Opus 4.5 max (1m tokens) and above were the tipping point for me, before that, I agree with 100% of what you said.
But even with Opus 4.6 max / GPT 5.4 high it takes time, you need to provide the right context, add skills / subagents, include tribal knowledge, have a clear workflow, just like you onboard a new developer. But once you get there, you can definitely get it to do larger and larger tasks, and you definitely get (at least the illusion) that it "understands" that it's doing.
It's not perfect, but definitely can code entire features, that pass rigorous code review (by more than one human + security scanners + several AI code reviewers that review every single line and ensure the author also understands what they wrote)
Wow, that's such a drastic different experience than mine. May I ask what toolset are you using? Are you limited to using your home grown "AcmeCode" or have full access to Claude Code / Cursor with the latest and greatest models, 1M context size, full repo access?
I see it generating between 50% to 90% accuracy in both small and large tasks, as in the PRs it generates range between being 50% usable code that a human can tweak, to 90% solution (with the occasional 100% wow, it actually did it, no comments, let's merge)
I also found it to be a skillset, some engineers seem to find it easier to articulate what they want and some have it easier to think while writing code.
I used to think that the people who keep saying (in March 2026) that AI does not generate good code are just not smart and ask stupid prompts.
I think I've amended that thought. They are not necessarily lacking in intelligence. I hypothesize that LLMs pick up on optimism and pessimism among other sentiments in the incoming prompt: someone prompting with no hope that the result will be useful end up with useless garbage output and vice versa.
That sounds a lot more like confirmation bias than any real effect on the AI's output.
Gung-ho AI advocates overlook problems and seem to focus more on the potential they see for the future, giving everything a nice rose tint.
Pessimists will focus on the problems they encounter and likely not put in as much effort to get the results they want, so they likely see worse results than they might have otherwise achieved and worse than what the optimist saw.
That's a valid sounding argument. However many people with no strong view either way are producing functional, good code with AI daily, and the original context of this thread is about someone who has never been able to produce anything committable. Many, many real world experiences show something excellent and ready to go from a simple one shot.
This is kinda like that thing about how psychic mediums supposedly can't medium if there's a skeptic in the room. Goes to show that AI really is a modern-day ouija board.
Don’t know why you’re getting downvoted, this is a fascinating hypothesis and honestly super believable. It makes way more sense than the intuitive belief that there’s actually something under the human skin suit understanding any of this code.
It's probably more to do with the intelligence required to know when a specific type of code will yield poor future coding integrations and large scale implementation.
It's pretty clear that people think greenfield projects can constantly be slopified and that AI will always be able to dig them another logical connection, so it doesn't matter which abstraction the AI chose this time; it can always be better.
This is akin to people who think we can just keep using oil to fuel technological growth because it'll some how improve the ability of technology to solve climate problems.
It's akin to the techno capitalist cult of "effective altruism" that assumes there's no way you could f'up the world that you can't fix with "good deeds"
There's a lot of hidden context in evaluating the output of LLMs, and if you're just looking at todays success, you'll come away with a much different view that if you're looking at next year's.
Optimism is only then, in this case, that you believe the AI will keep getting more powerful that it'll always clean up todays mess.
I call this techno magic, indistinguishable from religious 'optimism'
What you said about "we're all cooked" and "AI is useless" is literally me and everyone I know switching between the two on an hourly basis...
I find it the most exciting time for me as a builder, I can just get more things done.
Professionally, I'm dreading for our future, but I'm sure it will be better than I fear, worse than I hope.
From a toolset, I use the usual, Cursor (Super expensive if you go with Opus 4.6 max, but their computer use is game changing, although soon will become a commodity), Claude code (pro max plan) - is my new favorite. Trying out codex, and even copilot as it's practically free if you have enterprise GitHub. I'm going to probably move to Claude Code, I'm paying way too much for Cursor, and I don't really need tab completion anymore... once Claude Code will have a decent computer use environment, I'll probably cancel my Cursor account. Or... I'll just use my own with OpenClaw, but I'm not going to give it any work / personal access, only access to stuff that is publicly available (e.g. run sanity as a regular user). Playing with skills, subagents, agent teams, etc... it's all just markdowns and json files all the way down...
About our professional future:
I'm not going to start learning to be a plumber / electrician / A/C repair etc, and I am not going to recommend my children to do so either, but I am not sure I will push them to learn Computer Science, unless they really want to do Computer Science.
What excites me the most right now is my experiments with OpenClaw / NanoClaw, I'm just having a blast.
tl;dr most exciting yet terrifying times of my life.
I've gone back and forth on it a lot myself, but lately I've been more optimistic, for a couple of reasons.
While the final impact LLMs will have is yet to be determined (the hype cycle has to calm down, we need time to see impacts in production software, and there is inevitably going to be some kind of collapse in the market at some point), its undoubtable that it will improve overall productivity (though I think it's going to be far more nuanced then most people think). But with that productivity improvement will come a substantial increase in complexity and demand for work. We see this playout every single time some tool comes along and makes engineers in any field more productive. Those changes will also take time, but I suspect we're going to see a larger number of smaller teams working on more projects.
And ultimately, this change is coming for basically all industries. The only industries that might remain totally unaffected are ones that rely entirely on manual labor, but even then the actual business side of the business will also be impacted. At the end of the day I think it's better to be in a position to understand and (even to a small degree) influence the way things are going, instead of just being along for the ride.
If the only value someone brings is the ability to take a spec from someone else and churn out a module/component/class/whatever, they should be very very worried right now. But that doesn't describe a single software engineer I know.
A LOT of people are taking this seriously and not getting the (no so?) subtle satire in this. I fell for it at first glance too, had to do a double take. Some of the smartest people I know asked me for my thoughts on this.
The scary part - what's today is satire, is tomorrow's stealth mode startup.
Linking it here since it's easy to miss. It seems he is using the popularity of this to help a friend recovering from brain surgery, I think this makes this project even more awesome in my book.
In my very humble view, the mythical 10x developer can now be a 100x developer, and the 2x developer usually stays a 2x developer. We live in two parallel worlds right now. Some run an army of agents and ship somehow working and testable code, and some try to prove AI is not as good as them.
I know it sounds like a good take, but I don’t really see it happening much in real life anymore.
It’s more like the 1x developer gets frustrated and defensive, and shows the 5 stages of grief, try using AI and finds all the reason why it’s bad. Then goes ahead and refactors everything and breaks production.
I’m at this point too. I desperately want to hate AI, but it’s so incredibly competent. People who say LLM’s aren’t good generally just aren’t good at them
I haven’t seen any evidence of an army of agents producing unicorn companies. If this was the case we’d see a rash of < 10 employee startups being worth $1 billion, and to my knowledge that’s zero
I hate predictions, but when the dust settles, Copilot will take the lead. They are deep in the enterprise ecosystem, and they practically give it for free.
After the "upside down cup" debacle, and the "walk vs drive to the carwash" conundrum, and so many other examples where GPT 5.2 thinking failed miserably and Opus 4.6 and (even Sonnet 4.6 extended thinking) nailed it, I think they earned people wanting to cancel their subscription regardless of yesterday's events.
Great idea. It sadly doesn’t work for me, I went to pick a topic and some of the answers were pre-submitted, some of the questions did nothing after clicking “check”, then it just ended. Maybe it’s the HN hug of death?
I like coding, I really do. But like you, I like building things more than I like the way I build them. I do not find myself miss writing code by hand as much.
I do find it that the developers that focused on "build the right things" mourn less than those who focused on "build things right".
But I do worry. The main question is this - will there be a day that AI will know what are "the right things to build" and have the "agency" (or illusion of) to do it better than an AI+human (assuming AI will get faster to the "build things right" phase, which is not there yet)
My main hope is this - AI can beat a human in chess for a while now, we still play chess, people earn money from playing chess, teaching chess, chess players are still celebrated, youtube influencers still get monetized for analyzing games of celebrity chess players, even though the top human chess player will likely lose to a stockfish engine running on my iPhone. So maybe there is hope.
> will there be a day that AI will know what are "the right things to build" and have the "agency" (or illusion of) to do it better than an AI+human (assuming AI will get faster to the "build things right" phase, which is not there yet)
Of course, and if LLMs keep improving at current rates it will happen much faster than people think.
Arguably you don't need junior software engineers anymore. When you also don't need senior software engineers anymore it isn't that much of a jump to not needing project managers, managers in general or even software companies at all anymore.
Most people, in order to protect their own ego, will assume *their* job is safe until the job one rung down from them disappears and then the justified worrying will begin.
People on the "right things to build" track love to point out how bad people are at describing requirements, so assume their job as a subject matter expert and/or customer-facing liaison will be safe, but does it matter how bad people are at describing requirements if iteration is lightning fast with the human element removed?
Yes, maybe someone who needs software and who isn't historically some sort of software designer is going to have to prompt the LLM 250 times to reach what they really want, but that'll eventually still be faster than involving any humans in a single meeting or phone call. And a lot of people just won't really need software as we currently think about it at all, they'll just be passing one-off tasks to the AI.
The real question is what happens when the labor market for non-physical work completely implodes as AI eats it all. Based on current trends I'm going to predict in terms of economics and politics we handle it as poorly as possible leading to violent revolution and possible societal collapse, but I'd love to be wrong.
> The real question is what happens when the labor market for non-physical work completely implodes as AI eats it all. Based on current trends I'm going to predict in terms of economics and politics we handle it as poorly as possible leading to violent revolution and possible societal collapse, but I'd love to be wrong.
Exactly and the world has to start talking about it. Eventually everybody will, including all sorts of politicians who advocate to 'finally tackle the problem', which will be too late.
> I do find it that the developers that focused on "build the right things" mourn less than those who focused on "build things right".
I've always been strongly in the first category, but... the issue is that 10x more people will be able to build the right things. And if I build the right thing, it will be easy to copy. The market will get crowded, so distribution will become even harder than it is today. Success will be determined by personal brand, social media presence, social connections.
For me, photography is the metaphor - https://raskie.com/post/we-have-ai-at-home - We've had the technology to produce a perfect 2D likeness of a subject for close to two centuries now, and people are still painting.
Video didn't kill the radio star either. In fact the radio star has become more popular than ever in this, the era of the podcast.
While what you're saying is true, I think it is important to recognize that painting in a way that generates a livable income is mostly a marketing gig.
Likewise, being a podcaster, or "influencer" in general, is all about charisma and marketing.
So with value destruction for knowledge workers (and perhaps physical workers too once you factor in robotics) we may in fact be moving into a real "attention economy" where all value is related to being a charismatic marketer, which will be good for some people for a while, terrible for the majority, but even for the winners it seems like a limited reprieve. Historically speaking charismatic marketers can only really exist through the patronage of people who mostly aren't themselves charismatic marketers. Without patrons (who have disposable income to share) the charismatic marketers are eventually just as fucked as everyone else.
Art, photography, acting, music - none of them are good career choices. You'll either be one of the fortunate few, or you'll struggle to make a living. Sucks but that's how it is.
> will there be a day that AI will know what are "the right things to build" and have the "agency" (or illusion of) to do it better than an AI+human
I share this sentiment. It's really cool that these systems can do 80% of the work. But given what this 80% entails, I don't see a moat around that remaining 20%.
I can't explain it in a way that will make sense or base it on any data, but I do see that moat all the time. For example, Microsoft / GitHub somehow taking a 3 year lead (with Co-Pilot) and losing it to Cursor in months, and still catching up.
Microsoft / GitHub have no real limitation to doing better/faster, maybe it's the big company mentality, moving slower, fear of taking risks where you have a lot to lose, or when the personal incentive for a product manager at github is much much lower than the one of a co-founder of a seed stage startup. Co-Pilot was a microscopic line item for Microsoft as a whole, and probably marginal for GitHub too. But for Cursor, this was everything.
This is why we have innovation, if mega-corps didn't promote people to their level of incompetence, if bureaucracy and politics didn't ruin every good thing, if private equity didn't bleed every beloved product to the last penny, we would have no chance for any innovation or entrepreneurship because these company have practically close to unlimited resources.
So my only conclusion from this is - the moat is sometimes just the will to do better, to dare to say, I don't care if someone has a billion dollars to compete with me, I'll still do better.
In other words, don't underestimate the power of big companies to make colossal mistakes and build crappy products. My only worry is, that AI would not make the same mistakes an we'll basically have a handful of companies in the world (the makers of models, owner of tokens e.g. OpenAI, Anthropic, Google, Amazon, Meta, xAi), if AI led product teams will be able to not make the mistakes of modern corporations of ruining everything good that they got in their hands, then maybe software related entrepreneurship will be dead.
> "build the right things" [vs] "build things right"
I think this (frequent) comparison is incorrect. There are times when quality doesn't matter and times that it does. Without that context these discussions are meaningless.
If I build my own table no one really gives a shit about the quality besides me and maybe my friends judging me.
But if I sell it, well then people certainly care[0] and they have every right to.
If I build my own deck at my house people do also care and there's a reason I need to get permits for this, because the danger it can cause to others. It's not a crazy thing to get your deck inspected and that's really all there is to it.
So I don't get these conversations because people are just talking past one another. Look, no one gives a fuck if you poorly vibe code your personal website, or at least it is gonna be the same level as building your own table. But if Ikea starts shipping tables with missing legs (even if it is just 1%) then I sure give a fuck and all the customers have a right to be upset.
I really think a major part of this concern with vibe coding is about something bigger. It is about slop in general. In the software industry we've been getting sloppier and sloppier and LLMs significantly amplify that. It really doesn't matter if you can vibe code something with no mistakes, what matters is what the businesses do. Let's be honest, they're rushing and don't care about quality because they have markets cornered and consumers are unable to accurately evaluate products prior to purchase. That's the textbook conditions for a lemon market. I mean the companies outsource tech support so you call and someone picks up who's accent makes you suspicious of their real name being "Steve". After all, it is the fourth "Steve" you've talked to as you get passed around from support person to support person. The same companies who contract out coders from poor countries and where you find random comments in another language. That's the way things have been going. More vaporware. More half baked products.
So yeah, when you have no cake the half baked cake is probably better than nothing. At home it also doesn't matter if you're eating a half baked cake or one that competes with the best bakers in the world. But for everyday people who can't bake their own cakes, what do they do? All they see is a box with a cake in it, one is $1, another for $10, and another other is $100. They look the same but they can't know until they take a bite. You try enough of the $1 cakes and by the time you give up the $10 cakes are all gone. By the time you get so frustrated you'll buy the $100 cake they're gone too.
I don't dislike vibe coding because it is "building things the wrong way" or any of that pretentious notion. I, and I believe most people with a similar opinion, care because "the right things" aren't being built. Most people don't care how things were built, but they sure do care about the result. Really people only start caring about how the sausage is made when they find out that something distasteful is being served and concealed from them. It's why everyone is saying "slop".
So when people make this false dichotomy it just feels like people aren't listing to what's actually being said.
[0] Mind you, it is much easier for an inexperienced person to judge the quality of a table than software. You don't need to be a carpenter to know a table's leg is missing or that it is wobbly but that doesn't always hold true for more sophisticated things like software or even cars. If you haven't guessed already, I'm referencing lemon markets: https://en.wikipedia.org/wiki/The_Market_for_Lemons
Of course. I mean, my view is that it needs to be "build the right things right", vs "build things right and then discover if they are the right things". It's a stab at premature optimisation, focusing on code elegance more than delivering working software. Code simplicity, good design, scalability, are super important for maintainability, even in the age of AI (maybe even more so).
But considering that AI will more and more "build things right" by default, it's up to us humans to decide what are the "right things to build".
Once AI knows what are the "right things to build" better than humans, this is AGI in my book, and also the end of classical capitalism as we know it. Yes, there will still be room for "human generated" market, like we have today (photography didn't kill painting, but it made it a much less of a main employment option)
In a way, AI is the great equality maker, in the past the strongest men prevailed, then when muscles were not the main means to assert force, it was the intellect, now it's just sheer want. You want to do something, now you can, you have no excuses, you just need to believe it's possible, and do it.
As someone else said, agency is eating the world. For now.
> it needs to be "build the right things right", vs "build things right and then discover if they are the right things"
I still think this is a bad comparison and I hoped my prior comment would handle this. Frankly, you're always going to end up in the second situation[0] simply because of 2 hard truths. 1) you're not omniscient and 2) even if you were, the environment isn't static.
> But considering that AI will more and more "build things right" by default
And this is something I don't believe. I say a lot more here[1] but you can skip my entire comment and just read what Dijkstra has to say himself. I dislike that we often pigeonhole this LLM coding conversation into one about a deterministic vs probabilistic language. Really the reason I'm not in favor of LLMs is because I'm not in favor of natural language programming[2]. The reason I'm not in favor of natural language programming has nothing to do with its probabilistic nature and everything to do with its lack of precision[3].
I'm with Dijkstra because, like him, I believe we invented symbolic formalism for a reason. Like him, I believe that abstraction is incredibly useful and powerful, but it is about the right abstraction for the job.
[2] At the end of the day, that's what they are. Even if they produce code you're still treating it as a transpiler: turning natural language into code.
[3] Okay, technically it does but that's because probability has to do with this[4] and I'm trying to communicate better and most people aren't going to connect the dots (pun intended) between function mapping and probabilities. The lack of precision is inherently representable through the language of probability but most people aren't familiar with terms like "image" and "pre-image" nor "push-forward" and "pull-back". The pedantic nature of this note is precisely illustrative of my point.
> The reason I'm not in favor of natural language programming has nothing to do with its probabilistic nature and everything to do with its lack of precision
Yeah, even if they're made to be 100% deterministic, you've now got a programming language whose rules are deterministic, but hard to understand. You've effectively pinned the meaning of the natural language in some way, but not a way that anyone can effectively learn, and one that doesn't necessarily match their understanding of the actual natural language.
And it's weird that this even needs to be argued given that our long explanations are needed to even convey fairly simple concepts. Not to mention that it still relies upon correct interpretation.
The result of natural language programming is either an extremely limited programming language or an extremely verbose one (again, look at law). Presumably it'll result in both.
It's a nice idea but ignores the reason we invented symbolic languages in the first place. They were invented after natural language. It's not like code is some vestigial language raiment. We're trying to replace it because it's hard and annoying. But I'm certain that's mainly due to the level of abstraction we're trying to work with more than due to the language we're using
I'm highly worried that you are right. But what gives me hope is that people still play chess, I'd argue even more than ever. People still buy paper books and vinyl records. People still appreciated handwritten greeting cards over printed ones, pay extra to listen to live music where the recorded one is free and will likely sound much better. People are willing to pay an order of magnitude more for a sit in a theater for a live play, or pay premium for handmade products over their almost impossible to distinguish knock offs.
But even with Opus 4.6 max / GPT 5.4 high it takes time, you need to provide the right context, add skills / subagents, include tribal knowledge, have a clear workflow, just like you onboard a new developer. But once you get there, you can definitely get it to do larger and larger tasks, and you definitely get (at least the illusion) that it "understands" that it's doing.
It's not perfect, but definitely can code entire features, that pass rigorous code review (by more than one human + security scanners + several AI code reviewers that review every single line and ensure the author also understands what they wrote)
reply