Well thats quite easy to say is it? Most professionally developed software is decades old, when standards and patterns were different. So only greenfield developed software may measure up to your ideas of good software.
And AI can only “improve” because of human created software, otherwise it would gobble itself up (Model Collapse).
Not to mention it will not just gobble up itself now, but also all the bad code on the internet.
If you ask ChatGPT or Codestral how to safe a file in memory it will in almost all cases save those data chunks you’re reading from somewhere in a list and suggest appending every chunk to that list.
Yeah sure you can do that instead of using io.bytesIO() (probably, until weird things happen), but what the fuck. And that’s Python, literally the language those models are supposed to excel in.
I was being deliberately combative, but there’s a kernel of truth. Our discipline had been at it for decades and we have no idea how to write maintainable, secure, usable software. Most of the things we write are neither of those things, despite our very best efforts.
As an engineering discipline, we failed miserably.
That said, AI is definitely not going to improve things. It’s going to make things worse. Much much worse.
What? Who are we in this case? Because we as a community do know how to. The question is does our Stakeholders know that it may cost more and do they even care?
So it circles back to the meme. Does the vibe coder care if the AI Slop is maintainable, safe and secure? Or does they even know how to spot it?
What? Who are we in this case? Because we as a community do know how to. The question is does our Stakeholders know that it may cost more and do they even care?
This really pisses me off. Do you think civil engineers would build unsafe bridges because “stakeholders” told them to?
No, because they would go to fucking jail.
We are a fucking disgrace. We need software liability laws because apparently we can’t stand up for ourselves and our profession.
Do you think civil engineers would build unsafe bridges because “stakeholders” told them to? No, because they would go to fucking jail.
sure the first iteration of the bridge would be up to spec as they need to be inspected first to be approved. So the Stakeholders would do anything to get it done, buuuuuuut maybe not with the best materials and the best engineers/workers but certainly on time in mind.
But afterwards? when it comes to maintaining the bridge; eeeh, its good enough, here a patchwork; who cares if there are potholes everywhere, you can still use it right?
We are a fucking disgrace. We need software liability laws because apparently we can’t stand up for ourselves and our profession.
Can’t really stand up when CEOs constantly lay off teams. So why would one guy who stands up for what is just and right matters to them?
Well thats quite easy to say is it? Most professionally developed software is decades old, when standards and patterns were different. So only greenfield developed software may measure up to your ideas of good software.
And AI can only “improve” because of human created software, otherwise it would gobble itself up (Model Collapse).
Not to mention it will not just gobble up itself now, but also all the bad code on the internet.
If you ask ChatGPT or Codestral how to safe a file in memory it will in almost all cases save those data chunks you’re reading from somewhere in a list and suggest appending every chunk to that list.
Yeah sure you can do that instead of using io.bytesIO() (probably, until weird things happen), but what the fuck. And that’s Python, literally the language those models are supposed to excel in.
I was being deliberately combative, but there’s a kernel of truth. Our discipline had been at it for decades and we have no idea how to write maintainable, secure, usable software. Most of the things we write are neither of those things, despite our very best efforts. As an engineering discipline, we failed miserably.
That said, AI is definitely not going to improve things. It’s going to make things worse. Much much worse.
What? Who are we in this case? Because we as a community do know how to. The question is does our Stakeholders know that it may cost more and do they even care?
So it circles back to the meme. Does the vibe coder care if the AI Slop is maintainable, safe and secure? Or does they even know how to spot it?
This really pisses me off. Do you think civil engineers would build unsafe bridges because “stakeholders” told them to? No, because they would go to fucking jail.
We are a fucking disgrace. We need software liability laws because apparently we can’t stand up for ourselves and our profession.
sure the first iteration of the bridge would be up to spec as they need to be inspected first to be approved. So the Stakeholders would do anything to get it done, buuuuuuut maybe not with the best materials and the best engineers/workers but certainly on time in mind.
But afterwards? when it comes to maintaining the bridge; eeeh, its good enough, here a patchwork; who cares if there are potholes everywhere, you can still use it right?
Can’t really stand up when CEOs constantly lay off teams. So why would one guy who stands up for what is just and right matters to them?