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

111–120 of 432 posts

Re: Nobody knows how to build with AI yet

#111

Post by jon-wood »

verisimilidude wrote: post #86
lordnacho wrote: post #33

Earlier quoted context omitted.

But you also can't not swim with the tide. If you drove a horse-buggy 100 years ago, it was probably worth your while to keep your eye on whether motor-cars went anywhere. I was super skeptical about a year ago. Copilot was making nice predictions, that was it. This agent stuff is truly impressive.

AI's superpower is doing mediocre work at high speed. That's okay. Great, even. There's lots of mediocre work to do. And mediocre still clears below average. But! There's still room for expertise. And this is where I disagree about swimming with the tide. There will be those who are uninterested in using the AI. They will struggle. They will hone their craft. They will have muscle memory for the tasks everyone else f…

My ongoing concern is that most of us probably got to being able to do good work via several years of doing mediocre work. We put in the hours and along the way learned what good looks like, and various patterns that allow us to see the path to solving a given problem.

What does the next generation do when we’ve automated away that work? How do they learn to recognise what good looks like, and when their LLM has got stuck on a dead end and is just spewing out nonsense?

Re: Nobody knows how to build with AI yet

#112

Post by asadotzler »

I seem to have missed the part where he successfully prompted for security, internationalizability, localizability, accessibility, usability, etc., etc.

This is a core problem with amateurs pretending to be software producers. There are others, but this one is fundamental to acceptable commercial software and will absolutely derail vibe coded products from widespread adoption.

And if you think these aspects of quality software are easily reduced to prompts, you've probably never done serious work in those spaces.

Re: Nobody knows how to build with AI yet

#113

Post by ModernMech »

wilkystyle wrote: post #79
ModernMech wrote: post #54

Earlier quoted context omitted.

You prompt. You go live your life. You come back to ten thousand lines of code. You spend 5 minutes reading. One sentence of feedback. Another ten thousand lines appear while you're making lunch. Yeah, it strikes me the author writes prose the same way they're generating code. 20k lines? That's enough code for a whole compiler or an operating system kernel. I'd love to see what those 20k lines actually do -- notably,…

I definitely did a double take when I got to this section. I am neither an AI optimist nor an AI pessimist (probably slightly on the optimistic side of the midpoint) but this sounds insane to me for any software that people might truly depend on. Five minutes of review for 10,000 lines, happening multiple times per day?

I did some digging, and this seems to be the project the author is using AI to work on: https://github.com/sublayerapp/sublayer

You look at the PRs... there are 786(!) AI generated pull requests and an associated AI generated code review for each one. Each PR is about ~20-100 lines of Ruby (including comments) that implements an "action" for the sublayer system as a Ruby class. So probably something that could be handled by a macro expander. Or at least it's AI used as a fancy macro expander.

But yeah, there's about 20k lines of code right there easily. Although, because it's Ruby, it's not (much) of an exaggeration to say ~50% of the generated lines are a single "end" keyword.

The author is someone who before AI, would publish ~300 commits a year to Github. This year they are on track for 3000 commits using AI. But the result seems to be that PRs are accumulating in their repo, implementing hundreds of features. I'm wondering why the PRs are accumulating and not getting merged if the code is good? Is the bottleneck now review? What would happen if AI took over PR merging as well as PR creation?

Re: Nobody knows how to build with AI yet

#114

Post by vitaflo »

karel-3d wrote: post #24

Reading articles like this feels like being in a different reality. I don't work like this, I don't want to work like this and maybe most importantly I don't want to work with somebody who works like this. Also I am scared that any library that I am using through the myriad of dependencies is written like this. On the other hand... if I look at this as some alternate universe where I don't need to directly or indirec…

>I don't want to work with somebody who works like this.

You will most likely get your wish but not in the way you want. In a few years when this is fully matured there will be little reason to hire devs with their inflated salaries (especially in the US) when all you need is someone with some technical know-how and a keen eye on how to work with AI agents. There will be plenty of those people all over the globe who will demand much less than you will.

Hate to break it to you but this is the future of writing software and will be a reckoning for the entire software industry and the inflated salaries it contains. It won't happen overnight but it'll happen sooner than many devs are willing to admit.

Re: Nobody knows how to build with AI yet

#116

Post by rhdunn »

zwnow wrote: post #18
lordnacho wrote: post #15

I'm loving the new programming. I don't know where it goes either, but I like it for now. I'm actually producing code right this moment, where I would normally just relax and do something else. Instead, I'm relaxing and coding. It's great for a senior guy who has been in the business for a long time. Most of my edits nowadays are tedious. If I look at the code and decide I used the wrong pattern originally, I have to…

So you are relaxing and the AI is coding? Neat! Way to replace yourself, hope you won't cry after your job once it is gone.

I'm using AI assistants as an interactive search and coding assistant. I'm still driving the development and implementing the code.

Where I use it for is:

1. Remembering what something is called -- in my case the bootstrap pills class -- so I could locate it in the bootstrap docs. Google search didn't help as I couldn't recall the right name to enter into it. For the AI I described what I wanted to do and it gave the answer.

2. Working with a language/framework that I'm familiar with but don't know the specifics in what I'm trying to do. For example:

- In C#/.NET 8.0 how do I parse a JSON string?

- I have a C# application where I'm using `JsonSerializer.Deserialize` to convert a JSON string to a `record` class. The issue is that the names of the variables are capitalized -- e.g. `record Lorem(int Ipsum)` -- but the fields in the JSON are lowercase -- e.g. `{"ipsum": 123}`. How do I map the JSON fields to record properties?

- In C# how do I convert a `JsonNode` to a `JsonElement`?

3. Understanding specific exceptions and how to solve them.

In each case I'm describing things in general terms, not "here's the code, please fix it" or "write the entire code for me". I'm doing the work of applying the answers to the code I'm working on.

Re: Nobody knows how to build with AI yet

#117

Post by mrandish »

Quitschquat wrote: post #35
Flatcircle wrote: post #22

My theory on AI is it's the next iteration of google search, a better more conversational, base layer over all the information that exists on the internet. Of course some people will lose jobs just like what happened to several industries when search became ubiquitous. (newspapers, phone books, encyclopedias, travel agents) But IMHO this isn't the existential crisis people think it is. It's just a tool. Smart, clever…

Google doesn’t have to change search. It already returns AI generated crap before anything useful.

Append -ai to your query to omit AI results.

Re: Nobody knows how to build with AI yet

#118

Post by beefnugs »

lordnacho wrote: post #33
karel-3d wrote: post #24

Reading articles like this feels like being in a different reality. I don't work like this, I don't want to work like this and maybe most importantly I don't want to work with somebody who works like this. Also I am scared that any library that I am using through the myriad of dependencies is written like this. On the other hand... if I look at this as some alternate universe where I don't need to directly or indirec…

But you also can't not swim with the tide. If you drove a horse-buggy 100 years ago, it was probably worth your while to keep your eye on whether motor-cars went anywhere. I was super skeptical about a year ago. Copilot was making nice predictions, that was it. This agent stuff is truly impressive.

This doesn't make aaaaany sense: IF this actually worked, then why would all the biggest companies in the world be firing people? They would be forcing them all to DO THE TIDE and multiple their 10 billion dollar dominance to 100 billion dollar or more dominance.

The truth is something like: for this to work, there is huge requirements in tooling/infrastructure/security/simulation/refinement/optimization/cost-saving that just could never be figured out by the big companies. So they are just like... well lets trick as many investors and plebs to try to use this as possible, maybe one of them will come up with some breakthrough we can steal

Re: Nobody knows how to build with AI yet

#119

Post by recursive »

gabrieledarrigo wrote: post #94
karel-3d wrote: post #24

Reading articles like this feels like being in a different reality. I don't work like this, I don't want to work like this and maybe most importantly I don't want to work with somebody who works like this. Also I am scared that any library that I am using through the myriad of dependencies is written like this. On the other hand... if I look at this as some alternate universe where I don't need to directly or indirec…

I know, it's scary. But I guess it's the direction we are aiming for.

Just to clarify, I'm not a member of that "we".

Re: Nobody knows how to build with AI yet

#120

Post by mnky9800n »

prinny_ wrote: post #61
mnky9800n wrote: post #49

Earlier quoted context omitted.

I think the agent stuff is impressive because we are giving the AI scaffold and tools and things to do. And that is why it is impressive because it has some directive. But it is obvious if you don't give it good directives it doesn't know what to do. So for me, I think a lot of jobs will be making agents do things, but a lot won't. i think its really strange that people are all so against all this stuff. it's cool ne…

A lot of people join this profession because they like building stuff. They enjoy thinking about a problem and coming up with a solution and then implementing and testing it. Prompting is not the same thing and it doesn't scratch the same itch and at the end of the day it's important to enjoy your job, not only be efficient at it. I have heard the take that "writing code is not what makes you an engineer, solving pro…

Yeah but I write the code that is interesting to solve and let the LLM solve the problems that are not so important. Like making yet another webscraper tool is not the most exciting part of the process when you are trying to make some kind of real time inference tool for what people post on the internet.
Post reply on HN