Or somebody could vibecode the software that runs on a missile silo and no one notices it has a bug until the warhead goes off before it gets launched.
Yep, most likely ai will kill us by turning our brains to mush, to a point where eg construction workers and nuclear safety engineers are asking “Chat, how do I …”. Software developers are front runners here.
It will never be a smart idea to use LLMs to approve PRs, no matter how good they get.
I will say this categorically. If there is not at least one human at the company who understands a piece of code, that piece of code is a liability, no matter how well it works. The reason that AI can’t replace programmers isn’t that it isn’t good enough at generating code (although I’d still argue it isn’t), it’s that even code written by humans who are no longer at the company is an unmaintainable mess to the next batch of programmers who replace them. Changes in one part of the code often affect other parts of the code in ways that are hard to predict unless you know all of it at once, and that’s something AI is really bad at spotting. Some human, somewhere, has to understand the thing, and if you’re automating the review process, sooner or later it’s gonna let a bug through and and it’s going to be borderline impossible to find.
Also, just the idea that you would trust an AI to decide what changes are good for your codebase… eugh.
This statement is true but somewhat for the wrong reason, LLM is a tool. And if a none dev uses it to create code, and just let their hands of the wheel then yes, shit will break and it will be bad. But enviornments like a missile silo, is already within itself a solved problem, but also has procedures to counteract the thing you yourself highlighted. Bad code and bad code practices existed before AI, and will continue to exist after AI.
It does not decide anything, it is an LLM. But for example what do you bring to a code review of a 10k+ lines to replace one parameter with another? The AI can do a very quick and efficient search for basic or clear bugs, and sometimes even logical errors, but yes should not remove the human altogether. But claiming that an AI would have no use in a PR workflow is ignorant.
Or somebody could vibecode the software that runs on a missile silo and no one notices it has a bug until the warhead goes off before it gets launched.
Yep, most likely ai will kill us by turning our brains to mush, to a point where eg construction workers and nuclear safety engineers are asking “Chat, how do I …”. Software developers are front runners here.
I’ve been saying this. LLMs are just turbocharged StackOverflow. People who exclusively copy paste from it should not be employed
Luckily AI is at least somewhat good at also checking the code! :D
It will never be a smart idea to use LLMs to approve PRs, no matter how good they get.
I will say this categorically. If there is not at least one human at the company who understands a piece of code, that piece of code is a liability, no matter how well it works. The reason that AI can’t replace programmers isn’t that it isn’t good enough at generating code (although I’d still argue it isn’t), it’s that even code written by humans who are no longer at the company is an unmaintainable mess to the next batch of programmers who replace them. Changes in one part of the code often affect other parts of the code in ways that are hard to predict unless you know all of it at once, and that’s something AI is really bad at spotting. Some human, somewhere, has to understand the thing, and if you’re automating the review process, sooner or later it’s gonna let a bug through and and it’s going to be borderline impossible to find.
Also, just the idea that you would trust an AI to decide what changes are good for your codebase… eugh.
Except for when it is.
This statement is true but somewhat for the wrong reason, LLM is a tool. And if a none dev uses it to create code, and just let their hands of the wheel then yes, shit will break and it will be bad. But enviornments like a missile silo, is already within itself a solved problem, but also has procedures to counteract the thing you yourself highlighted. Bad code and bad code practices existed before AI, and will continue to exist after AI.
It does not decide anything, it is an LLM. But for example what do you bring to a code review of a 10k+ lines to replace one parameter with another? The AI can do a very quick and efficient search for basic or clear bugs, and sometimes even logical errors, but yes should not remove the human altogether. But claiming that an AI would have no use in a PR workflow is ignorant.