Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
dartos
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
21 ms
·
301.
▲
by
dartos
2y ago
How would ensuring the IRS is able to correctly do its job make more money for the ultra wealthy?
302.
▲
by
dartos
2y ago
Also from that article > IBM is giving its mainframe customers a tool infused with generative AI to translate COBOL code to Java Hard to wholesale believe claims when they are literally the ones selling that capability. It probably works
303.
▲
by
dartos
2y ago
AI is okay at it. It is one of the things they are best at. Tho it works best with small chunks of well isolated code… I doubt it’d be very successful with the IRS’s ball of COBOL, especially considering how little cobol code is available f
304.
▲
by
dartos
2y ago
RWKV. It’s a Linux foundation project as well.
305.
▲
by
dartos
2y ago
Ah… finance reporting.
306.
▲
by
dartos
2y ago
> I'd really like to know how they see themselves. Extreme cheap voice work for ad campaigns and asset flips. Countless YouTube ads use AI voice cloning and I’m pretty sure elevenlabs is the biggest player in that space.
307.
▲
by
dartos
2y ago
The issue isn’t code quality, it’s intentionality. There’s nothing wrong with the code on this site, from what I can tell. It’s just obvious that nobody put any care towards making it (not everything needs to be a work of art, mind you).
308.
▲
by
dartos
2y ago
> On search listings, make clicking anywhere in the rounded-rectangle for the search hit go to the details view These are the kinds of small issues which compound into a shitty product that you get from wholesale relying on AI.
309.
▲
by
dartos
2y ago
Is there anything about it that screams cursor to you? Or was that just a throwaway statement EDIT: nvm literally says at the bottom of the page
310.
▲
by
dartos
2y ago
> people who don't put any time into studying Degrees don’t mean that either. I’ve been studying textbooks and papers on real time rendering techniques for the past 4 or so years. I think one could learn something from listening to
311.
▲
by
dartos
2y ago
> Sam Altman (a high-school graduate) “People who didn’t pass a test aren’t worth listening to” I have no love for Altman, but this is kind of elitism is insulting.
312.
▲
by
dartos
2y ago
> At the invention of the printing press, there were people with this same energy. Obviously those people were wrong. What energy? What were they wrong about? The luddite type groups have historically been correct in their fears. It just
313.
▲
by
dartos
2y ago
> I have a solid understanding of AI > and it will destroy the value of most human work From my view, as someone studying and working with ML and AI since 2017, these two statements seem in conflict. Either you’re overestimating what
314.
▲
by
dartos
2y ago
Okay, but this obviously isn’t the same as that. C compilers emit assembly as a direct result of the code written. The same compiler given the same code will output the same assembly. This would be true for LLMs if you set the temperature t
315.
▲
by
dartos
2y ago
No it isn’t. For so many reasons, it isn’t. What a wildly incorrect take. Here’s the most obvious reason why that isn’t the case: There is a direct cause effect relationship with the code you write and the resulting program. That is not so
316.
▲
by
dartos
2y ago
Are they mostly the same 5k packages as 2017? That seems to be the crux of it.
317.
▲
by
dartos
2y ago
Docker pre-kube was really great, imo. Sticking a bunch of docker containers on a box behind a reverse proxy or load balancer felt like the sweet spot of complexity and scalability for most apps.
318.
▲
by
dartos
2y ago
Marketing is really good at their job. That coupled with new money and retail investors being thinking they’re in a gold rush and you get the environment we’re in.
319.
▲
by
dartos
2y ago
I don’t think “prompt engineering” will remain its own field. It’s just modern SEO and SEO will eat it, eventually. Prompt engineering as a service makes more sense than having on-staff people anyway, since your prompt’s effectiveness can c
320.
▲
by
dartos
2y ago
The secret sauce is using a big LLM to generate many “responses” for some given intents, then use phi or something cheap to do intent detection and pick one of the pre made responses. You can generate 100s of responses per intent, so the us
321.
▲
by
dartos
2y ago
As a programmer (not a designer) I like tailwind bc I can keep my styles and markup in the same file. It’s a nice workflow. Idk what the “tailwind look” is. Tailwind is basically a bunch of css classes with a single rule per. I bet lots of
322.
▲
by
dartos
2y ago
Why is this clickbait on the front page?
323.
▲
by
dartos
2y ago
The real reason is because it’s faster (or was in 2014) to diff 2 versions of an optimized in-memory tree than it is to use the dom API to diff live html (where any change might trigger the browser’s layout and rendering pipeline) That’s wh
324.
▲
by
dartos
2y ago
Tailwind and bootstrap are near opposite approaches to frontend design … What’re you on about?
325.
▲
by
dartos
2y ago
Because Meta called llama that shortly after it got leaked and it stuck. The AI crowd doesn’t care much for licenses anyway.
326.
▲
by
dartos
2y ago
> no internet and all, which means there wont be a market Please tell me that you know that markets have existed and exist without the internet. In those markets, gold is and was valued… > Historically gold was valuable because it was
327.
▲
by
dartos
2y ago
Metro had it right
328.
▲
by
dartos
2y ago
> “the value would be mostly erased” implies that the price goes down to 0 No, it means the value would be much much closer to 0 than it is now. That’s why I said “mostly erased” and not “entirely erased” > Gold wouldnt survive your d
329.
▲
by
dartos
2y ago
Procrastination may be useful. [1] (first thing from a Google search of useful procrastination) At the very least it’s not universally a bad thing. Instead of ignoring that comment, take it to mean “users may want a feature to allocate an a
330.
▲
by
dartos
2y ago
At that point, wouldn’t the man pages be better than asking 4o? It was already wrong once.
More ›