New Software Developer Generation Is Cooked. OpenAI Keys Exposed | Hacker News Reader

New Software Developer Generation Is Cooked. OpenAI Keys Exposed

twitter.com

5 pointsitsdevdaniel14 comments

I'm curious how new technologies that make things easier, sort of bifurcate the populous. I guess on the one hand, the calculator made it so that you could do a job that required a little bit of math, without having to be good at math. And that's good! Maybe there are a bunch of jobs that only require a little bit of math.

But like, a person who is good at math with a calculator will probably go much farther. If you think that LLM's mean you don't have to learn coding, you're never going to do as well as the person who goes above and beyond. And the existence of new technologies doesn't change this underlying behavior. Maybe in the end nothing much really changes, either you're the type of person who wants to learn the underlying skill or not.

There's probably a third category of people, and I don't know how big it is, but like, being able to make music on your computer probably radically altered the number of people making music, partially because, if your core skill was wanting to compose music and not play music, the computer actually enabled a new skill. Not only did you not have to learn how to play a traditional instrument (though those that do are probably at a significant advantage) but it would be super expensive to have a band or orchestra at your disposal to play that music, prior to computer music.

I'm curious how big this third group is in the AI world. Probably not tiny, especially if your goal is to say test the viability of a bunch of CRUD app ideas. And lots of businesses are CRUD apps.

But actually, there may be another use case for LLMs and that's really modular applications. I had this idea for a new kind of in-browser DAW a while ago. I got the basics going, but adding on to it was going to be a slog and I kind of got disinterested, but since the DAW is modular and wants to include a whole mess of relatively simple components, that's probably a great problem space for LLMs, since each component requires relatively little context.

I've already experienced this success building myself a custom dashboard for work. Each dashboard component is basically just a really simple repeatable application (RSS reader, todo list, music player etc.). So, if the app design is modular, it's pretty easy for an LLM to successfully create the dashboard panels.

> being able to make music on your computer probably radically altered the number of people making music

I don't think this is substantially true. I'm old enough to remember back before it was possible for most people to make music on a computer. My view is that if the computer affected this at all, it slightly reduced the number of people making music, not increased it. But I really think that the percentage of people making music hasn't changed that much at all. People who want to compose or perform music have always done so, and always will, regardless of the toolset. The barrier to entry isn't high.

And it seems to me that most regular people who are making music now don't seem to be doing it on a computer.

Interesting, do you think that computer music cannibalized other forms of music?

I may be biased, because I feel like personally, I would not have made music without the computer, because my main interest is in like full work composition of multiple instruments, and the barrier to entry to really learn an instrument does turn me off.

There are definitely less bands playing together nowadays than 20 years ago and way less famous ones. While there are much more single musicians using computer music as base.

> do you think that computer music cannibalized other forms of music?

I don't know, but my gut tells me that is has not to any serious degree. That is, I don't think that many people who were making music without computers stopped doing that to make music with computers. Some may have added computers to their instrument repertoire, though.

Utter clickbait. Taking this one mistake as an indication that a "generation" is writing bad code. One more reason to not have an X account.

one mistake that most vibe coders are overseeing

kids did this before LLMS

yes, but now its worse

how old are you? Maybe it is the first time you are seeing this cycle so you think now is worse. Kids these days...

irrelevant, do some research for yourself, after the use of LLMs its more prone. The mistakes might be the same, but they’re happening faster, more often, and by people who don’t even realize there’s a mistake.

You that are paying more attention, now is not different and the world won't end.

"Do some research for yourself" is not actually a rebuttal.

Supply some data, maybe?

Its not hard to look it up yourself.

https://www.csoonline.com/article/3953927/ai-programming-cop...

"GitGuardian’s State of Secrets Sprawl 2025 report revealed a 25% increase in leaked secrets year-over-year, with 23.8 million new credentials detected on public GitHub in 2024 alone."

"“These tools often lack contextual awareness of security practices and, without proper oversight, can generate insecure code and persistent vulnerabilities,” Smith said. “This becomes a systematic issue as LLM-generated code spreads and creates flaws throughout the supply chain, with over 70% of critical security debt now stemming from third-party code,”

Reply on news.ycombinator.com