Such a great question. If someone uses AI to build the entire product they themselves have no idea how any of it works, are they just going to vibe code the bug fixes too? What do they add to the prompt “and fix the bugs”?
Exactly this. We’re def seeing some growing pains, but as AI improves and more stuff gets vibe-coded, these common problems are gonna become easier to catch and prevent. Let’s say AI-built apps keep having the same shitty security flaw. Once enough people identify it and build the fix into their workflows, AI tools will start warning users about it and offering the fix automatically.
What fascinates me is how quickly shit has advanced. A little over a year ago, people said AI would never be able to program anything useful. A year before that, it struggled with basic math. Look at how much it has improved in only two years. Now fucking imagine where it could be three years from now.
In 3 years, the current state of flaws will just be a “Hey 'member when…”
I know Lemmy is anti-AI, which is ironic considering how many users here are computer nerds with jobs in tech. But let’s face it, AI is going to keep improving. It’s an amazing tool, and I love watching it develop. I use it every fucking day. The haters need to adapt to it or count on being unemployed. That’s just reality.
I also think most Lemmys use it every day, despite the show the put on about hating it. If they’re programmers, then they are definitely using it because most pro jobs require/force it to be used now.
I’m sure it’s a great use of time telling the AI to fix the bugs it generated without being able to verify if it actually fixes something and doesn’t introduce new bugs because it just regurgitates whatever its been trained on. Just use your customers for testing!
Just use your customers for testing!
Which is how most software has been tested for last several years. Long before AI. So…
I’m not saying that’s good. But that idea isn’t new just because of vibe coding. lol
