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

AI and Home-Cooked Software

mrkaran.dev

Post reply on HN
Print view View original

11–20 of 44 posts

Re: AI and Home-Cooked Software

#11

Post by djmips »

bitwize wrote: post #7

Ah, another "Now that we have AI, people can do [thing people could do for decades]" article. If there was something you wanted a computer to do, that it did not yet do, you programmed it. And if you didn't know how, you learned. BASIC was always there. But the industry as a whole moved away from the idea that end users are to program computers sometime in the 80s or 90s (the glorious point and click future was not e…

I feel like you've never actually tried to make tool with Claude Code or similar because BASIC is not it - that's viewing the past with rose coloured glasses. However I understand your central thesis that we could have actually put effort into making something that average folk could use to effectively leverage computers in a way that requires 'code'. But you know we have tried - we have Scratch - we have all of the node graph spaghetti in Unreal Engine and others - I am a programmer but I finally sat down and went through the process of making a working finished tool in a language I'm unfamliar with and using Claude Code and it went really well. And if folks like Ben Krasnow of Applied Science channel are using AI coding tools that they would formally take them 3 to 5 x longer to struggle through unfamiliarity then practically it's working - although I also take a nod to your 'at what cost'. But the idea that we could have been living in some Utopian BASIC derived alternate uinverse seems a little bit optimistic to me. I like AI coding (if I don't have to think of the costs)

Re: AI and Home-Cooked Software

#12

Post by moomoo11 »

_aavaa_ wrote: post #2

“Every line of AI-generated code is a plausible-looking liability. It may pass basic tests, only to fail spectacularly in production with an edge case you never considered.” Every time I read something along the lines I have to wonder whose code these people review during code reviews. It’s not like the alternative is bulletproof code.

They set up a GitHub action that has AI do an immediate first pass (hallucinates high on drugs and not the good kind) and leave a review.

Considering 80% of team mates are usually dead weight or mid at best (every team is carried by that 1 or 2 guys who do 2-3x), they will do the bare minimum review. Let’s be real.. PIP is real. Job hopping because bad is real.

It’s a problem. I have dealt with this and had to fire.

Re: AI and Home-Cooked Software

#13

Post by ivanech »

AI tools have been so good for me for making home-cooked software. As a new-ish parent, it’s so much easier to do stuff. I don’t need to go into extra-deep focus mode to learn how to center a div for the hundredth time, I can spend that precious focus time on the problems that matter / the core motivation.

Re: AI and Home-Cooked Software

#14

Post by bitwize »

jjmarr wrote: post #9

Long-term SWEs at non-tech companies will spend much of their time reviewing vibe coded features/prototypes/scripts from non-technical employees and scaling them once they become critical infrastructure. This'll eliminate jobs in the "develop CRUD app" industry but will create better jobs in security/scalability/quality expertise. But it'll take a few years as all these vibe coded business process scripts start to fa…

That is literally the exact promise of CASE tools in the 80s and the early 90s; UML code generation tools in the 2000s, and "low-code/no-code" platforms in the 2010s. It turned out to be a disaster every time, especially when the Idea Persons chucked their creations over the wall to SWEs to bash them into actual products because the Idea Persons had Far More Important Things To Do than maintain their coalesced brain farts.

We're repeating history but with more energy consumption.

Re: AI and Home-Cooked Software

#15

Post by bitwize »

djmips wrote: post #11
bitwize wrote: post #7

Ah, another "Now that we have AI, people can do [thing people could do for decades]" article. If there was something you wanted a computer to do, that it did not yet do, you programmed it. And if you didn't know how, you learned. BASIC was always there. But the industry as a whole moved away from the idea that end users are to program computers sometime in the 80s or 90s (the glorious point and click future was not e…

I feel like you've never actually tried to make tool with Claude Code or similar because BASIC is not it - that's viewing the past with rose coloured glasses. However I understand your central thesis that we could have actually put effort into making something that average folk could use to effectively leverage computers in a way that requires 'code'. But you know we have tried - we have Scratch - we have all of the…

I'm not saying BASIC is it. But it was good enough in its day—my father used it to write engine simulations. It was the first language to attack the problem "getting computing nonprofessionals to write their own programs for their own needs" and it achieved that very well by the standards of the 01960s-01980s. But the fact that every computer shipped with a language that allowed users to get started with programming right away, was a noble thing we should have sought to preserve even in the present day. Scratch is for kids, and node spaghetti presents the usual no-code issues. HolyC comes close, but you know, Terry Davis. An acquired taste.

I was kinda hoping that language would be Python, but even that requires ceremony these days.

Re: AI and Home-Cooked Software

#16

Post by chasd00 »

macNchz wrote: post #10

I've been thinking in a similar way over the past year or so—we're seeing the emergence of more widespread access to custom software for use cases that previously never would have justified the investment. There are so many situations where a little program can save one or a few people hours of work on tedious tasks, but wouldn't make sense to build or support given traditional software development overhead, but that…

i'm not an llm superfan but i do find them useful for coming up with one-off scripts to process datafiles or other small tasks that would have taken me a couple hours to get right. The llm produces something 85% of the way there including command line argument processing and all that stuff i hate to type out. I just fix the broken bits and make some adjustments and i have what i need in about 20 min. For that kind of task they are very useful IMO.

Re: AI and Home-Cooked Software

#17

Post by infinitezest »

I find LLMs really useful on a daily basis but I keep wondering, what's going to happen when the VC money dries up and the real cost of inference kicks in? Its relatively cheap now but its also being heavily subsidized. The usual answer is to just jam ads into your product and slowly increase the price over time (see: Netflix) but I don't know how that'll work for LLMs.

Re: AI and Home-Cooked Software

#18

Post by ndileas »

bitwize wrote: post #14
jjmarr wrote: post #9

Long-term SWEs at non-tech companies will spend much of their time reviewing vibe coded features/prototypes/scripts from non-technical employees and scaling them once they become critical infrastructure. This'll eliminate jobs in the "develop CRUD app" industry but will create better jobs in security/scalability/quality expertise. But it'll take a few years as all these vibe coded business process scripts start to fa…

That is literally the exact promise of CASE tools in the 80s and the early 90s; UML code generation tools in the 2000s, and "low-code/no-code" platforms in the 2010s. It turned out to be a disaster every time, especially when the Idea Persons chucked their creations over the wall to SWEs to bash them into actual products because the Idea Persons had Far More Important Things To Do than maintain their coalesced brain…

I wasn't around for the second millennium versions. At some point, doesn't there exist a kind of activation energy threshold where enough money/promise etc is gained from the prototype that this pattern works for good ideas and not for bad ones?

Re: AI and Home-Cooked Software

#19

Post by bonsai_spool »

bitwize wrote: post #14
jjmarr wrote: post #9

Long-term SWEs at non-tech companies will spend much of their time reviewing vibe coded features/prototypes/scripts from non-technical employees and scaling them once they become critical infrastructure. This'll eliminate jobs in the "develop CRUD app" industry but will create better jobs in security/scalability/quality expertise. But it'll take a few years as all these vibe coded business process scripts start to fa…

That is literally the exact promise of CASE tools in the 80s and the early 90s; UML code generation tools in the 2000s, and "low-code/no-code" platforms in the 2010s. It turned out to be a disaster every time, especially when the Idea Persons chucked their creations over the wall to SWEs to bash them into actual products because the Idea Persons had Far More Important Things To Do than maintain their coalesced brain…

I do think that there’s a difference in kind here - we’re not producing UML graphs that require programmer time to implement (or sending the diagrams to SE Asia and then code reviewing).

The code ‘works’ - and the folks who are improving the prototype can also benefit from the tools that the Idea Person used.

Re: AI and Home-Cooked Software

#20

Post by peacebeard »

_aavaa_ wrote: post #2

“Every line of AI-generated code is a plausible-looking liability. It may pass basic tests, only to fail spectacularly in production with an edge case you never considered.” Every time I read something along the lines I have to wonder whose code these people review during code reviews. It’s not like the alternative is bulletproof code.

A lot of people seem to equate using AI tools and deploying code that you don’t understand. All code should be fully understood by the person using the AI tool, then again by the reviewer. The productivity benefit of these tools is still massive, and there is benefit to doing the research and investigation to understand what the LLM is doing if it was not clear up front.
Post reply on HN