[budsniffer952]

Popped over to HackerNews, read two comment sections and the top comments in both articles were users saying the same thing: "AI can't write code! The whole thing will come crumbling down any minute! Just you wait!"

I've never seen this community like this. Are these people cooked? We are years into this and they haven't been able to figure it out? They are going to continue to tell people using these tools successfully every day that, actually, it's just a mirage?

[techblueberry]

This is roughly in line with where the thought leadership is:

https://newsletter.pragmaticengineer.com/p/context-engineeri...

But you know, maybe you’re having a better experience? You should be a consultant.

[discreteevent]

> You should be a consultant.

It's a bot. They are everywhere trying to keep the bubble inflated.

[csomar]

I highly suspect they are real humans being paid to do it. AI will have produced a much more detectable result.

[Kiro]

Yeah, it's baffling. I can't relate to these statements at all. What are people doing? Surely the smart people of HN would have been able to figure this out a long time ago.

I also don't find these people in real life. Even the most junior developers I know are able to navigate this without creating this supposed mess.

[gregoryl]

Its a bit unkind to talk like this - the obvious and equally unproductive response is to question if you are really as good as you think you are. Are those junior developers not making a mess, or do you lack the insight to see it?

[anal_reactor]

Big codebases tend to become a mess anyway so if your company has experience dealing with shit, the fact that now shit is AI-generated doesn't substantially change anything.

And I'm not joking. Imagine it this way - managing a group of skilled professionals is a completely different skill from managing a bunch of alcoholics doing a minimum-wage job, and sometimes the latter situation is just the reality you find yourself in.

[therealdrag0]

Good shit, anal reactor.

[throwaway314155]

Or perhaps the ones complaining aren’t as good as they think they are at coding, prompting, etc.

[lelele]

But if so, then where is the secret prompt sauce for getting those results, please? It's not like we complainers don't read and put in practice all we can about how to refine a spec prompt after prompt, yet our results are still disappointing. To me, that's no wonder at all, as we haven't reached AGI yet; hence, no matter how impressive and productive these tools are, I totally don't expect them to actually know what they are doing.

To be clear: LLMs write about 99% of my code, but with constant steering.

[therealdrag0]

A lot of this is just talking past each other. AI being good enough to write 99% of your code but then you still complain about it sucking. Glass half full vs half empty.

[lelele]

This.

[Philip-J-Fry]

I've seen two camps of people within the same company I work at. Some think AI is generating acceptable code. Others think it's generating slop.

I sit in the middle a lot of the time as my opinion on code quality vary depending on how important the code actually is.

I will say one thing, the people who I generally deemed as worse developers prior to the wide AI roll out are the ones that are more accepting of AI code. And consequently, they're the ones spending a magnitude more money each day. The people who were writing better code and, in my opinion, were stronger developers, are spending less on AI and are generating more acceptable code when they use it. And they can spot issues from a mile away, because they actually understand the code being generated. They're not just committing lots of redundant code.

We had an AI outage for a day or two a few weeks back. The reaction from some people was like it was the end of the world and they couldn't do any work. Some projects got put on hold because no one actually knew how to work on the code base without AI. Other, better in my opinion, developers just went shrugged it off and got on with their job.

AI over-dependence is a big issue that people will face more and more. Weaker juniors used to just be a bit slower and stagnate a bit, but they could still kinda work independently and understand what they were doing. Now they are almost entirely prompt monkeys, take away their Claude Code terminal and they are completely stumped.

[flohofwoe]

> I sit in the middle a lot of the time as my opinion on code quality vary depending on how important the code actually is.

This is really the key insight that most people on the extreme ends of the discussion don't seem to grasp and then talk in absolutisms because they think that their tiny village is the whole world.

It's the same kind of 'extremism' that dominates most discussions about memory safety btw.

For some types of programs, code quality simply doesn't matter as long the program does its job, and generating those programs even with the most sloppiest vibecoding approaches is totally fine.

And in some areas of the industry (like web development) these types of programs are basically the norm - but this sort of code was already mostly hastily cobbled together slop even before AI.

I also see extreme differences in usefulness of LLM code generation. For anything JS or TS it works great because there's a giant corpus. For C code (or even more obscure languages) especially when using libraries which change all the time my experience is totally different (at least for code generation, bug scanning works quite well).

That's my experience so far anyway, literally "hit and miss" with some areas where there are more hits than misses, and other areas where it's the other way around.

[skydhash]

It’s not a mess as in spaghetti code, which you will find with novice programmers. It’s a mess as in complex and disjointed codebase. Happy path works somewhat, but it crumbles if you run it long enough or encounters an edge case.

You need a very good level of insight to build a codebase to do what is supposed to do, to not do what it shouldn’t do, and to still be comprehensible. Because a software is a system and building system well is what engineering is about.

[pmg101]

Isn't "spaghetti code" a casual term to describe "complex and disjointed codebase"? What's the difference? Have I been using the term wrong all this time?

[skydhash]

Not really. Spaghetti code is usually related to the big ball of mud. Meaning anything you need to understand, you need to go up and down the files while being distracted by irrelevant concerns. There’s no abstraction, no separation of concerns,…

Complex and disjointed codebase is the usual over engineering. The requirement calls for a quick printf, but the implemented module is equivalent to ncurses. And then every calling point turns into a complex ceremony. And most of them are subtly wrong in some way.

Those are two distinct mode of failures. The first was the sin of novice programmers before LLM. The second was the sin of not so novice programmers before LLMs, at least until the pain of living with their mistakes taught them better.

But with agentic tooling, the output is often the second. I believe mostly because no one publish the simple generic version of their codebase. Instead it’s the one that is aligned with their specs, simple in that context, but too complex in any other.

[pmg101]

Thanks. I do think I see the distinction you're making.

[tim-projects]

Says more about you buddy. I didn't say I'm anti AI. I said run the prompt. If you had run the prompt then you would now have better code. Instead you decided to spark off some dumb debate that has nothing to do with what I said in my comment.

Kind of ironic no? To be the harbinger of the very thing you object to.

[eudamoniac]

I've posted this before and it's still what I think:

I'm convinced that the polarization is that one's impression of AI has a direct 1:1 mapping with one's previous level of skill and sensitivity to quality. Most people are by definition average and they are impressed.

Is there anyone in the industry noted for their skill, quality, and taste, e.g. Jonathon Blow, Casey Muratori, who is impressed and thinks the AI is really good? I haven't seen any. In my personal circle, the best devs I know are either micromanaging or shunning AI; none of them think the agents are capable or really good. The mediocre devs I know are largely on board. This applies both online and off.

Couple this with the fact that no AI focused project has come out, not a single one, that meets a high quality bar with nontrivial complexity.

I am an AI quality sceptic. They can be useful if you don't care for quality, but I never don't care for quality. I live for quality.

[bakugo]

> The whole thing will come crumbling down any minute

It is crumbling down as we speak. There's more software being pumped out than ever, but it's also buggier, more bloated, and just generally worse than ever.

A few days ago I had to revert to an older version of an open source project I use regularly, because the author turned full vibe coder and has pushed hundreds of AI generated commits in just a few weeks, at least one of which completely broke an important feature I relied on.

Whether or not you consider putting out software that technically meets the minimum requirements but is a buggy, broken mess to be a success depends entirely on your personal definition of success. Maybe you do, and you are entitled to your opinion. But once again I feel the need to remind people that we humans did not get where we are today by always doing the bare minimum. If everyone throughout history shared this mentality, we'd still be banging rocks in caves.

[brabel]

> also buggier, more bloated, and just generally worse than ever.

Citation needed. I have seen enormous improvements in bug fixes across our projects and many other companies (one HN headline says Google fixed more bugs in July than in previous 2 years). I would’ve expected even more given how easy it is to fix things now. LLMs can understand code better than anyone I’ve met, judging by their power to explain something and then optionally go ahead and fix stuff. People used to ask me about our project as I am the most senior developer and been working on our product for 12 years. But when I want to know something for sure I myself ask Claude. It is amazingly good.

[twister2920]

> Citation needed.

that's not where the burden of proof lies buddy. you prove to us where AI has caused a spike in quality, functioning software

[bakugo]

> Citation needed.

Citation is me and my own personal experiences with vibe coded software. If you have examples of stable, efficient, polished software created by AI, I'll gladly try them out.

Regardless, your comment is a perfect exhibit of the problem. You are seeing 100 AI generated PRs merged in a day and interpreting that as 100 improvements to the user experience, which probably sounds great when you tell it to your manager before asking for a promotion, but doesn't actually translate to reality and never has. Before AI, it was generally agreed upon that measuring quality of work by lines of code written, PRs merged, etc. made no sense, yet that wisdom seems to have been completely forgotten in favor of just watching the line go up.

How do you even know the improvements are "enormous" if you don't understand the code? Do you just occasionally glance at your terminal to see Claude going on about how it just found the smoking gun and made a critical fix to a load-bearing seam and go "whoa, that sounds important"? If those improvements and bug fixes really were so important, why were they not considered worth the effort to make before AI came along, or was your team simply incapable of making them before?

If you're the most senior developer working on the same codebase for 12 years, yet proudly state that an LLM understands it better than you do, I can't help but interpret that as an admission of incompetence, so letting AI do your work for you may really be an improvement, but it shouldn't be.

[ModernMech]

What was the project? I'd like to examine the before and after.

[bakugo]

Adminer. The code was never great, but it worked, and I have not been able to find a suitable replacement that isn't slow and bloated.

The specific functionality that was broken was the plugin that allows picking from a fixed list of servers instead of entering the address every time.

[IAmGraydon]

Interesting. What product and/or project have you successfully launched with it?

[Bootvis]

I have solo built this:

https://github.com/bobjansen/Ibex

Ibex is a usable DSL for table manipulation that is quite fast

https://bobjansen.github.io/Ibex/benchmarks.html

I'm still managing to add more features, such as multithreading. The LLM's do make mistakes but they have gotten better and they tend to self correct these days. I'm also comparing results to mature products such as Polars and DuckDB and the results match.

[twister2920]

how many users do you have?

[Bootvis]

Zero. I don’t think that implies the product is bad though. I haven’t marketed it much yet and I’m not great at it either. Even if it is marketed well, integrating the tool into existing workflows would take some effort. This has nothing to do with the suitability of LLMs though.

[themgt]

Blub is good enough for him, because he thinks in Blub.

[dakolli]

They aren't that great, they can be helpful, but they mostly make lazy people seem and feel less lazy. Maybe that's you? Maybe that's me as well, idk, but when you look close, these things can easily cause more problems than they solve (more often than not ime). No I'm not holding it wrong, I got more token consumption than most here probably, I have access to all the models, unlimited, I'm not that impressed.

I've personally seen the consequences of someone who naively thought these tools could help them execute on their ideas and I wouldn't be surprised if they end up in prison because of it.

[bigstrat2003]

> Are these people cooked? We are years into this and they haven't been able to figure it out?

There's nothing to figure out. It's not rocket surgery to prompt the clanker, anyone who can write half decent technical documentation can to so. The things simply do not produce stuff at a decent level of quality.

[tovej]

What people? What is the definition of success?

You can't appeal to something happening in the real world if you don't say what you mean.

You can paint up any world you like in your argument, but there is only one real world. Please tell us who it is, in the real world, that the hacker news crowd is gaslighting.

[cyanydeez]

you have to remember, 70 million people voted for the current US regime and many of them billionaires.

The ability to delude ones self and not pay attention to the "facts on the code" is moumentally built into the human condition.

Everyones experience is some unique snowflake of just how easy their mental gymnastics are.

[philipallstar]

> you have to remember, 70 million people voted for the current US regime and many of them billionaires.

Many of the 70 million Republican voters were billionaires? What are you talking about?

[cyanydeez]

"of the american billionaires who voted, many of them voted republican"

I get it, but it's not _that hard_ to parse.

[smolder]

> Are these people cooked?

Yes. This is a weird kind of denial about the LLMs that cost nation-state levels of energy doing human nerd work.

They aren't perfect at all, but there's very little reason to be anything but a prompt engineer and proof reader anymore.

[techblueberry]

Prompt engineering died like 2 years ago.

[goatlover]

This is an absurdly overgeneralized take for a wide and diverse field, and it's also false. Of course for some people it's true, but hardly for everyone. I doubt it's even the majority of all programmers.

[gmueckl]

The naysayers aren't cooked at all. They have really, really good points that align with my own experience so perfectly It's eerie. Vine coding has a price tag attached that will have to be paid with interest eventually. The trouble is that it takes a whole damn lot of experience see these signs early.

That doesn't mean that LLM slop code can't be useful. But you just can't build a whole business entirely by surfing a wave of vibe coded slop. That wave will break over you and drown you eventually.

[bakugo]

Prompt engineering? Proof reading? That's so last month, who even does that anymore? What a nerd.

All this condescending drivel about how "anyone who still writes code by hand is a loser, but not me, I'm a prompt engineer because I'm smart and forward-thinking" is so painfully tone-deaf, as if this "prompt engineering" has any lasting value and isn't just a momentary stepping stone towards a techno-feudalist future where every tech company consists of a small group of very wealthy executives, typing what they want in plain words into a text box and having the AI do everything for them, with no need for anyone below them.