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

211–220 of 1001 posts

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

#211

Post by simonw »

ruszki wrote: post #209
CraftingLinks wrote: post #187

Earlier quoted context omitted.

Keep believing. To the bitter end. For such human slop codebases AI slop additions will do equally fine. Add good testing and the code might even improve over the garbage that came before.

Generating also the tests happens a little bit too often for any kind of improvement. simonw posted here a generated “something” the other day, which he didn’t know whether it’s really working or not, but he was happy that his generated, completely unchecked tests are green, and yet some other root commenter here praises him. It needs a lot of work to not be skeptical, when when I try it, it generates shit, especiall…

Which thing was that?

(If you need help finding it try visiting https://tools.simonwillison.net/hn-comments-for-user and searching for simonw - you can then search my 1,000 most recent comments in one place.)

If my tests are green then it tells me a LOT about what the software is capable of, even if I haven't reviewed every line of the implementation.

The next step is to actually start using it for real problems. That should very quickly shake out any significant or minor issues that sneaked past the automated tests.

I've started thinking about this by comparing it to work I've done within larger companies. My team would make use of code written by other teams without reviewing everything those other teams had written. If their tests passed we would build against their stuff, and if their stuff turned out not to work we would let them know or help debug and fix it ourselves.

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

#212

Post by auggierose »

babarock wrote: post #61
dom96 wrote: post #2

> As a programmer, I want to write more open source than ever, now. I want to write less, just knowing that LLM models are going to be trained on my code is making me feel more strongly than ever that my open source contributions will simply be stolen. Am I wrong to feel this? Is anyone else concerned about this? We've already seen some pretty strong evidence of this with Tailwind.

I don't know if you're "wrong", but I do feel differently about this. I've written a ton of open source code and I never cared what people do with it, both "good" or "bad". I only want my code to be "useful". Not just to the people I agree with, but to anyone who needs to use a computer. Of course, I'd rather people use my code to feed the poor than build weapons, but it's just a preference. My conviction is that my…

And in that case, use MIT license or something like that for your code, and all is good. If I use AGPL, on the other hand, AI companies should not be allowed to train on that and then use the result of that training while ignoring the license.

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

#213

Post by falloutx »

galdauts wrote: post #24

I feel like the use of the term "anti-AI hype" is not really fully explored here. Even limiting myself to tech-related applications - I'm frankly sick of companies trying to shove half-baked "AI features" down my throat, and the enshittification of services that ensues. That has little to do with using LLMs as coding assistants, and yet I think it is still an essential part of the "anti-AI hype".

The dreaded summarize feature, its in places you wouldn't expect, and not to mention the whole lets record every meeting and then summarize it for leaders. Big Brother in work is back and its even more powerful.

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

#214

Post by senko »

zahlman wrote: post #146
senko wrote: post #57

The anti-AI hype, in the context of software development, seems to focus on a few things: > AI code is slop, therefore you shouldn't use it You should learn how to responsibly use it as a tool, not a replacement for you. This can be done, people are doing it, people like Salvatore (antirez), Mitchell (of Terraform/Ghostty fame), Simon (swillison) and many others are publicly talking about it. > AI can't code XYZ It's…

> at least three quarters are boring boilerplate. And making that part interesting is one of the worst things you can do in software development! That path lies resume-driven development, architecture astronautics, abusing design patterns du jour, and other sins that will make code maintenance on that thing a nightmare! You apparently see "making the boilerplate interesting" as doing a bunch of overengineering. Stran…

I agree with what you said, but I think we might be talking about slightly different things. Let me give a few examples in an attempt to better explain myself:

A company I worked with a while ago had a microservices architecture, and have decided to not use one of a few standard API serialization/deserialization options, but write their own, because was going to be more performant, easier to maintain, better fit for their use case. A few years on, after having grown organically to support all the edge cases, it's more convoluted, slower, and buggy than if they went with the boring option that ostensibly had "a bit more boilerplate" from the start.

A second example is from a friend, whose coworker decided to write a backend-agnostic, purpose-agnostic, data-agnostic message broker/routing library. They spent a few months of this, delivered a beautifully architected solution in a few dozen k lines of code. The problem is the solution solves many problems the company didnt and wouldn't have, and will be a maintenance drag from then forevermore. Meanwhile, they could have done it in a few hundred lines of code that would be coupled to the problem domain, but still farily decend from most people's point of view.

These two are from real projects. But you can also notice that in general people are often picking a fancy solution over a boring one, ostensibly because it has something "out of the box". The price of the "out of the box"-ness (aside from potential SaaS/infra costs and vendor lock in), is that you now need to adapt your own code to work with the mental model (domain) of the fancy solution.

Or to harp on something trivial, you end up depending on left-pad because writing it yourself was boring.

> fix the system such that it doesn't require boilerplate any more.

I think perhaps I used a more broad meaning for "boilerplate" than you had in mind. If we're talking about boilerplate as enumerating all the exceptions a Java method may raise, or whatever unholy sad thing we have to do in C to use GTK/GObject, then I agree.

But I also meant something more closer to "glue code that isn't the primary carrier of value of the project", or to misuse financial language in this context, the code that's a cost center, not a profit center.

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

#215

Post by jeroenhd »

uyzstvqs wrote: post #88
dom96 wrote: post #2

> As a programmer, I want to write more open source than ever, now. I want to write less, just knowing that LLM models are going to be trained on my code is making me feel more strongly than ever that my open source contributions will simply be stolen. Am I wrong to feel this? Is anyone else concerned about this? We've already seen some pretty strong evidence of this with Tailwind.

Then why open source something in the first place? The entire point is to make it public, for anyone to use however is useful to him or her, and often to publicly collaborate on a project together. If I made something open source, you can train your LLM on it as much as you want. I'm glad my open source work is useful to you.

Plenty of people will gladly give you their hard work for free if you promise you'll return the favor. Or if you promise not to take your work and make others pay for it when they could just get it for free. Basically, help the people that want to embrace the freedoms of open source, but not the ones that are just in it for the free labour. Or at the very, very least, include a little "thank you" note.

AI doesn't hold up its end of the bargain, so if you're in that mindset you now have to decide between going full hands-off like you or not doing any open source work at all.

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

#216

Post by falloutx »

esperent wrote: post #42

> But I'm worried for the folks that will get fired. It is not clear what the dynamic at play will be: will companies try to have more people, and to build more? This is the crux. AI suddenly became good and society hasn't caught on yet. Programmers are a bit ahead of the curve here, being closer to the action of AI. But in a couple of years, if not already, all the other technical and office jobs will be equally aff…

In one of the scenarios programmers get replaced then the progress slows, thus saving jobs of writers, lawyers, marketing, scientists, artists. At this point I am okay with that scenario seeing how programmers have showed no solidarity while every other field has been rejecting AI. Lawyers have even started hiring junior lawyers back and Art industry has basically shoved AI into a bin of irrelevance.

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

#217

Post by jillesvangurp »

dom96 wrote: post #2

> As a programmer, I want to write more open source than ever, now. I want to write less, just knowing that LLM models are going to be trained on my code is making me feel more strongly than ever that my open source contributions will simply be stolen. Am I wrong to feel this? Is anyone else concerned about this? We've already seen some pretty strong evidence of this with Tailwind.

A common intention with opensource is to allow people, and AI tools they use, to reuse, recombine, etc. OSS code in any way they see fit. If that's not what you want, don't open source your work. It's not stealing if you gave it away and effectively told people "do whatever you want". Which is one way licenses such as the MIT license are often characterized.

It's very hard to prevent specific types of usage (like feeding code to an LLM) without throwing out the baby with the bathwater and also preventing all sorts of other valid usages. AGPLv3, which is what antirez and Redis use goes to far IMHO and still doesn't quite get the job done. It doesn't forbid people (or tools) to "look" at the code which is what AI training might be characterized as. That license creates lots of headaches for corporate legal departments. I switched to Valkey for that reason.

I actually prefer using MIT style licenses for my own contributions precisely because I don't want to constrain people or AI usage. Go for it. More power to you if you find my work useful. That's why I provide it for free. I think this is consistent with the original goals of open source developers. They wanted others to be able to use their stuff without having to worry about lawyers.

Anyway, AI progress won't stop because of any of this. As antirez says, that stuff is now part of our lives and it is a huge enabler if you are still interested in solving interesting problems. Which apparently he is. I can echo much of what he says. I've been able to solve larger and larger problems with AI tools. The last year has seen quite a bit of evolution in what is possible.

> Am I wrong to feel this?

I think your feelings are yours. But you might at least examine your own reasoning a bit more critically. Words like theft and stealing are big words. And I think your case for that is just very weak. And when you are coding yourself are you not standing on the shoulders of giants? Is that not theft?

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

#219

Post by simonw »

rvz wrote: post #74
bsaul wrote: post #38

Earlier quoted context omitted.

An ecosystem is being built around AI : Best prompting practices, mcps, skills, IDE integration, how to build a feedback loop so that LLM can test its output alone, plug to the outside world with browser extensions, etc... For now i think people can still catch up quickly, but at the end of 2026 it's probably going to be a different story.

> Best prompting practices, mcps, skills, IDE integration, how to build a feedback loop so that LLM can test its output alone, plug to the outside world with browser extensions, etc... Ah yes, an ecosystem that is fundamentally inherently built on probabilisitic quick sand and even with the "best prompting practices", you still get agents violating the basics of security and committing API keys when they were told no…

One of the skills needed to effectively use AI for code is to know that telling AI "don't commit secrets" is not a reliable strategy.

Design your secrets to include a common prefix, then use deterministic scanning tools like git hooks to prevent then from being checked in.

Or have a git hook that knows which environment variables have secrets in and checks for those.

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

#220

Post by jeroenhd »

simonw wrote: post #160
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.

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…

So far every new AI product and even model update has required me to relearn how to get decent results out of them. I'm honestly kind of sick of having to adjust my work flow every time.

The intuition just doesn't hold. The LLM gets trained and retrained by other LLM users so what works for me suddenly changes when the LLM models refresh.

LLMs have only gotten easier to learn and catch up on over the years. In fact, most LLM companies seem to optimise for getting started quickly over getting good results consistently. There may come a moment when the foundations solidify and not bothering with LLMs may put you behind the curve, but we're not there yet, and with the literally impossible funding and resources OpenAI is claiming they need, it may never come.

Post reply on HN