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

Nobody knows how to build with AI yet

worksonmymachine.substack.com

Post reply on HN
Print view View original

421–430 of 432 posts

Re: Nobody knows how to build with AI yet

#421

Post by yencabulator »

pluc wrote: post #383
tacker2000 wrote: post #328

Earlier quoted context omitted.

> There's no pride, no joy, just productivity. I think it’s more nuanced than that. Not every project one does will be or should be considered art or a source of joy and pride. The boring CRUD apps that put the “bread on the table” are just that, a means to an end, they will not be your main source of pride or fulfillment. But somewhere in between there will be projects where you can put all your heart in and turn of…

Well, that's a good example. Why would you get a DJ when you can say "Siri, play Weddings Classics"? There's no humanity involved, no skills to read the room or cater to audiences. So you get a DJ; what if your DJ thinks his job or your event is boring and generates the same playlist you could have done yourself? You need passion, you need interest, you need to be involved. Otherwise every job becomes tedious, and hu…

One thing that differentiates a (good) DJ from a playlist is that a DJ will react to the crowd. That'll influence song selection, mixing, live looping, and so on.

Which means clearly we need to feed video of the dancefloor to a vision model and output MIDI tokens!

Re: Nobody knows how to build with AI yet

#422

Post by Anamon »

tcldr wrote: post #345
myblake wrote: post #299

Earlier quoted context omitted.

> There's no pride, no joy, just productivity. This is exactly what bothers me about the present moment. Not that the pride of craftsmanship is everything, but dialing it down to zero with extreme pressure to stay that way is a bit sad. But we’ve clearly gone through this with other mediums before, perhaps someday people will appreciate hand written code the way we appreciate hand carved wood. Or perhaps we were all…

I think the audience who can appreciate handcrafted code will be vastly smaller than the audience who appreciates hand carved wood.

What about the audience which appreciates software that actually works without one billion subtle bugs and devastating security issues, and which also can be built upon and extended?

Re: Nobody knows how to build with AI yet

#423

Post by Anamon »

mmcconnell1618 wrote: post #186
ivanjermakov wrote: post #32

Earlier quoted context omitted.

> Search has just become Chat I think chat-like LLM interfacing is not the most efficient way. There has to be a smarter way.

English and other languages come with lots of ambiguity and assumptions. A significant benefit of programming languages is they have explicit rules for how they will be converted into a running program. An LLM can take many paths from the same starting prompt and deliver vastly different output.

A nice article on this was in the January issue of the Communications of the ACM [1]. With a reference to a piece by Dijkstra predicting that this is never going to be effective, back in 1979.

Being able to write code in a programming language is a feature, not a flaw. If we had always had to program in natural language, the precision and unambiguity of programming languages would be an eagerly welcomed revolution.

[1] https://cacm.acm.org/opinion/on-program-synthesis-and-large-...

Re: Nobody knows how to build with AI yet

#424

Post by Anamon »

brap wrote: post #323

I don’t know which models you guys are working with, but all of the SOTA ones I’ve tried in different configurations ended up producing mostly garbage. I find myself having to spend more time guiding the model in the right direction and fixing its mistakes than I would’ve spent building it all myself. Every time I read one of these stories I feel like maybe you guys have models from 2035, because the ones we have tod…

It works for people who don't really look at the code, or try to understand what the LLM spat out. It's right there in this article: if it compiles and runs, just commit and push.

They usually don't write this kind of blog post anymore after they tried to vibe-code an extension or modification onto their vibe-coded application.

There was a good one on HN a while ago, though, of someone who decided to actually take a look at their vibe-coded codebase, finding an unbelievable mess of inconsistencies and redundancies.

People who just want to put something together and hate programming can be fine with this. Anyone who wants to build something clean that actually lasts, won't.

Re: Nobody knows how to build with AI yet

#426

Post by tcldr »

Anamon wrote: post #422
tcldr wrote: post #345

Earlier quoted context omitted.

I think the audience who can appreciate handcrafted code will be vastly smaller than the audience who appreciates hand carved wood.

What about the audience which appreciates software that actually works without one billion subtle bugs and devastating security issues, and which also can be built upon and extended?

Maybe not possible with today's SOTA AI but I have no doubt it's within reach.

Re: Nobody knows how to build with AI yet

#429

Post by bopbopbop7 »

Zacharias030 wrote: post #420
bopbopbop7 wrote: post #403

Earlier quoted context omitted.

It’s been two more years for the past three years. What’s going to happen in those two years? Are we going to find 10 more internets to train on?

two years ago it didn’t do anything that made me want to use it during work. The data wall idea of needing 10 more internets is like the peak oil theory of the 2000s, imho. I don’t think 10 more internets are required, and all those tasks like proper security and i18n, none of these require creativity of any kind. These are ideal candidates for LLMs to solve quite soon. But to talk about the data wall more: Current S…

The current training data is not the full internet worth because most of the internet is garbage data. It is removed for a reason. But sure, those 200T tokens of Facebook user data is the answer to smarter LLMs.

Re: Nobody knows how to build with AI yet

#430

Post by gngoo »

dlisboa wrote: post #357
gngoo wrote: post #311

To me it feels like I’m in the camp of people who has already figured it out. And I have now learned the hard way that it’s almost impossible to teach others (I organized several meetups on the topic). The ability seems like pure magic. I know that there are others who have it very easy now building even complex software with AI and delivering project after project to clients at record speed at no less of quality as…

Just record yourself doing it and post online. If the projects are indeed complex and you’ve found a way to be 20x more productive people will learn from it. The problem is not having any evidence or basis on which to compare claims. Alchemists claimed for centuries to synthesize gold, if they only had video we could’ve ruled that out fast.

For which reason exactly? Everyone will catch up to this eventually.
Post reply on HN