lionkor

12 ms·

301.
by lionkor 6mo ago
Presumably they aren't falling for their (extremely obvious) "grassroots" marketing, and know, like any good engineer, that LLMs are not the right tool for this. It's easy to just see Bun as a marketing stunt, as well.
302.
by lionkor 6mo ago
No, deterministic tools so far have not fucked up completely by themselves. There's always a bug, maybe a fix, and maybe a regression test.
303.
by lionkor 6mo ago
Yes, but it should be fairly easy to "simply" attack the common technologies that LLMs keep parroting. NextJS, or some Rust tools, or whatever other tools LLMs "love" using, are all great targets. Once millions of comple
304.
by lionkor 6mo ago
It DOES matter, because the claim that LLMs are a layer of abstraction implies that it's somehow more than a random word generator. It does a great job at generating words in the right order, and often, given enough time, datacenter re
305.
by lionkor 6mo ago
With a sufficiently complex prompt and a sufficiently complex codebase, LLMs consistently fail and make mistakes, "forget" parts of the prompt, etc. There's no comparison to be made between this and, for example, a compiler.
306.
Ask HN: Where are SWE's being replaced?
4 points by lionkor 6mo ago|1 comments
307.
by lionkor 6mo ago
Why not? Are there any software-related industries in the West where software engineers are not needed or won't be needed?
308.
by lionkor 6mo ago
LLMs are not another abstraction. ALL OTHER LAYERS you named are fully deterministic, understood, debuggable, etc. You cannot be serious.
309.
by lionkor 6mo ago
> Who cares? I care, because it landed on the front-page of a website that I thought was about engineering and hacking, not prompting a no-code tool and spending 400 bucks on it, to get something nobody else wants.
310.
by lionkor 6mo ago
CHasm[0] is a bit of a joke, right? Did it take a lot of Claude massaging to get it to ignore basic engineering knowledge, like that compilers write better assembly than any other tool or person will (in the general case, which this is)? Th
311.
by lionkor 6mo ago
Software made for one, made by LLMs which regurgitate the average of existing tools, are going to have more security issues, not less.
312.
by lionkor 6mo ago
Are you illiterate? There are step by step instructions, it's accessible too.
313.
Approaching Zero Bugs?(daniel.haxx.se)
4 points by lionkor 6mo ago|0 comments
314.
by lionkor 6mo ago
short commit messages, but you would usually allow details in the details part of the commit message, right? Which is where this goes?
315.
by lionkor 6mo ago
You're likely in an echo chamber! Barely anyone I know uses AI as more than a fallible tool.
316.
by lionkor 6mo ago
Yes, of course. Every single bit of data you send to OpenAI is stored, catalogued, indexed, analayzed, and trained on. It'll simply be a "oops, we miscatalogued and accidentally trained GPT 6 on all data, not just data we got cons
317.
by lionkor 6mo ago
I struggle to find anything on this post that wouldn't be caught by some kind of unit test or manual review, especially when comparing with the GNU source for the coreutils. The whole coreutils rewrite is a terrible idea[1] and clearly
318.
by lionkor 6mo ago
Can't wait to see how the next election(s) turn out--I'm unsure that a properly well funded campaign would skip the opportunity.
319.
by lionkor 6mo ago
Its most likely a "You're totally right, this fix broke production! Let me fix it"
320.
by lionkor 6mo ago
But let's be real, the AI output today is largely complete and utter shit. Like, really, without careful promoting and oversight, it's not fun to read, it's not smart, it's not accurate, it's not thorough, and it&#x
321.
by lionkor 6mo ago
You're right, I was wrong with my example. > The fact that AI is more efficient at this? So what? That does not in any way affect the principle. Well it's not a human, so exceptions for humans shouldn't apply
322.
by lionkor 6mo ago
Weird hypothetical. Not sure who "you folx" is. For me, that would solve my issue with it.
323.
by lionkor 6mo ago
Or not, it doesn't really matter and nobody seems to care. This isn't for the good of humanity, there's no indication that it is
324.
by lionkor 6mo ago
Again, somehow other governments in the world have figured out how to do things for the people , without a company having to lobby for it. For example USB-C ports on all devices, I don't think Xiaomi lobbied with billions and that&#x2
325.
by lionkor 6mo ago
Do the ends always justify the means?
326.
by lionkor 6mo ago
No, of course not. The issue is that code was copied and used, without adhering to the license, as training data. Even before training started, that's not right. That's the issue. All of this would not be possible if laws were adh
327.
by lionkor 6mo ago
The ideal world would be one where, to train on art, you have to buy a license to that art. Sure, for most artists they would maybe put a low price tag, but that isn't the point. The point isn't about money. It's that copies
328.
by lionkor 6mo ago
No, they aren't. Clause 1 of the "modern" BSD license is > 1. Redistributions of source code must retain the above copyright notice, this list of conditions and the following disclaimer. It's a license, not a free giv
329.
by lionkor 6mo ago
The whole point of software licenses is that the copyright holder DOESN'T change. The author retains the rights, and LICENSES them. So, in fact, no rights are given away, they are licensed.
330.
by lionkor 6mo ago
If you look at my comment history (don't, you'll fall over from boredom), you'll see I'm also against that. I've researched and selected specific licenses for all the code I've open sourced, which is quite a lo

More ›