where my LLM is like "The original writeup is inaccurate; it should say X" and I'm like "Are you sure? That contradicts what I know about related concept Y", and then the LLM is like "Good point, I dug into it more and the real answer is Z." (And even then, god only knows if Z is correct. I still have to get real developer eyeballs on anything before it goes live.)
The granularity at which documentation parity is kept makes a big difference.
For example, brekaign things down into explicit assertion blocks in the code, and a link to the documentaiton that covers that assertion. So block-to-block tracking works well. Tracking completely separate documentation to independent codebase... idk I haven't tried it but my guess it that the context windows are too small for that.
But minor things like making sure there are unique error codes, that the codes somewhat make sense, etc. are all helpful. If nothing else, the model being stupid and acting as a rubber duck for pair-programming is useful for knowing what to document given a different perspective.
70 years of working on the 'how do I reuse code?'. And the solution just takes all the electricity in the world.
No its not. No one serious truly believes
1. AI is going away
2. its obvious that the total market cap of AI related stocks will go down more than 60-70%
Its obviously true that there are some related AI companies that will go bust as with any new technology. And its also true that the rise won't be monotonic.
What we instead get is perpetually pushed timelines for the bubbles and a religious belief that the end will come and people would pay for it.
To drive my point further, I ask for a single metric you would like to see that should be a certain way for you to believe it is not a bubble. Should the revenues have been 2x higher? Should the costs be 3x lower? Anything?