Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Don't fall into the anti-AI hype

antirez.com

Post reply on HN
Print view View original

591–600 of 1001 posts

Re: Don't fall into the anti-AI hype

#591

Post by onlyrealcuzzo »

totallykvothe wrote: post #569

I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…

I'm not sure if I got in this weird LLM bubble where they give me bad advice to drive engagement, because I can't resist trying to correct them and tell them how absurdly wrong they are.

But it is astounding how terrible they are at debugging non-trivial assembly in my experience.

Anyone else have input here?

Am I in a weird bubble? Or is this just not their forte?

It's truly incredible how thoughtless they can be, so I think I'm in a bubble.

Re: Don't fall into the anti-AI hype

#592

Post by mvkel »

newsoftheday wrote: post #589
mvkel wrote: post #580

Earlier quoted context omitted.

> Other people are just less picky than I am I think this is part of it. When coding style has been established among a team, or within an app, there are a lot of extra hoops to jump through, just to get it to look The Right Way, with no detectable benefit to the user. If you put those choices aside and simply say: does it accomplish the goal per the spec (and is safe and scalable[0]), then you can get away with a lo…

> When coding style has been established It feels like you're diminishing the parent commenter's views, reducing it to the perspective of style. Their comment didn't mention style.

Style = syntax, taste, architecture choices, etc. Things you would see on a 15-year-old Java app.

i.e. not a greenfield project.

Re: Don't fall into the anti-AI hype

#593

Post by __float »

CuriouslyC wrote: post #575
totallykvothe wrote: post #569

I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…

I think it comes down to what you mean by sub par code. If you're talking a mess of bubblesorts and other algorithmic problems, that's probably a prompting issue. If you're talking "I just don't like the style of the code, it looks inelegant" that's not really a prompting issue, models will veer towards common patterns in a way that's hard to avoid with prompts. Think about it like compiler output. Literally nobody c…

There's a huge amount you're missing by boiling down their complaint to "bubble sorts or inelegant code". The architecture of the new code, how it fits into the existing system, whether it makes use of existing utility code (IMO this is a huge downside; LLMs seem to love to rewrite a little helper function 100x over), etc.

These are all important when you consider the long-term viability of a change. If you're working in a greenfield project where requirements are constantly changing and you plan on throwing this away in 3 months, maybe it works out fine. But not everyone is doing that, and I'd estimate most professional SWEs are not doing that, even!

Re: Don't fall into the anti-AI hype

#595

Post by jryle70 »

totallykvothe wrote: post #569

I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…

If you follow antirez's post history, he was a skeptics until maybe a year ago. Why don't you look at his recent commits and judge for yourself. I suppose the majority of his most recent code is relevant for this discussion.

https://github.com/antirez?tab=overview&from=2026-01-01&to=2...

Re: Don't fall into the anti-AI hype

#597

Post by totallykvothe »

jryle70 wrote: post #595
totallykvothe wrote: post #569

I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…

If you follow antirez's post history, he was a skeptics until maybe a year ago. Why don't you look at his recent commits and judge for yourself. I suppose the majority of his most recent code is relevant for this discussion. https://github.com/antirez?tab=overview&from=2026-01-01&to=2...

I don't think I'd be a good judge because I don't have the years of familiarity and expertise in his repos that I do at my job. A lot of the value of me specifically vs an LLM at my job is that I have the tribal knowledge and the LLM does not. We have gotten a lot better at documentation, but I don't think we can _ever_ truly eliminate that factor.

Re: Don't fall into the anti-AI hype

#598

Post by agentultra »

This essay starts with a weak premise, not facts.

There is enough evidence to support claims that AI is a black hole where money gets evaporated.

It’s great that you can delegate some tasks to it now and not have to write all of the code yourself. There is some evidence showing that it doesn’t benefit junior developers nearly as much. If you didn’t generate the specification test that demonstrates the concurrency issue you were trying to solve in Redis but you read the code it generated and understood it then you didn’t need to learn anything. How is a junior developer who has never solved such problems supposed to learn so they can do the same thing?

But worse, UBI and such are the solutions of libertarian oligarchs that dream of a world without people, according to Doctorow and I think he’s right. It seems like the author also wants this? He doesn’t seem to know what will happen to the jobless but we should vote in some one who will start a government program to take care of them. How long until the author is replaced as well?

Lastly… who’s “hyping” anti-AI and what do they gain from making false claims?

I think the real problem for programming is when these companies all collapse and take the rest of the economy down with them… are there going to be enough programmers left to maintain everything? Or will we be sifting though the mountains of tech debt never to see the light of day again?

Re: Don't fall into the anti-AI hype

#599

Post by daxfohl »

totallykvothe wrote: post #569

I don't understand the stance that AI currently is able to automate away non-trivial coding tasks. I've tried this consistently since GPT 3.5 came out, with every single SOTA model up to GPT 5.1 Codex Max and Opus 4.5. Every single time, I get something that works, yes, but then when I start self-reviewing the code, preparing to submit it to coworkers, I end up rewriting about 70% of the thing. So many important deta…

Agreed, but:

There's been a notable jump over the course of the last few months, to where I'd say it's inevitable. For a while I was holding out for them to hit a ceiling where we'd look back and laugh at the idea they'd ever replace human coders. Now, it seems much more like a matter of time.

Ultimately I think over the next two years or so, Anthropic and OpenAI will evolve their product from "coding assistant" to "engineering team replacement", which will include standard tools and frameworks that they each specialize in (vendor lock in, perhaps), but also ways to plug in other tech as well. The idea being, they market directly to the product team, not to engineers who may have specific experience with one language, framework, database, or whatever.

I also think we'll see a revival of monolithic architectures. Right now, services are split up mainly because project/team workflows are also distributed so they can be done in parallel while minimizing conflicts. As AI makes dev cycles faster that will be far less useful, while having a single house for all your logic will be a huge benefit for AI analysis.

Re: Don't fall into the anti-AI hype

#600

Post by xboxnolifes »

systemf_omega wrote: post #8

What I don't understand about this whole "get on board the AI train or get left behind" narrative, what advantage does an early adopter have for AI tools? The way I see it, I can just start using AI once they get good enough for my type of work. Until then I'm continuing to learn instead of letting my brain atrophy.

Early adopters get the advantage of only having to learn a trickle of new things every few weeks instead of everything all at once.

Part of the problem with things that iterate quickly is that iterations tend to reference previous versions. So, you try learning the new hotness (v261), but there are implied references to v254, v239, and v198. Then you realize, v1, v5, v48, v87, v138, v192, and v230 have cute identifiers that you aren't familiar with and are never explained anywhere. New concepts get introduced in v25, v50, v102, and v156 that later became foundational knowledge that is assumed to be understood by the reader and is never explained anywhere.

So, if you feel confident something will be the next hotness, it's usually best to be an early adopter, so you gain your knowledge slowly over years instead of having to cram when you need to pick it up.

Post reply on HN