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

701–710 of 1001 posts

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

#701

Post by growt »

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…

It might be 1), being an early adopter doesn’t help much with AI. So much is changing constantly. If you put a good description of your architecture and coding guidelines in the right .md files and work on your prompts the output should be much better. In the other hand your project being legacy code probably also doesn’t help.

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

#702

Post by erk__ »

never_inline wrote: post #676
irthomasthomas wrote: post #366

Earlier quoted context omitted.

In my feed 'AI hype' outnumbers 'anti-AI hype' 5-1. And anti-hype moderates like antirez and simonw are rare. To be a radical in ai is to believe that ai tools offer a modest but growing net positive utility to a modest but growing subset of hackers and professionals

The only AI bloggers who don't have something to sell seems to be simonw, the flask guy, and this redis guy. Any other blog recommendations from HN?

tbh I think it is just a question about time before flask guy has something to sell: https://earendil.com/

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

#703

Post by fmbb »

khafra wrote: post #667
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…

> Non-trivial coding tasks A coding agent just beat every human in the AtCoder Heuristic optimization contest. It also beat the solution that the production team for the contest put together. https://sakana.ai/ahc058/ It's not enterprise-grade software, but it's not a CRUD app with thousands of examples in github, either.

> It's not enterprise-grade software, but it's not a CRUD app with thousands of examples in github, either.

Optimization is a very simple problem though.

Maintaining a random CRUD app from some startup is harder work.

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

#704

Post by Sateeshm »

adityaathalye wrote: post #27

Don't fall into the "Look ma, no hands" hype. Antirez + LLM + CFO = Billion Dollar Redis company, quite plausibly. /However/ ... As for the delta provided by an LLM to Antirez, outside of Redis (and outside of any problem space he is already intimately familiar with), an Apples to Apples comparison would be he trying this on an equally complex codebase he has no idea about. I'll bet... what Antirez can do with Redis…

AI is basically Leonard from Memento. Very capable. Knows how the world works broadly. Can't make new memories. Need context (tattoos, notes, and polaroids). Misunderstandings things.

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

#705

Post by fmbb »

growt wrote: post #701
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…

It might be 1), being an early adopter doesn’t help much with AI. So much is changing constantly. If you put a good description of your architecture and coding guidelines in the right .md files and work on your prompts the output should be much better. In the other hand your project being legacy code probably also doesn’t help.

Do you have links to texts that describe which markdown files, and what to write in them? What is good and what is bad etc.

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

#706

Post by intended »

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…

Thank you for providing data which can actually be used to collate! I strongly suspect that experience is a huge determinant of what utility is seen from LLMs.

It seems that theres more people writing and finishing projets, but not many have reached the point where they have to maintain their code / deal with the tech debt.

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

#707

Post by krautburglar »

p0w3n3d wrote: post #674

> How do I feel, about all the code I wrote that was ingested by LLMs? I feel great to be part of that, because I see this as a continuation of what I tried to do all my life: democratizing code, systems, knowledge. I don't see it as democratic or democratising. TBH the knowledge is stored in three giga companies that used sometimes almost non-lawful (if not lawful?) methods to gain it, scraping it off the gpl projec…

"democratizing" as in "I steal everybody's shit, make most content creators go bankrupt, then put it all in an LLM behind a paywall." Privatization of all human knowledge--past, present, & future. They own both parties, so it's not like anyone is going to vote their way out of this one--unless one considers guillotines a form of voting.

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

#708

Post by beaker52 »

tymscar wrote: post #531
simonw wrote: post #160

Earlier quoted context omitted.

This is a pretty common position: "I don't worry about getting left behind - it will only take a few weeks to catch up again". I don't think that's true. I'm really good at getting great results out of coding agents and LLMs. I've also been using LLMs for code on an almost daily basis since ChatGPT's release on November 30th 2022. That's more than three years ago now. Meanwhile I see a constant flow of complaints fro…

I think you are right in saying that there is some deep intuition that takes months, if not years, to hone about current models, however, the intuition some who did nothing but talk and use LLMs nonstop two years ago would be just as good today as someone who started from scratch, if not worse because of antipatterns that don’t apply anymore, such as always starting a new chat and never using a CLI because of context…

> because of antipatterns that don’t apply anymore, such as always starting a new chat

I’m keen to understand your reasoning on this. I don’t agree, but maybe I’m just stuck with old practices, so help me?

What’s your justification as to why starting a new chat is an antipattern?

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

#709

Post by KronisLV »

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 end up rewriting about 70% of the thing.

Doesn't match my experience, that figure is closer to about 20-40% to me, though a lot of those changes I want are possible by just further prompting OR turning to a different model, or adding some automated checks that promptly fail and the AI can do a few more loops of fixes.

> Other people are just less picky than I am, or they have a less thorough review culture that lets subpar code slide more often.

This is also likely, or you are just doing stuff that is worse represented in the training data, or working on novel things where the output isn't as good. But I'm leaning towards people just being picky about what they view as "good code" (or underspecifying how the AI is supposed to output it) at least roughly since Sonnet 4, since with some people I work with it's just endless and oftentimes meaningless discussions and bikeshedding when in code review.

You can always be like: "This here pattern in these 20 files is Good Code™, use the same collection of approaches and code style when working on this refactoring/new feature."

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

#710

Post by lmeyerov »

growt wrote: post #701
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…

It might be 1), being an early adopter doesn’t help much with AI. So much is changing constantly. If you put a good description of your architecture and coding guidelines in the right .md files and work on your prompts the output should be much better. In the other hand your project being legacy code probably also doesn’t help.

We find across our team different people are able to use these things at different levels. Unsurprisingly, more senior coders with both more experience in general and more experience in ai coding are able to do more with ai and get more ambitious things done more quickly.

A bummer is that we have a genai team (louie.ai) and a gpu/viz/graph analytics team (graphistry), and those who have spent the last 2-3 years doing genai daily have a higher uptake rate here than those who aren't. I wouldn't say team 1 is better than team 2 in general: these are tools, and different people have different engineering skill and ai coding skill, including different amounts of time doing both.

What was a revelation for me personally was taking 1-2mo early in claude code's release was to go full cold turkey on manual coding, similar to getting immersed in a foreign language. That forced eliminating a lot of bad habits wrt effective ai coding both personally and in state of our repo tooling. Since then, it's been steady work to accelerate and smooth that loop, eg, moving from vibe coding/engineering to now more eval-driven ai coding loops: https://media.ccc.de/v/39c3-breaking-bots-cheating-at-blue-t... . That takes a LOT of buildout.

Post reply on HN